By Topic

Non-Linear Signal and Image Processing (Ref. No. 1998/284), IEE Colloquium on

Date 22 May 1998

Filter Results

Displaying Results 1 - 15 of 15
  • Morphological properties of rank ordering filters

    Page(s): 2/1 - 210
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1320 KB)  

    When performing tasks such as image segmentation and region coding, it is important to preserve the shape of the regions as closely as possible to those of the original object shapes. For the purposes of data compression it may then be prudent to allow some simplification of the region boundaries in a controlled manner (Beaumont 1996). To improve the segmentation of the image it is usual to filter the image to remove noise, and limit the generation of unnatural regions. The author found that the standard methods for spatial noise filtering such as low pass filtering and median filtering proved unsatisfactory. This is because they also filtered the shape of the region boundaries. A set of nonlinear filters were developed that preserved more of the structure of region boundaries. The constraints of the project also required the filters to be local (i.e mask based) and perform in real time (i.e. computationally fast). While developing the above filters the author uncovered the following 2 classes of filters which did not seem to be well documented in the literature: rank ordering clipping filters-nonlinear low pass; and low deviation filters-nonlinear edge enhancement View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Video and film restoration using mathematical morphology

    Page(s): 3/1 - 3/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (780 KB)  

    All those who have visited the cinema will be familiar with the problem of film dirt, i.e. dark artefacts corrupting the images viewed. It is obviously desirable to remove this corruption, especially if one wishes to sell digital versions of the film. This type of film dirt is only present in the luminance fields of the sequence. So, in order to remove this corruption, a suitable filter is applied only to the luminance field of each frame in the sequence. However, it is necessary to determine exactly which filter to use. In this paper we consider the class of soft morphological filters (Koskinen and Astola 1994). But, it is still necessary to determine which, out of the whole class of soft morphological filters, is the optimal filter for this application, i.e. the filter that will be able to remove most noise whilst preserving the underlying structure. This topic is discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Reducing chromatic grain noise in film sequences

    Page(s): 5/1 - 5/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (304 KB)  

    Grain noise is one of the most common distortions in cinematographic film sequences and is caused by the crystal structure of the chemical coating of the film material. The colour sensitive crystals can be considered as three separate populations. Thus noise in the three channels is uncorrelated and similarly noise between frames is uncorrelated. Conversely, the signal (i.e. the projected view volume) is highly correlated between channels and over time. We explore methods of using this constraint to reduce noise within an adaptive filter framework using the popular Widrow-Hopf LMS algorithm. As a film sequence typically includes many moving elements such as actors on a moving background, motion estimation techniques are used to eliminate as much as possible the effect of grey-level variations on the adaptive filter. An optical flow technique is used to extract pixel motions prior to the application of the noise reduction View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image distortions produced by mean, median and mode filters

    Page(s): 6/1 - 6/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (360 KB)  

    Median filters are probably the most widely used of the noise suppression filters that are commonly applied to digital images. They are vastly superior to mean filters since the latter are well known to blur images as well as removing noise. Median filters do not suffer from this disadvantage. Although median filters do not blur images, it has been found that they can cause a certain amount of image distortion by shifting edges (Davies 1997). This paper reviews previous work on the computation of shifts produced by mean, median and mode filters using continuum models. The paper proceeds to show how, in the case of median filters, it can be extended using discrete models to give more accurate predictions that are in line with experimental measurements View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Measuring noise in colour images

    Page(s): 8/1 - 8/4
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (196 KB)  

    Algorithms exist for removing unwanted noise from colour images; however little has been said about how to actually quantify the effect of these algorithms or generally how to quantify the amount of noise in a colour image. The standard methods include calculating the normalised mean square error (NMSE) between the original and the filtered image, calculating the mean chromaticity error (MCRE) between the original and the filtered image and of course visual inspection. The problem of MCRE and NMSE is that they only quantify the differences between two images-they are not a specific measure for noise in an image. Visual inspection is problematic because it is subjective. In this paper we propose a novel method for specifically calculating the amount of noise in a colour image. We give a general definition of how pixels could be classified as `noisy' based on well-established vector processing methodologies. We then describe how the sum of these `noisy' pixels can then be calculated and normalised to give a general measure of the percentage of neighbourhoods containing noisy pixels, which can then be used as a non-subjective estimate of the amount of noise in a colour image View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • IEE Colloquium on Non-Linear Signal and Image Processing (Ref. No.1998/284)

    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (72 KB)  

    The following topics were dealt with: non-linear processing technique for coherent interference artefacts removal; rank ordering filters morphological properties; video and film restoration; condition monitoring; Volterra nonlinear systems identification using circular inputs; advanced PRNN based non-linear prediction/system identification; hybrid non-linear filters for locating speckled contaminants in grain; FPGAs; Isaac Newton programme; film sequences chromatic grain noise reduction; mean, median and mode filters image distortions; non-linear signal processing; colour image noise measurement; Bayesian approach to signal modelling View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The Bayesian approach to signal modelling

    Page(s): 9/1 - 9/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (336 KB)  

