By Topic

Vision, Image and Signal Processing, IEE Proceedings -

Popular Articles (November 2014)

Includes the top 50 most frequently downloaded documents for this publication according to the most recent monthly usage statistics.
  • 1. FPGA implementations of fast Fourier transforms for real-time signal and image processing

    Page(s): 283 - 296
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (636 KB)  

    Applications based on the fast Fourier transform (FFT), such as signal and image processing, require high computational power, plus the ability to experiment with algorithms. Reconfigurable hardware devices in the form of field programmable gate arrays (FPGAs) have been proposed as a way of obtaining high performance at an economical price. However, users must program FPGAs at a very low level and have a detailed knowledge of the architecture of the device being used. They do not therefore facilitate easy development of, or experimentation with, signal/image processing algorithms. To try to reconcile the dual requirements of high performance and ease of development, the paper reports on the design and realisation of a high level framework for the implementation of 1D and 2D FFTs for real-time applications. A wide range of FFT algorithms, including radix-2, radix-4, split-radix and fast Hartley transform (FHT) have been implemented under a common framework in order to enable system designers to meet different system requirements. Results show that the parallel implementation of 2D FFT achieves linear speed-up and real-time performance for large matrix sizes. Finally, an FPGA-based parametrisable environment based on 2D FFT is presented as a solution for frequency-domain image filtering application. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 2. Intelligent distributed surveillance systems: a review

    Page(s): 192 - 204
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (643 KB)  

    This survey describes the current state-of-the-art in the development of automated visual surveillance systems so as to provide researchers in the field with a summary of progress achieved to date and to identify areas where further research is needed. The ability to recognise objects and humans, to describe their actions and interactions from information acquired by sensors is essential for automated visual surveillance. The increasing need for intelligent visual surveillance in commercial, law enforcement and military applications makes automated visual surveillance systems one of the main current application domains in computer vision. The emphasis of this review is on discussion of the creation of intelligent distributed automated surveillance systems. The survey concludes with a discussion of possible future directions. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 3. Estimation of image noise variance

    Page(s): 80 - 84
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (852 KB)  

    A novel algorithm for estimating the noise variance of an image is presented. The image is assumed to be corrupted by Gaussian distributed noise. The algorithm estimates the noise variance in three steps. At first the noisy image is filtered by a horizontal and a vertical difference operator to suppress the influence of the (unknown) original image. In a second step a histogram of local signal variances is computed. Finally a statistical evaluation of the histogram provides the desired estimation value. For a comparison with several previously published estimation methods an ensemble of 128 natural and artificial test images is used. It is shown that with the novel algorithm more accurate results can be achieved View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 4. Complex gradient and Hessian

    Page(s): 380 - 383
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (184 KB)  

    The gradient and Hessian are often used in analytical and numerical function optimisation complex valued parameter estimation problems. In a number of signal processing applications the function is a real function of complex variables. Then the optimisation is usually carried out with respect to the real and imaginary part of these variables; therefore, the gradient and Hessian concerned are real. The reason for this approach is to avoid difficulties with the definition and interpretation of the gradient and Hessian with respect to the complex variables. Definitions of a complex gradient and Hessian are proposed to solve these difficulties. The proposed and the real gradient and Hessian are fully compatible and are related by simple linear transformations. The results presented are an extension of a result by Brandwood (1983) concerning a complex gradient View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 5. Visual gesture recognition

    Page(s): 101 - 106
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (540 KB)  

    Presents a method for recognising human-hand gestures using a model based approach. A finite state machine is used to model four qualitatively distinct phases of a generic gesture. Fingertips are tracked in multiple frames to compute motion trajectories. The trajectories are then used for finding the start and stop position of the gesture. Gestures are represented as a list of vectors and are then matched to stored gesture vector models using table lookup based on vector displacements. Results are presented showing recognition of seven gestures using images sampled at 4 Hz on a SPARC-1 without any special hardware. The seven gestures are representatives for actions of left, right, up, down, grab, rotate, and stop View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 6. Shape based leaf image retrieval

    Page(s): 34 - 43
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (623 KB)  

    The authors present an efficient two-stage approach for leaf image retrieval by using simple shape features including centroid-contour distance (CCD) curve, eccentricity and angle code histogram (ACH). In the first stage, the images that are dissimilar with the query image are first filtered out by using eccentricity to reduce the search space, and fine retrieval follows by using all three sets of features in the reduced search space in the second stage. Different from eccentricity and ACH, the CCD curve is neither scaling-invariant nor rotation-invariant. Therefore, normalisation is required for the CCD curve to achieve scaling invariance, and starting point location is required to achieve rotation invariance with the similarity measure of CCD curves. A thinning-based method is proposed to locate starting points of leaf image contours, so that the approach used is more computationally efficient. Actually, the method can benefit other shape representations that are sensitive to starting points by reducing the matching time in image recognition and retrieval. Experimental results on 1400 leaf images from 140 plants show that the proposed approach can achieve a better retrieval performance than both the curvature scale space (CSS) method and the modified Fourier descriptor (MFD) method. In addition, the two-stage approach can achieve a performance comparable to an exhaustive search, but with a much reduced computational complexity. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 7. Regular implementation algorithms of time domain aliasing cancellation

