Abstract: A systematic method for the detection and segmentation of microcalcifications in mammograms is presented. It is important to preserve size and shape of the individual calcifications as exactly as possible. A reliable diagnosis requires both rates of false positives as well as false negatives to be extremely low. The proposed approach uses a two-stage algorithm for spot detection and shape extraction. The first stage applies a weighted difference of Gaussians filter for the noise-invariant and size-specific detection of spots. A morphological filter reproduces the shape of the spots. The results of both filters are combined with a conditional thickening operation. The topology and the number of the spots are determined with the first filter, and the shape by means of the second. The algorithm is tested with a series of real mammograms, using identical parameter values for all images. The results are compared with the judgement of radiological experts, and they are very encouraging. The described approach opens up the possibility of a reproducible segmentation of microcalcifications, which is a necessary precondition for an efficient screening program. >

Abstract: A model for texture analysis and segmentation using multiple oriented channel filters is analyzed in the general framework. Several different arguments are applied leading to the conclusion that the two-dimensional Gabor filters possess strong optimality properties for this task. Properties of the multiple-channel segmentation approach are analyzed. In particular, perturbations of textures from an ideal model are found to have important effects on the segmentation that can usually be ameliorated by simply preceding the segmentation process by a logarithmic operation and using a low-pass postfilter prior to making region assignments. The difficult problems of space-variant textures and multiple component textures are also considered. Local spatial frequency estimation approaches are suggested that use the responses as constraints in estimating the locally emergent texture frequencies. Complex texture aggregates containing multiple shared frequency components can be analyzed if the textures are distinct and few in number. >

Abstract: A generalized algorithm for fuzzy clusterwise regression (GFCR) is proposed that incorporates both benefit segmentation and market structuring within the framework of preference analysis. The metho...

Abstract: A knowledge-based approach for Landsat image segmentation is proposed. The image segmentation problem is solved by extracting kernel information from the input image to provide an initial interpretation of the image and by using a knowledge-based hierarchical classifier to discriminate between major land-cover types in the study area. The proposed method is designed in such a way that a Landsat image can be segmented and interpreted without any prior image-dependent information. The general spectral land-cover knowledge is constructed from the training land-cover data, and the road information of an image is obtained through a road-detection program. >

Abstract: A method for automatic segmentation of speech into phones is described. The incoming utterance is split up into more or less stationary parts, and these stationary parts are labelled as phones using the phonetic transcription of the utterance. An implicit segmentation algorithm splits up the utterance into segments on the basis of the degree of similarity between the frequency spectra of neighboring frames. An explicit algorithm does the same, but on the basis of the degree of similarity between the frequency spectra of the frames in the utterance and reference spectra. A combination algorithm compares the two segmentation results and produces the final segmentation. Automatically determined phone boundaries are compared with manually determined ones. The result of a perception test is described. >

Abstract: Despite the fact that complex visual scenes contain multiple, overlapping objects, people perform object recognition with ease and accuracy. One operation that facilitates recognition is an early segmentation process in which features of objects are grouped and labeled according to which object they belong. Current computational systems that perform this operation are based on predefined grouping heuristics. We describe a system called MAGIC that learns how to group features based on a set of presegmented examples. In many cases, MAGIC discovers grouping heuristics similar to those previously proposed, but it also has the capability of finding nonintuitive structural regularities in images. Grouping is performed by a relaxation network that attempts to dynamically bind related features. Features transmit a complex-valued signal (amplitude and phase) to one another; binding can thus be represented by phase locking related features. MAGIC's training procedure is a generalization of recurrent backpropagation to complex-valued units.

Abstract: Multichannel filtering techniques are presented for obtaining both region- and edge-based segmentations of textured images. The channels are represented by a bank of even-symmetric Gabor filters that nearly uniformly covers the spatial-frequency domain. Feature images are obtained by subjecting each (selected) filtered image to a nonlinear transformation and computing a measure of energy around each pixel. Region-based segmentations are obtained by using a square-error clustering algorithm. Edge-based segmentations are obtained by applying an edge detector to each feature image and combining their magnitude responses. An integrated segmentation technique that combines the strengths of the previous two techniques while eliminating their weaknesses is proposed. The integrated approach is truly unsupervised, since it eliminates the need for knowing the exact number of texture categories in the image. >

Abstract: We present a technique for the construction of multi-scale representations of grey-level images. Unlike conventional representations the scales are discrete as opposed to continuous and their level is solely determined by the data. The technique is based upon connecting singular points in the image with maximum gradient paths. We also describe two segmentation methods which use the maximum gradient paths generated during the construction of the multi-scale representation. In both segmentation techniques the paths are used to determine significant ridges and troughs. The first technique operates directly on the image, while the second technique uses the magnitude of the image derivative.

