By Topic

Vision, Image and Signal Processing, IEE Proceedings -

Issue 1 • Date Feb 1995

Filter Results

Displaying Results 1 - 11 of 11
  • New separable transform

    Page(s): 27 - 30
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (176 KB)  

    A separable version in two dimensions of the transform previously presented by Boussakta and Holt (1992) is introduced and generalised to the multi-dimensional case. This transform is fast and symmetric, has the convolution property and is separable, making if a good candidate for the calculation of two-dimensional convolutions and correlations for image processing purposes View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Texture classification using a spatial-point process model

    Page(s): 1 - 6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (580 KB)  

    A Bayesian statistical classifier for the segmentation of texture is presented, which models the quantised image data as a set of independent spatial Poisson processes. Two data sets are examined, namely Gaussian white noise textures, and textures contained in a sidescan sonar image of the seabed. The Poisson model is demonstrated to be applicable in both these cases, and a maximum likelihood discriminant function is developed. Finally, results are presented for the classification of both data sets View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Unusual-length number-theoretic transforms using recursive extensions of Rader's algorithm

    Page(s): 31 - 34
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (248 KB)  

    A novel decomposition of NTT block-lengths is proposed using repeated applications of Rader's (1968) algorithm to reduce the problem to that of realising a single small-length NTT. An efficient implementation of this small-length NTT is achieved by an initial basis conversion of the data, so that the new basis corresponds to the kernel of the small-length NTT. Multiplication by powers of the kernel become rotations and all arithmetic is efficiently performed within the new basis. More generally, this extension of Rader's algorithm is suitable for NTT or DFT applications where an efficient implementation of a particular small-length NTT/DFT module exists View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Isolated Mandarin syllable recognition using segmental features

    Page(s): 59 - 64
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (508 KB)  

    A segment-based speech recognition scheme is proposed. The basic idea is to model explicitly the correlation among successive frames of speech signals by using features representing contours of spectral parameters. The speech signal of an utterance is regarded as a template formed by directly concatenating a sequence of acoustic segments. Each constituent acoustic segment is of variable length in nature and represented by a fixed dimensional feature vector formed by coefficients of discrete orthonormal polynomial expansions for approximating its spectral parameter contours. In the training, an automatic algorithm is proposed to generate several segment-based reference templates for each syllable class. In the testing, a frame-based dynamic programming procedure is employed to calculate the matching score of comparing the test utterance with each reference template. Performance of the proposed scheme was examined by simulations on multi-speaker speech recognition for 408 highly confusing isolated Mandarin base-syllables. A recognition rate of 81.1% was achieved for the case using 5-segment, 8-reference template models with cepstral and delta-cepstral coefficients as the recognition features. It is 4.5% higher than that of a well-modelled 12-state, 5-mixture CHMM method using cepstral, delta cepstral, and delta-delta cepstral coefficients View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fast computation of the two-dimensional generalised Hartley transforms

    Page(s): 35 - 39
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (280 KB)  

    The two-dimensional generalised Hartley transforms (2-D GDHTs) are various half-sample generalised DHTs, and are used for computing the 2-D DHT and 2-D convolutions. Fast computation of 2-D GDHTs is achieved by solving (n1+(n01/2))k1+(n2+(n02 /2))k2=(n+(½))k mod N, n01, n02 =1 or 0. The kernel indexes on the left-hand side and on the right-hand side belong to the 2-D GDHTs and the 1-D H3, respectively. This equation categorises N×N-point input into N groups which are the inputs of a 1-D N-point H3. By decomposing to 2-D GDHTs, an N×N-point DHT requires a 3N/2i 1-D N/2i-point H3, i=1, ..., log2N-2. Thus, it has not only the same number of multiplications as that of the discrete Radon transform (DRT) and linear congruence, but also has fewer additions than the DRT. The distinct H 3 transforms are independent, and hence parallel computation is feasible. The mapping is very regular, and can be extended to an n-dimensional GDHT or GDFT easily View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dynamic partial search scheme for stochastic codebook of FS1016 CELP coder

    Page(s): 52 - 58
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (460 KB)  