    Page(s): 387 - 392
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (616 KB)  

    The authors propose two highly regular algorithms for realising the time domain aliasing cancellation (TDAC) technique. The first TDAC implementation, which is based on the fast discrete cosine transform, effectively adopts analysis and synthesis window functions in the transform structure. This implementation algorithm achieves the least computational complexity in TDAC processes. The second TDAC implementation, which extends Goertzel's concept, uses a simple selectable-fixed-coefficient second-order infinite impulse response (IIR) filter to recursively achieve multichannel audio encoding and decoding processes. With a properly selected coefficient, this recursive implementation achieves a lower round-off-error than the current fast implementations and the direct implementation in finite wordlength. In recently developed high quality consumer products, the first algorithm is suitable to be realised in digital signal processing chips and the second one will be a better choice for VLSI implementation View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 8. Continuous restricted Boltzmann machine with an implementable training algorithm

    Page(s): 153 - 158
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (649 KB)  

    The authors introduce a continuous stochastic generative model that can model continuous data, with a simple and reliable training algorithm. The architecture is a continuous restricted Boltzmann machine, with one step of Gibbs sampling, to minimise contrastive divergence, replacing a time-consuming relaxation search. With a small approximation, the training algorithm requires only addition and multiplication and is thus computationally inexpensive in both software and hardware. The capabilities of the model are demonstrated and explored with both artificial and real data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 9. Fingerprint enhancement by directional Fourier filtering

    Page(s): 87 - 94
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (832 KB)  

    A new method of enhancing fingerprint images is described, based upon nonstationary directional Fourier domain filtering. Fingerprints are first smoothed using a directional filter whose orientation is everywhere matched to the local ridge orientation. Thresholding then yields the enhanced image. Various simplifications lead to efficient implementation on general-purpose digital computers. Results of enhancement are presented for fingerprints of various pattern classifications. A comparison is made with the enhancement used within the automated fingerprint identification system (AFIS) developed by the UK Home Office. Use of the proposed enhancement method leads to significant improvements in the speed and accuracy of the AFIS View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Full text access may be available. Click article title to sign in or learn about subscription options.
  • 11. Combined thresholding and neural network approach for vein pattern extraction from leaf images

    Page(s): 881 - 892
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1693 KB)  

    Living plant recognition based on images of leaf, flower and fruit is a very challenging task in the field of pattern recognition and computer vision. There has been little work reported on flower and fruit image processing and recognition. In recent years, several researchers have dedicated their work to leaf characterisation. As an inherent trait, leaf vein definitely contains the important information for plant species recognition despite its complex modality. A new approach that combines a thresholding method and an artificial neural network (ANN) classifier is proposed to extract leaf veins. A preliminary segmentation based on the intensity histogram of leaf images is first carried out to coarsely determine vein regions. This is followed by a fine segmentation using a trained ANN classifier with ten features extracted from a window centred on the object pixel as its inputs. Compared with other methods, experimental results show that this combined approach is capable of extracting more accurate venation modality of the leaf for the subsequent vein pattern classification. The approach can also reduce the computing time compared with a direct neural network approach View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 12. High capacity image steganographic model

    Page(s): 288 - 294
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (805 KB)  

    Steganography is an ancient art of conveying messages in a secret way that only the receiver knows the existence of a message. So a fundamental requirement for a steganographic method is imperceptibility; this means that the embedded messages should not be discernible to the human eye. There are two other requirements, one is to maximise the embedding capacity, and the other is security. The least-significant bit (LSB) insertion method is the most common and easiest method for embedding messages in an image. However, how to decide on the maximal embedding capacity for each pixel is still an open issue. An image steganographic model is proposed that is based on variable-size LSB insertion to maximise the embedding capacity while maintaining image fidelity. For each pixel of a grey-scale image, at least four bits can be used for message embedding. Three components are provided to achieve the goal. First, according to contrast and luminance characteristics, the capacity evaluation is provided to estimate the maximum embedding capacity of each pixel. Then the minimum-error replacement method is adapted to find a grey scale as close to the original one as possible. Finally, the improved grey-scale compensation, which takes advantage of the peculiarities of the human visual system, is used to eliminate the false contouring effect. Two methods, pixelwise and bitwise, are provided to deal with the security issue when using the proposed model. Experimental results show effectiveness and efficiency of the proposed model. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 13. Robust rotation-invariant texture classification: wavelet, Gabor filter and GMRF based schemes

    Page(s): 180 - 188
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1476 KB)  