    In this paper, an introduction to Bayesian methods in signal processing is given. The paper starts by considering the important issues of model selection and parameter estimation and derives analytic expressions for the model probabilities of two simple models. The idea of marginal estimation of certain model parameters is then introduced and expressions are derived for the marginal probability densities for frequencies in white Gaussian noise and a Bayesian approach to general change point analysis is given. Numerical integration methods are introduced based on Markov chain Monte Carlo techniques and the Gibbs sampler in particular View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Identification of Volterra nonlinear systems using circular inputs

    Page(s): 10/1 - 10/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (284 KB)  

    It is shown that the minimum-mean square estimate of the parameters of a Volterra model can be obtained in closed form when the input is circularly symmetric. The proposed method is applicable to systems with arbitrary orders and memory. Theoretical results are illustrated via Monte-Carlo simulations View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Two applications of simple non-linear signal processing

    Page(s): 7/1 - 7/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (380 KB)  

    This paper describes two different applications of nonlinear signal processing. In both cases, computational simplicity was a major goal and a nonlinear approach significantly increased performance. The two applications (PSK demodulation of a noisy signal from sign-only data, and frequency estimation of a single complex tone) are both of practical importance. It is also hoped that these two examples will encourage designers of real systems to investigate non-linear approaches. Finally, the examples may be of interest to theoreticians since in neither case has a full theoretical analysis been completed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Advanced PRNN based nonlinear prediction/system identification

    Page(s): 11/1 - 11/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (328 KB)  

    Insight into the core of the pipelined recurrent neural network (PRNN) in prediction applications is provided. It is shown that modules of the PRNN contribute to the final predicted value at the output of the PRNN in two ways, namely through the process of nesting, and through the process of learning. A measure of the influence of the output of a distant module to the amplitude at the output of the PRNN is analytically found, and the upper bound for it is derived. Furthermore, an analysis of the influence of the forgetting factor in the cost function of the PRNN to the process of learning is undertaken, and it is found that for the PRNN, the forgetting factor can even exceed unity in order to obtain the best predictor. Simulations on three speech signals support that approach, and outperform the other stochastic gradient based schemes View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Condition monitoring with non-linear signal processing

    Page(s): 4/1 - 4/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (352 KB)  

    Condition monitoring plays a vital and indispensable role in the operation of any type of plant or machinery. In general, condition monitoring is achieved by processing and interpreting some characteristic signal from the plant under consideration. Traditionally, only linear second order statistical analysis has been used to process the characteristic signals from plant, however, a new approach to condition monitoring called higher order statistics is becoming increasingly prevalent. A small body of research has been concentrated on using higher order statistical techniques within a condition monitoring environment but the success of this work has been hampered by the lack of generality of the technique. This submission investigates a method of capitalising on the attractive properties of such techniques in an intelligent and simple fashion in the construction of a comprehensive, generalised condition monitoring tool. This is demonstrated by practical investigation of the condition monitoring of an induction machine View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Nonlinear and nonstationary signal processing

    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (80 KB)  

    Presents a brief discussion of the need for and evolution to nonlinear and nonstationary signal processing. Applications where these are useful are mentioned View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fast implementation of non-linear filters using FPGAs

    Page(s): 13/1 - 13/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (396 KB)  

    Most adaptive image and signal processing tasks are performed on specialist digital signal processing chips. These devices are highly optimised for efficient computation of the core multiply and accumulate operations required by current algorithms. Attempts to synthesise these types of algorithms on FPGAs have resulted in few competitive implementations. FPGAs generally fail to realise efficient arithmetic functions except in the most constrained cases such as constant coefficient multipliers. The approach adopted in this paper is based on the use of stack filters that avoid these difficulties by employing logical algorithms that do not rely on any arithmetic functions View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A nonlinear processing technique for removing coherent interference artefacts

    Page(s): 1/1 - 1/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (568 KB)  

    Ultrasound pulses propagating through human tissue appear to retain most of their initial coherence, and are coherently scattered from the many inhomogeneities within a tissue. A complex echo field is generated which exhibits many interference effects, the most familiar of these manifests itself as the ubiquitous speckle artefact. Speckle pervades almost all medical ultrasound pulse-echo signals and imposes a fundamental limit on signal and image quality. It is commonly assumed that the removal of speckle will produce a great advantage in a large number of practical applications. The novel approach developed here provides a general descriptive framework for interference effects, and is based on a description of interference by the presence of what we have termed structure zeros in the analytic continuation of the real data into the complex frequency, and complex time domains. The structure zeros may be uniquely identified if the form of the interrogating ultrasound pulse is precisely known. In practice, the latter requirement cannot be satisfied, and the presence of noise introduces a further element of uncertainty, but the structure zeros which make the dominant contribution to signal corruption may be unambiguously identified (via a sensitivity index) when short data segments are considered. Appropriate manipulation of the structure zero locations results in a specific, desired correction to the signal View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hybrid non-linear filters for locating speckled contaminants in grain

    Page(s): 12/1 - 12/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (464 KB)  

    This paper discusses the possible application of morphological operators to the inspection of grain. The particular task that is involved is that of locating contaminants such as rodent droppings. In the following section we describe the background to the problem and show the result of applying morphological operators and nonlinear filters to grain images in order to locate contaminants. We analyse the situation in more detail and show quite where the basic morphological approach needs supplementation View full abstract»

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