    The authors present a new classified dynamic partial search structure for the stochastic codebook of the FS1016 CELP coder to replace the fixed partial search for selecting the best excitation vector of the stochastic codebook. In the proposed scheme, the conventional one-stage stochastic codebook search is substituted with a two-stage dynamic method for reducing the computational complexity without degrading the voice quality. The establishment of this structure is based on two classifiers, one for the line spectrum pairs (LSP) of the input signals, and the other for the autocorrelation coefficients (AC) of the stochastic codebook search target. In addition, the stochastic codebook is classified into K subcodebooks, and with these two classifiers it is possible to determine dynamically which subcodebook needs to be searched. This method achieves a reduction in the search procedure by a factor of 2-8. The efficiency of these two classifiers is discussed and the comparison of the performance between the fixed partial search and the proposed technique is also addressed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Modelling of acoustic transfer functions for echo cancellers

    Page(s): 47 - 51
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (360 KB)  

    ARMA modelling (infinite impulse response filters having both poles and zeros), rather than MA modelling, of acoustic transfer functions is potentially capable of reducing the number of coefficients required to achieve a given level of echo cancellation. The amount of reduction obtainable, however, has received little attention in the literature. To arrive at quantitative figures for the complexity reduction possible, equation error and output error ARMA and MA modelling of acoustic impulse responses are compared experimentally. It is shown, for both fullband and subband implementations, that the potential for reducing the number of coefficients required is small View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parallel-decomposition algorithm for discrete computational problems and its application in developing an efficient discrete convolution algorithm

    Page(s): 40 - 46
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (372 KB)  

    The authors present a parallel-decomposition algorithm for discrete computational problems. An efficient convolution algorithm is developed under the proposed decomposition algorithm. The decomposition operation is based on integer modular arithmetic. Congruent sets are generated from integer modular operation over the index of the problem and constitute a partition. The partition is used to decompose the problem into several small subproblems. The partition under the integer modular operation is unique. Complete reconstruction of the problem is guaranteed by the uniqueness of the partition. Since the algorithm is established on the foundation of all algebraic systems, i.e. number and set theories, it is highly problem-independent, and provides an uniform approach to parallel decomposition of general problems on the discrete domain. Because the subproblems are highly regular, it is suitable for VLSI hardware implementation. The computational complexity of the developed convolution algorithm is reduced by a factor of p2, and (p2)n for n-dimensional problems (p can be any common factor of the input sequences). Compared to the popular FFT methods, where round-off error is introduced, the convolution result from the algorithm is exact. In addition, it does not have restrictions on the length of the input sequences, as in the well known block convolution algorithm. As a result, the algorithm is highly efficient for convolution of large input sequences or correlation operations View full abstract»

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

    Page(s): 7 - 14
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (476 KB)  

    A hierarchical line and segment extraction algorithm, based on a pyramid, is described. Initially, lines are detected in small windows using the Hough transform. The detected lines are then merged using a distance criteria thus avoiding a reaccumulation process at each level of the pyramid. The hierarchical merging process is efficiently performed on lines rather than on segments (since there are many more segments than lines). The detected lines are broken into segments, at the top of the pyramid. The proposed approach is compared to similar approaches based on hierarchical feature extraction. The authors show that their approach combines the advantages of other works and avoids their drawbacks such as quantisation effect and lack of robustness View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Development of simple orthogonal transforms for image compression

    Page(s): 22 - 26
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (308 KB)  

    A new family of orthogonal transform called dyadic transform (DT) is proposed for image compression by using the principle of dyadic symmetry. Two order-8 DTs are generated and their implementation is very simple in that it only requires operations of addition and binary shift. Performance evaluations using a statistical model and real images show that the performance of the proposed transforms is very close to that of DCT View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image segmentation for compression of images and image sequences

    Page(s): 15 - 21
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (524 KB)  

    Region-based image coding schemes, the so-called second generation techniques, have gained much favour in recent years. For still picture coding, they can increase the compression ratio obtained by transform coding by an order of magnitude, while maintaining adequate image representation. The success of these techniques relies on the ability to describe regions in an image succinctly by their shape and size. The algorithms presented describe methods for segmenting images. Unlike most other region based algorithms, these algorithms incorporate knowledge of the border coding process in deciding how to partition the image. The extension from single image compression to sequential image compression is also considered. A new, efficient segmentation scheme is proposed that exploits temporal redundancies between successive images, and reduces some problems associated with error accumulation in error images View full abstract»

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