    Three novel feature extraction schemes for texture classification are proposed. The schemes employ the wavelet transform, a circularly symmetric Gabor filter or a Gaussian Markov random field with a circular neighbour set to achieve rotation-invariant texture classification. The schemes are shown to give a high level of classification accuracy compared to most existing schemes, using both fewer features (four) and a smaller area of analysis (16×16). Furthermore, unlike most existing schemes, the proposed schemes are shown to be rotation invariant demonstrate a high level of robustness noise. The performances of the three schemes are compared, indicating that the wavelet-based approach is the most accurate, exhibits the best noise performance and has the lowest computational complexity View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 14. Digital integrator design using Simpson rule and fractional delay filter

    Page(s): 79 - 86
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (164 KB)  

    The IIR digital integrator is designed by using the Simpson integration rule and fractional delay filter. To improve the design accuracy of a conventional Simpson IIR integrator at high frequency, the sampling interval is reduced from T to 0.5T. As a result, a fractional delay filter needed to be designed in the proposed Simpson integrator. However, this problem can be solved easily by applying well-documented design techniques of the FIR and all-pass fractional delay filters. Several design examples are illustrated to demonstrate the effectiveness of the proposed method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 15. Forward sequential algorithms for best basis selection

    Page(s): 235 - 244
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (404 KB)  

    The problem of signal representation in terms of basis vectors from a large, over-complete, spanning dictionary has been the focus of much research. Achieving a succinct, or `sparse', representation is known as the problem of best basis representation. Methods are considered which seek to solve this problem by sequentially building up a basis set for the signal. Three distinct algorithm types have appeared in the literature which are here termed basic matching pursuit (BMP), order recursive matching pursuit (ORMP) and modified matching pursuit (MMP). The algorithms are first described and then their computation is closely examined. Modifications are made to each of the procedures which improve their computational efficiency. The complexity of each algorithm is considered in two contexts; one where the dictionary is variable (time-dependent) and the other where the dictionary is fixed (time-independent). Experimental results are presented which demonstrate that the ORMP method is the best procedure in terms of its ability to give the most compact signal representation, followed by MMP and then BMP which gives the poorest results. Finally, weighing the performance of each algorithm, its computational complexity and the type of dictionary available, recommendations are made as to which algorithm should be used for a given problem View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 16. Multimodal biometric user-identification system for network-based applications

    Page(s): 409 - 416
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (573 KB)  

    The authors describe the design and development of a prototype of a multimodal biometric system for the automatic identification of an individual. Unlike in other proposed multimodal biometric systems, biometric features are acquired from the same image, using a low-cost scanner, at the same time. It makes the system suitable for home and for many network-based applications. After the pre-processing phase, the hand-geometry, finger and palm-print features invariant to hand translation and rotation on the scanner are extracted. Fusion at the matching-score level is obtained by means of the total similarity measure. In the decision module, three rules are used to establish identity. The system was tested on a database of 130 persons. The test performance, FAR=0% and FRR=0.2%, suggests that the system can be used in medium/high-security Internet environments. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 17. Properties determining choice of mother wavelet

    Page(s): 659 - 664
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (155 KB)  

    Properties of wavelets with finite as well as infinite support are summarised to facilitate mother wavelet selection in a chosen application. The quantitative guidelines reduce dependence on trial-and-error schemes resorted to for selection and underscore the importance of such selection in any application of interest. In wavelet-based image sequence superresolution, studied during the last four years, the use of a B-spline mother wavelet is justified. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 18. Image restoration using the particle filter: handling non-causality

    Page(s): 650 - 656
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (305 KB)  

    A recursive-state estimation scheme for image restoration using the particle filter is described. Handling non-causal blurs within a recursive framework is a challenging problem. Most recursive image restoration schemes assume the blur to be causal or semi-causal in nature, but this is unrealistic. A novel choice for the state vector and a concurrent block estimation technique to incorporate full-plane regions of support for the image model as well as the blur are proposed. The particle filter-based framework enables general types of degradations to be tackled. The method assumes that the functional form of the distortion as well as the noise statistics are known but does not place any restrictions on them. Several experimental results are presented to validate the approach View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 19. Human gait recognition in canonical space using temporal templates

    Page(s): 93 - 100
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (716 KB)  

    A system for automatic gait recognition without segmentation of particular body parts is described. Eigenspace transformation (EST) has already proved useful for several tasks including face recognition, gait analysis, etc; it is optimal in dimensionality reduction by maximising the total scatter of all classes but is not optimal for class separability. A statistical approach that combines EST with canonical space transformation (CST) is proposed for gait recognition using temporal templates from a gait sequence as features. This method can be used to reduce data dimensionality and to optimise the class separability of different gait sequences simultaneously. Incorporating temporal information from optical-flow changes between two consecutive spatial templates, each temporal template extracted from computation of optical flow is projected from a high-dimensional image space to a single point in a low-dimensional canonical space. Using template matching, recognition of human gait becomes much faster and simpler in this new space. As such, the combination of EST and CST is shown to be of considerable potential in an emerging new biometric View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 20. Bispectral and trispectral features for machine condition diagnosis

    Page(s): 229 - 234
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (260 KB)  