Abstract: We present a neural network algorithm that simultaneously performs segmentation and recognition of input patterns that self-organizes to detect input pattern locations and pattern boundaries. We demonstrate this neural network architecture on character recognition using the NIST database and report on results herein. The resulting system simultaneously segments and recognizes touching or overlapping characters, broken characters, and noisy images with high accuracy.

Abstract: A split-and-merge algorithm is proposed for the segmentation of the digitized surface of a range image into planar regions. The geometric data structure used is a triangular tessellation of image domain. This data structure, combined with an adaptive surface approximation technique, allows a better adaptation of the range image segmentation to the surface boundaries. It also provides an efficient neighborhood referencing mechanism, thus resulting in a fast algorithm. >

Abstract: The continuous model of an image to segment can be used to put the segmenting problem as that of the evolution of a 2D spline curve or 3D surface under strengths comparable to those used in “active contours” or “snakes” (hence the name of “ snake-spline ”). The advantage of this method is that this evolution can be modelled with a simple differential system, the variables of which are the coefficients of the B-splines representing the frontier. Moreover, this method is by nature adaptative : it is easy to control the number of B-splines used to represent the frontier, which makes it possible to improve progressively the result of the segmentation. Besides, a priori knowledge can easily be taken into account if provided in the form of a CAD model of the object to segment.

Abstract: A technique for constructing shape representation from images using free-form deformable surfaces is presented. The authors model an object as a closed surface that is deformed subject to attractive fields generated by input data points and features. Features affect the global shape of the surface, while data points control its local shape. This approach is used to segment objects even in cluttered or unstructured environments. The algorithm is general in that it makes few assumptions on the type of features, the nature of the data, and the type of objects. Results for a wide range of applications are presented: reconstruction of smooth isolated objects such as human faces, reconstruction of structured objects such as polyhedra, and segmentation of complex scenes with mutually occluding objects. The algorithm has been successfully tested using data from different sensors including grey-coding range finders and video cameras, using one or several images. >

Abstract: A new segmentation-based image coding method is proposed. The encoder recursively partitions an image into convex n-gons, 3 >

Abstract: A novel image segmentation technique is presented which combines region growing, edge detection, and a novel edge preserving smoothing algorithm. The combined method helps to avoid characteristic segmentation errors which occur when using region growing or edge detection separately. The method is applied to segment MRI (magnetic resonance imaging) images for subsequent 3D visualization, and experimental results are presented. >

Abstract: A novel method for outdoor natural color image segmentation for road following is presented. From the results of experiments it can be seen that the hue, saturation, and intensity (HSI) system should be used rather than the red, green, blue (RGB) system, and segmentation can be done in S-I space with good performance. An automatic adaptive threshold selection method is developed. The preliminary results indicate the effectiveness and efficiency of this method. >

Abstract: Publisher Summary This chapter discusses the patterning of body segments of the zebrafish embryo. Descriptive studies in zebrafish have provided exceptionally clear information, often at the level of individual cells, about the structure, extent in the body, and development of the segments. Experimental analyses, including the use of mutations, reveal cellular interactions required for segmentation. Information from zebrafish provides a useful background for understanding how genes make vertebrate segments, an issue for the future, and for which this species also holds some promise. The developmental studies of zebrafish strengthen the notion that vertebrates seem to be metameric creatures. Attempts to dispel this notion in the past have pointed up the limited extent of vertebrate body segmentation. Segmentation does indeed seem to be limited, but it is even limited in annelids that are certainly segmented animals and that show no hint of segmental organization in lineages that produce the gut or the skin, organs that also appear unsegmented in vertebrates. Accordingly, the presence of nonsegmented tissues within a segmented body plan does not seem particularly problematic for the common segmented ancestor hypothesis; a most interesting question for both development and evolution is how the apparently unsegmented axial sets of cells, the notochord and the floor plate, come to be insinuated into the metameres. These axial tissues have dominant developmental roles and certainly have also been extremely important in chordate evolution.

Abstract: Approaches the texture segmentation problem by clustering feature vectors created from a Gabor transform data block. Given an N*N image, the authors compute 24 Gabor transforms using Gabor kernels with six orientations and four sizes. This results in a Gabor data block composed of N/sup 2/ feature vectors of length 24. The feature vectors are then grouped based on their distribution in the high-dimensional feature space. The authors hypothesize that the pixels in a given group have similar characteristics, and thus are part of the same texture. Experimental results for segmenting a synthetic railroad track image were encouraging; a clear-cut segmentation of the image was obtained. By clustering Gabor features, the authors were able to segment an image into regions of uniform texture without prior knowledge of the types of texture, or the frequency and orientation characteristics of these textures. The clustering algorithm is a modified Kohonen self-organizing feature map. >

Abstract: Several algorithms used for segmenting highly speckled high-resolution synthetic-aperture-radar (SAR) complex data into spatially and radiometrically homogeneous regions are presented. The procedure is based on two models, one for the speckled complex amplitudes and the other for the regions. The first model uses the physics of the SAR imaging and processing system to characterize the statistics of speckle, while the second model uses a Markov random field to describe the statistics of the regions. Based on the combination of these two models from Bayes theory, two possible optimality criteria are considered for the segmentation of the complex data into regions. The different algorithms are implemented on a parallel optimization network. Results from both simulated and actual SAR complex data are presented for a comparison of the different alternatives and evaluation of the performance of the segmentation techniques.