    The application of bispectral and trispectral analysis in condition monitoring is discussed. Higher-order spectral analysis of machine vibrations for the provision of diagnostic features is investigated. Experimental work is based on vibration data collected from a small test rig subjected to bearing faults. The direct use of the entire bispectrum or trispectrum to provide diagnostic features is investigated using a variety of classification algorithms including neural networks, and this is compared with simpler power spectral and statistical feature extraction algorithms. A more detailed investigation of the higher-order spectral structure of the signals is then undertaken. This provides features which can be estimated more easily in practice and could provide diagnostic information about the machines View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 21. Robust fabric defect detection and classification using multiple adaptive wavelets

    Page(s): 715 - 723
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (734 KB)  

    The wavelet transform has been widely used for defect detection and classification in fabric images. The detection and classification performance of the wavelet transform approach is closely related to the selection of the wavelet. Instead of predetermining a wavelet, a method of designing a wavelet to adapt to the detection or classification of fabric defects has been developed. For further improvement of the performance, the paper extends the adaptive wavelet-based methodology from the use of a single adaptive wavelet to multiple adaptive wavelets. For each class of fabric defect, a defect-specific adaptive wavelet was designed to enhance the defect region at one channel of the wavelet transform, where the defect region can be detected by using a simple threshold classifier. Corresponding to the multiple defect-specific adaptive wavelets, the multiscale edge responses to defect regions have been shown to be more efficient in characterising defects, which leads to a new approach to the classification of defects. In comparison with the single adaptive wavelet approach, the use of multiple adaptive wavelets yields better performance on defect detection and classification, especially for defects that are poorly detected by the single adaptive wavelet approach. The proposed method has been evaluated on the inspection of 56 images containing eight classes of fabric defects, and 64 images without defects. In defect detection, 98.2% detection rate and 1.5% false alarm rate were achieved, and in defect classification, 97.5% accuracy was achieved. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 22. Original approach for the localisation of objects in images

    Page(s): 245 - 250
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (432 KB)  

    An original approach is presented for the localisation of objects in an image which approach is neuronal and has two steps. In the first step, a rough localisation is performed by presenting each pixel with its neighbourhood to a neural net which is able to indicate whether this pixel and its neighbourhood are the image of the search object. This first filter does not discriminate for position. From its result, areas which might contain an image of the object can be selected. In the second step, these areas are presented to another neural net which can determine the exact position of the object in each area. This algorithm is applied to the problem of localising faces in images View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 23. Efficient technique for circle detection using hypothesis filtering and Hough transform

    Page(s): 292 - 300
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1000 KB)  

    A fast circle detection method using a variant of Hough-like technique is reported. The proposed technique is simple in implementation, efficient in computation and robust to noise. In general, to evaluate circle parameters for all possible point triplets in an edge image containing n points, nC3 enumerations of the points have to be examined. However, if specific relations of the circle points are sought, the required number of enumerations can be reduced. The authors propose one such scheme of detection with point triplets possessing right angle property and the required enumerations can be reduced to nC2. Moreover, a novel processing strategy known as hypothesis filtering is introduced. The strategy includes two hypothesis constraints termed consistency checking with gradient angles and neighbouring points validation. Experimental results are demonstrated to reveal the performance of the method in detecting circles in both synthetic and real images. Since the proposed method adopts a right angle criterion for hypothesis, circles occluded or broken by more than one half may not be detected. Test results show that the limitation of the proposed method appears to be acceptable. When compared with established Hough transform techniques, the main strengths of the proposed detection method are its attractively computational and memory complexities and good accuracy of detection View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Full text access may be available. Click article title to sign in or learn about subscription options.
  • 25. Uniform distribution of points on a hyper-sphere with applications to vector bit-plane encoding

    Page(s): 187 - 193
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (528 KB)  

    In vector bit-plane encoding schemes, codebooks must be uniformly distributed on a hyper-sphere. Shells of regular lattices are often used, but they provide only a limited choice of number of vectors K and dimension N. The authors propose a method to generate codebooks in dimension N with arbitrary number K of vectors, almost uniformly distributed on a hyper-sphere. The uniform distribution of an arbitrary number of points on the surface of a hyper-sphere is still an open problem. Some mathematicians indeed consider it one of the mathematical challenges of the 21st century. The proposed method uses a combination of geometric and stochastic approaches to generate approximate solutions. The generated codebooks are tested in vector bit-plane encoding schemes. The results show that the proposed method is effective in generating codebooks almost uniformly distributed on a hyper-sphere View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 26. Robust microphone arrays using subband adaptive filters

    Page(s): 17 - 25
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (874 KB)  

    A new adaptive beamformer which combines the idea of subband processing and the generalised sidelobe canceller structure is presented. The proposed subband beamformer has a blocking matrix that uses coefficient-constrained subband adaptive filters to limit target cancellation within an allowable range of direction of arrival. Simulations comparing the fullband and subband adaptive beamformers show that the subband beamformer has better performance than the fullband beamformer when the target and/or interfering signals are coloured. In reverberant environments, the proposed subband beamformer also performs better than its fullband counterpart View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 27. Genetic algorithms for feature selection in machine condition monitoring with vibration signals

    Page(s): 205 - 212
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (508 KB)  

    Artificial neural networks (ANNs) can be used successfully to detect faults in rotating machinery. Using statistical estimates of the vibration signal as input features. In any given scenario, there are many different possible features that may be used as inputs for the ANN. One of the main problems facing the use of ANNs is the selection of the best inputs to the ANN, allowing the creation of compact, highly accurate networks that require comparatively little preprocessing. The paper examines the use of a genetic algorithm (GA) to select the most significant input features from a large set of possible features in machine condition monitoring. Using a GA, a subset of six input features is selected from a set of 66 giving a classification accuracy of 99.8%, compared with an accuracy of 87.2% using an ANN without feature selection and all 66 inputs. From a larger set of 156 different features, the GA is able to select a set of six features to give 100% recognition accuracy View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 28. Design of two-channel linear-phase QMF banks based on real IIR all-pass filters

    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (350 KB)  

    The design of two-channel linear-phase quadrature mirror filter (QMF) banks constructed by real infinite impulse response (IIR) digital all-pass filters is considered. The design problem is appropriately formulated to result in a simple optimisation problem. Using a variant of Karmarkar's algorithm, the optimisation problem can be efficiently solved through a frequency sampling and iterative approximation method to find the real coefficients for the IIR digital all-pass filters. The resulting two-channel QMF banks possess an approximately linear phase response without magnitude distortion. The effectiveness of the proposed technique is achieved by forming an appropriate Chebyshev approximation of the desired phase response and then finding its solution from a linear subspace in a few iterations. Finally, several simulation examples are presented for illustration and comparison View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 29. Novel approach to automated fingerprint recognition

    Page(s): 160 - 166
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1276 KB)  

    The paper describes an enhanced fingerprint recognition system consisting of image preprocessing, feature extraction and matching that runs accurately and effectively on a personal computer platform. The image preprocessing includes histogram equalisation, modification of directional codes, dynamic thresholding and ridgeline thinning which are sufficient to enhance the image to a state ready for feature extraction. Only features extracted are stored in a file for fingerprint matching. The matching algorithm presented is a modification and improvement of the structural approach. Experimental results acquired for matching are accurate, reliable and fast for implementation using a PC and a fingerprint scanner. The proposed fingerprint recognition scheme can provide an efficient way of automated identification and can be extended to numerous other security or administration applications View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 30. Statistical modelling of the wavelet coefficients with different bases and decomposition levels

    Page(s): 203 - 206
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (237 KB)  

    Various wavelet coefficient statistics are useful to increase the compression efficiency of images. The distribution of the wavelet coefficients within a sub-band affects how the decompression values should be adjusted. A generalised Gaussian distribution model is seen to be applicable from a theoretical point of view, and a convenient way is provided to estimate the parameters from the empirical data. This knowledge is applied to examine the best statistical modelling of the wavelet coefficients for different wavelet bases and decomposition levels. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 31. Aerial inspection of overhead power lines using video: estimation of image blurring due to vehicle and camera motion

    Page(s): 157 - 166
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1744 KB)  

    One of the principal difficulties of video inspection of overhead power distribution lines from a helicopter is the blurring of the image due to rotation of the camera in its gimbals and the translational motion of the helicopter. The author presents a kinematic model describing the sightline geometry which includes the effect of the helicopter's rectilinear motion in three degrees of freedom. It is shown that, except for locations very near to the object being inspected the velocity flow field of the image is substantially uniform. The image Jacobian is used to predict how much blur is to be expected along a typical flight path during inspection operations. An important conclusion is that this will be greater than the tolerable limit of 1-2% unless the camera is rotated at a rate which compensates for the helicopter's rectilinear motion the implications for the design of an automated object tracking system are discussed. The maximum singular value of the image Jacobian is proposed as a quality measure for inspection from various locations of the helicopter relative to the target object. It is found to be an useful figure of relative merit but too conservative to provide a realistic upper bound on image blur View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 32. Analysing animal behaviour in wildlife videos using face detection and tracking

    Page(s): 305 - 312
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (695 KB)  

    An algorithm that categorises animal locomotive behaviour by combining detection and tracking of animal faces in wildlife videos is presented. As an example, the algorithm is applied to lion faces. The detection algorithm is based on a human face detection method, utilising Haar-like features and AdaBoost classifiers. The face tracking is implemented by applying a specific interest model that combines low-level feature tracking with the detection algorithm. By combining the two methods in a specific tracking model, reliable and temporally coherent detection/tracking of animal faces is achieved. The information generated by the tracker is used to automatically annotate the animal's locomotive behaviour. The annotation classes of locomotive processes for a given animal species are predefined by a large semantic taxonomy on wildlife domain. The experimental results are presented. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 33. Survey and comparative analysis of entropy and relative entropy thresholding techniques

    Page(s): 837 - 850
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1021 KB)  