Abstract: Image segmentation is an important area in the general field of image processing and computer vision. It is a fundamental part of the `low level' aspects of computer vision and has many practical applications such as in medical imaging, industrial automation and satellite imagery. Traditional methods for image segmentation have approached the problem either from localisation in class space using region information, or from localisation in position, using edge or boundary information. More recently, however, attempts have been made to combine both region and boundary information in order to overcome the inherent limitations of using either approach alone. In this thesis, a new approach to image segmentation is presented that integrates region and boundary information within a multiresolution framework. The role of uncertainty is described, which imposes a limit on the simultaneous localisation in both class and position space. It is shown how a multiresolution approach allows the trade-off between position and class resolution and ensures both robustness in noise and efficiency of computation. The segmentation is based on an image model derived from a general class of multiresolution signal models, which incorporates both region and boundary features. A four stage algorithm is described consisting of: generation of a low-pass pyramid, separate region and boundary estimation processes and an integration strategy. Both the region and boundary processes consist of scale-selection, creation of adjacency graphs, and iterative estimation within a general framework of maximum a posteriori (MAP) estimation and decision theory. Parameter estimation is performed in situ, and the decision processes are both flexible and spatially local, thus avoiding assumptions about global homogeneity or size and number of regions which characterise some of the earlier algorithms. A method for robust estimation of edge orientation and position is described which addresses the problem in the form of a multiresolution minimum mean square error (MMSE) estimation. The method effectively uses the spatial consistency of output of small kernel gradient operators from different scales to produce more reliable edge position and orientation and is effective at extracting boundary orientations from data with low signal-to-noise ratios. Segmentation results are presented for a number of synthetic and natural images which show the cooperative method to give accurate segmentations at low signal-to-noise ratios (0 dB) and to be more effective than previous methods at capturing complex region shapes.

Abstract: A novel texture segmentation technique for both supervised and unsupervised segmentation is presented. The textured images under study are modeled by a proposed hierarchical Markov random field (MRF) model. This model is formed by combining the binomial model for textures and the multilevel logistic model for region distributions. The supervised segmentation is achieved by a novel algorithm which can reach the global maxima of the posteriori distribution even if the textures are modeled by an MRF model. For unsupervised segmentation, a novel parameter estimation scheme is proposed for estimating the model parameters directly from a given image. The proposed technique is verified by a variety of textured images, such as synthesized textures, natural textures, and aerial images, in both the supervised and unsupervised segmentation cases. >

Abstract: The development of techniques for interpreting the structure of three-dimensional images, f ( x , y , z ), is useful in many applications. A key initial stage in the signal to symbol conversion process, essential for the interpretation of the data, is three-dimensional image segmentation involving the processes of partitioning and identification . Most segmentation and grouping research in computer vision has addressed partitioning of 2D images, f ( x , y ). In this paper, we present a parallel 3D image segmentation algorithm which, through the use of α- partitioning and volume filtering , segments 3D images such that the greylevel variation within each volume can be described by a regression model . Experimental results demonstrate the effectiveness of this algorithm on several real-World 3D images.

Abstract: We propose a unified approach to solving low, intermediate, and high level computer vision problems for 3D object recognition from range images. All three levels of computation are cast in an optimization framework and can be implemented on neural network style architectures. In the low level computation, the task is to estimate curvature images from the input range data. Subsequent processing at the intermediate level is concerned with segmenting these curvature images into coherent curvature sign maps. In the high level computation, image features are matched against model features based on an object description called an attributed relational graph (ARG). We show that the above computational tasks at the three different levels can all be formulated as optimizing a two-term energy function. The first term encodes unary constraints while the second term encodes binary ones. These energy functions are minimized using parallel and distributed relaxation-based algorithms which are well suited for neural network implementation. Some experimental results are presented for curvature-based segmentation, ARG matching, and 3D Surface matching.

Abstract: This paper describes an approach, called centered object integrated segmentation and recognition (COISR), for integrating object segmentation and recognition within a single neural network. The application is hand-printed character recognition. Two versions of the system are described. One uses a backpropagation network that scans exhaustively over a field of characters and is trained to recognize whether it is centered over a single character or between characters. When it is centered over a character, the net classifies the character. The approach is tested on a dataset of hand-printed digits. Very low error rates are reported. The second version, COISR-SACCADE, avoids the need for exhaustive scans. The net is trained as before, but also is trained to compute ballistic 'eye' movements that enable the input window to jump from one character to the next.