    Entropy-based image thresholding has received considerable interest in recent years. Two types of entropy are generally used as thresholding criteria: Shannon's entropy and relative entropy, also known as Kullback-Leibler information distance, where the former measures uncertainty in an information source with an optimal threshold obtained by maximising Shannon's entropy, whereas the latter measures the information discrepancy between two different sources with an optimal threshold obtained by minimising relative entropy. Many thresholding methods have been developed for both criteria and reported in the literature. These two entropy-based thresholding criteria have been investigated and the relationship among entropy and relative entropy thresholding methods has been explored. In particular, a survey and comparative analysis is conducted among several widely used methods that include Pun and Kapur's maximum entropy, Kittler and Illingworth's minimum error thresholding, Pal and Pal's entropy thresholding and Chang et al.'s relative entropy thresholding methods. In order to objectively assess these methods, two measures, uniformity and shape, are used for performance evaluation View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 34. Designing a fuzzy step size LMS algorithm

    Page(s): 261 - 266
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (560 KB)  

    A new approach in adjusting the step size of the least mean square (LMS) using the fuzzy logic technique is presented. It extends the earlier work of Gan (see Signal Process., vol.49, no.2, p.145-49, 1996) by giving a complete design methodology and guidelines for developing a reliable and robust fuzzy step size LMS (FSS LMS) algorithm. It also presents a computational study and simulation results of this newly proposed algorithm compared to other conventional variable step size LMS algorithms View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 35. Sub-pixel non-parametric PSF estimation for image enhancement

    Page(s): 285 - 292
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (844 KB)  

    Applying standard resolution enhancement and sub-pixel measurement techniques to an imaging system is problematic when the system characteristics are not known. The importance of precise system characterisation is often underestimated in resolution enhancement and sub-pixel measurement. The methods presented allow accurate sub-pixel measurements of system characteristics to be made with minimal assumptions. The nondeveloped parametric technique developed accurately characterises the properties of an imaging system. This is demonstrated by measuring the point spread function (PSF), along with static and dynamic distortions, for a high precision thermal imaging system to sub-pixel accuracy. The PSF is estimated to ±0.1 of a pixel and imaging system errors to the order of ±0.1 of a pixel are identified. The improved precision of PSF estimation is shown to benefit resolution enhancement. A novel feature of the method used to estimate the PSF (and to enhance the image) is that the estimation of the spatially invariant subpixel pixel PSF and of geometric distortion are performed independently View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 36. Novelty detection using extreme value statistics

    Page(s): 124 - 129
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (392 KB)  

    Extreme value theory is a branch of statistics that concerns the distribution of data of unusually low or high value, i.e. in the tails of some distribution. These extremal points are important in many applications as they represent the outlying regions of normal events against which we may wish to define abnormal events. In the context of density modelling, novelty detection or radial-basis function systems, points that lie outside of the range of expected extreme values may be flagged as outliers. There has been interest in the area of novelty detection, but decisions as to whether a point is an outlier or not tend to be made on the basis of exceeding some (heuristic) threshold. It is shown that a more principled approach may be taken using extreme value statistics View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 37. Watermark detection based on the properties of error control codes

    Page(s): 115 - 121
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (640 KB)  

    Watermark detection is a topic which is seldom addressed in the watermarking literature. Most authors concentrate on developing novel watermarking algorithms. In a practical watermarking system, however, one must be able to distinguish between watermarked and unwatermarked documents. Many existing systems belong to the class of so called 'yes/no' watermarks, where the detector correlates the candidate image with some known sequence to determine whether a mark is present. Unfortunately, these watermarks often carry no extra information and are not very useful. On the other hand, multi-bit watermarking schemes typically use a separate reference watermark and the payload of the watermark is decoded only when this reference watermark is successfully detected in the received image. It is shown that it is not necessary to use a reference watermark for detection purposes if the watermark payload is encoded with an error control code. One can then put all the energy into the payload watermark and increase its robustness. The turbo code is used as an example of error control codes in the work presented, and simulation results using an algorithm based on the authors' previous work verifies their theory. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 38. Novelty detection and neural network validation

    Page(s): 217 - 222
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (460 KB)  

    One of the key factors which limits the use of neural networks in many industrial applications has been the difficulty of demonstrating that a trained network will continue to generate reliable outputs once it is in routine use. An important potential source of errors is novel input data; that is, input data which differ significantly from the data used to train the network. The author investigates the relationship between the degree of novelty of input data and the corresponding reliability of the outputs from the network. He describes a quantitative procedure for assessing novelty, and demonstrates its performance by using an application which involves monitoring oil flow in multiphase pipelines View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 39. Time-frequency localisation of Shannon wavelet packets

    Page(s): 365 - 369
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (241 KB)  

    It is well known that the 2π minimally supported frequency scaling function φα(x) satisfying φˆα(ω)=χ(-α,2π-α)(ω), 0<α<2π, is not time localised. The Shannon wavelet packets φnα(x) are generated from φα(x) via an arbitrary pair of low-pass and high-pass filters (h, g), which are associated with an orthogonal multiresolution analysis. The authors prove that φnα(x) is time localised if α and n satisfy certain conditions. They also show that the decay properties of φnα(x) depend on the multiplicity of the zero ω=π of the symbol m0(ω) of the low-pass filter h. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 40. Noise-robust pitch detection method using wavelet transform with aliasing compensation

    Page(s): 327 - 334
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (548 KB)  

    An improved wavelet-based method is developed for extracting pitch information from noisy speech. It uses a modified spatial correlation function which is originally applied to wavelet-based signal denoising to improve the performance of pitch detection in a noisy environment. The modified spatial correlation function needed in the proposed pitch detection method makes use of an aliasing compensation algorithm to eliminate the aliasing distortion that arises from the downsampling and upsampling operations of the wavelet transform. As a consequence, this allows one to further increase the accuracy of pitch detection. It is shown in various experimental results that this new method gives a considerable performance improvement when compared with other conventional and wavelet-based methods. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 41. Source combined linear predictive analysis in pulse-based speech coders

    Page(s): 143 - 148
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (480 KB)  

    The conventional linear predictive analysis in pulse-based coders is replaced by the so-called source-combined linear predictive method to match the excitations considered in two steps: synthesis filter determination and excitation search. It differs from Atal's (1982) two-pass approach in that the synthesis filter is optimised jointly with the excitation prior to the analysis-by-synthesis (ABS) excitation search. However, it is a difficult task to obtain the optimum solution, and, thus, a suboptimal algorithm is developed. Initially, the algorithm starts with the covariance method and then corrects the synthesis filter using an estimate of the excitation. This is accomplished by using two coupled equations originally developed. The effectiveness of the approach in multipulse and regular pulse excited coders is demonstrated. Extensive simulation results, at several bit rates, with different excitations, are presented. Comparisons are made with the standard coder and the coder that employs Atal's approach. In all conditions, the proposed coder is found to give better results View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 42. Region-growing approach to colour segmentation using 3D clustering and relaxation labelling

    Page(s): 270 - 276
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1763 KB)  

    The paper presents a new segmentation algorithm for colour images based on a series of region growing and merging processes. This algorithm starts with the region growing process, which groups pixels into homogeneous regions by combining 3D clustering and relaxation labelling techniques. Each resulting small region is then merged to the region which is the nearest to it in terms of colour similarity and spatial proximity. One problem with region growing is its inherent dependency on the selection of the seed region, which can be avoided by using the relaxation labelling technique. Experimental results are presented to demonstrate the performance of the new method in terms of better segmentation and less sensitivity to noise, and in terms of computational efficiency. The segmentation results using the fuzzy c-means technique, the competitive learning neural network and a region growing and merging algorithm are also presented for comparison purposes. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 43. Variable-step-size LMS algorithm: new developments and experiments

    Page(s): 311 - 317
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (444 KB)  

    The variable-step-size least-mean-square (VSLMS) algorithm is explored and adopted for tracking of time-varying environments. Two implementations of the VSLMS algorithm are proposed. The emphasis is on implementations sizes with different step sizes at various taps of the adaptive filter. General analysis of the VSLMS algorithm appears to be somewhat involved. However, for one implementation a limited analysis of the algorithm is found possible. For this implementation it is shown that, when the input samples to the adaptive filter are zero-mean, Gaussian and uncorrelated with one another, the VSLMS algorithm can adapt itself to select the optimum set of step sizes which results in the best-tracking performance. Simulation experiments with the VSLMS algorithm show that, under fairly mild conditions, both of the proposed implementations adapt toward the optimum step sizes View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 44. Thresholding based on histogram approximation

    Page(s): 271 - 279
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (972 KB)  

    The authors propose two automatic threshold-selection schemes, based on functional approximation of the histogram. The first method is based on minimising the sum of square errors, and the second one is based on minimising the variance of the approximated histogram. Experimental results show that, on average, the latter scheme gives better results than the former one, at a small extra computational cost. A `goodness' measure is proposed to measure the effectiveness of the two schemes, and to compare them against the entropy-based approach and the moment-based approach View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 45. Colour image filters based on hypercomplex convolution

    Page(s): 89 - 93
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (560 KB)  

    There are very few examples of true vector filters known for colour images. The authors introduce a new class of filter based on convolution with hypercomplex masks, and present three colour edge detecting filters inspired by the Prewitt, Sobel and Kirsch filters. These filters, when applied to a colour image, produce an almost greyscale image with colour edges where the original image had a sharp change of colour. They rely on a three-space rotation about the grey line of RGB space, implemented with the rotation operator R[]R-1 where R is a quaternion and R-1 is its inverse. Pixels of similar colour corresponding to opposing positions in the filter masks cancel to give a grey or monochromatic result, while pixels of different colour are differenced in a vector sense to give coloured edge pixels. The linearity of the new filters is discussed and the paper concludes with a discussion of the impulse response of a hypercomplex filter View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 46. Robust M-estimate adaptive filtering

    Page(s): 289 - 294
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (528 KB)  

    An M-estimate adaptive filter for robust adaptive filtering in impulse noise is proposed. Instead of using the conventional least-square cost function, a new cost function based on an M-estimator is used to suppress the effect of impulse noise on the filter weights. The resulting optimal weight vector is governed by an M-estimate normal equation. A recursive least M-estimate (RLM) adaptive algorithm and a robust threshold estimation method are derived for solving this equation. The mean convergence performance of the proposed algorithm is also analysed using the modified Huber (1981) function (a simple but good approximation to the Hampel's three-parts-redescending M-estimate function) and the contaminated Gaussian noise model. Simulation results show that the proposed RLM algorithm has better performance than other recursive least squares (RLS) like algorithms under either a contaminated Gaussian or alpha-stable noise environment. The initial convergence, steady-state error, robustness to system change and computational complexity are also found to be comparable to the conventional RLS algorithm under Gaussian noise alone View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 47. Variable digital filter design using the outer product expansion

    Page(s): 123 - 128
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (348 KB)  

    Digital filters with adjustable frequency domain characteristics are referred to as variable digital filters. Variable filters are useful in the applications where the filter characteristics are required to be changeable during the course of signal processing. Especially in real time applications, variable filters are needed to change their coefficients instantaneously such that the real time signal processing can be performed. The present paper proposes a very efficient technique for variable 1D digital filter design. Generally speaking, the variable coefficients of variable digital filters are multidimensional functions of a set of spectral parameters which define the desired frequency domain characteristics. The authors first sample the given variable 1D magnitude specification and use the samples to construct a multidimensional array, then propose an outer product expansion method for expanding the multidimensional array as the sum of outer products of 1D arrays (vectors). Based on the outer product expansion, one can reduce the difficult problem of designing a variable 1D digital filter to the easy one that only needs constant 1D filter designs and 1D polynomial approximations. The technique can obtain variable 1D filters having arbitrary desired magnitude characteristics with a high design accuracy View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 48. Novel fuzzy reinforced learning vector quantisation algorithm and its application in image compression

    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (824 KB)  

    A new approach to the design of optimised codebooks using vector quantisation (VQ) is presented. A strategy of reinforced learning (RL) is proposed which exploits the advantages offered by fuzzy clustering algorithms, competitive learning and knowledge of training vector and codevector configurations. Results are compared with the performance of the generalised Lloyd algorithm (GLA) and the fuzzy K-means (FKM) algorithm. It has been found that the proposed algorithm, fuzzy reinforced learning vector quantisation (FRLVQ), yields an improved quality of codebook design in an image compression application when FRLVQ is used as a pre-process. The investigations have also indicated that RL is insensitive to the selection of both the initial codebook and a learning rate control parameter, which is the only additional parameter introduced by RL from the standard FKM View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 49. Efficient multilevel successive elimination algorithms for block matching motion estimation

    Page(s): 73 - 84
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (690 KB)  

    The authors present fast algorithms to reduce the computations of block matching algorithms, for motion estimation in video coding. Efficient multilevel successive elimination algorithms are based on the multilevel successive elimination. Efficient multilevel successive elimination algorithms consist of four algorithms. The first algorithm is given by the sum of absolute difference between the sum norms of sub-blocks in a multilevel successive elimination algorithm (MSEA) using the partial distortion elimination technique. By using the first algorithm, computations of MSEA can be reduced further. In the second algorithm, the sum of absolute difference (SAD) is calculated adaptively from large value to small value according to the absolute difference values between pixels of blocks. By using the second algorithm, the partial distortion elimination in the SAD calculation can occur early, therefore the computations of MSEA can be reduced. The second algorithm is useful not only with MSEA, but also with all kinds of block matching algorithms. In the third algorithm, the elimination level of the MSEA can be estimated. Accordingly, the computations of the MSEA related to the level lower than the estimated level can be reduced. The fourth algorithm is, first of all, to search the motion vector over the half sampled search points. At the second search, the authors search the unsampled search points around the tested search points where the motion vector may exist from the first search results. The motion estimation accuracy of the fourth algorithm is nearly 100% and the computations can be greatly reduced. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 50. Overview of compression and packet loss effects in speech biometrics

    Page(s): 372 - 376
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (222 KB)  

    An overview is presented of compression and packet loss effects in speech biometrics. These new problems appear particularly in recent applications of biometrics over mobile or Internet networks. The influence of speech compression on speaker recognition performance in mobile networks is investigated. In a first experiment, it is found that the use of GSM coding degrades the performance. In a second experiment, the features for the speaker recognition system are calculated directly from the information available in the encoded bit stream. It is found that a low LPC order in GSM coding is responsible for most performance degradations. A speaker recognition system was obtained which is equivalent in performance to the original one which decodes and reanalyses speech before performing recognition. The joint packet loss and compression effects over IP networks are also studied. It is experimentally demonstrated that the adverse effects of packet loss alone are negligible, while the encoding of speech, particularly at a low bit rate, coupled with packet loss, can reduce the verification accuracy considerably. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.