Showing papers on "Segmentation-based object categorization published in 1995"

Markov random field models for unsupervised segmentation of textured color images

Abstract: We present an unsupervised segmentation algorithm which uses Markov random field models for color textures. These models characterize a texture in terms of spatial interaction within each color plane and interaction between different color planes. The models are used by a segmentation algorithm based on agglomerative hierarchical clustering. At the heart of agglomerative clustering is a stepwise optimal merging process that at each iteration maximizes a global performance functional based on the conditional pseudolikelihood of the image. A test for stopping the clustering is applied based on rapid changes in the pseudolikelihood. We provide experimental results that illustrate the advantages of using color texture models and that demonstrate the performance of the segmentation algorithm on color images of natural scenes. Most of the processing during segmentation is local making the algorithm amenable to high performance parallel implementation. >

Adaptive image segmentation using a genetic algorithm

Abstract: We present the first closed loop image segmentation system which incorporates a genetic algorithm to adapt the segmentation process to changes in image characteristics caused by variable environmental conditions such as time of day, time of year, clouds, etc. The segmentation problem is formulated as an optimization problem and the genetic algorithm efficiently searches the hyperspace of segmentation parameter combinations to determine the parameter set which maximizes the segmentation quality criteria. The goals of our adaptive image segmentation system are to provide continuous adaptation to normal environmental variations, to exhibit learning capabilities, and to provide robust performance when interacting with a dynamic environment. We present experimental results which demonstrate learning and the ability to adapt the segmentation performance in outdoor color imagery.

Medical image segmentation using topologically adaptable surfaces

Abstract: This paper presents a technique for the segmentation of anatomic structures in medical images using a topologically adaptable snakes model. The model is set in the framework of domain subdivision using simplicial decomposition. This framework allows the model to maintain all of the strengths associated with traditional snakes while overcoming many of their limitations. The model can flow into complex shapes, even shapes with significant protrusions or branches, and topological changes are easily sensed and handled. Multiple instances of the model can be dynamically created, can seamlessly split or merge, or can simply and quickly detect and avoid collisions. Finally, the model can be easily and dynamically converted to and from the traditional parametric snakes model representation. We apply a 2D model to segment structures from medical images with complex shapes and topologies, such as arterial “trees”, that cannot easily be segmented with traditional deformable models.

Experiments in curvature-based segmentation of range data

Abstract: This paper focuses on the experimental evaluation of a range image segmentation system which partitions range data into homogeneous surface patches using estimates of the sign of the mean and Gaussian curvatures. The authors report the results of an extensive testing program aimed at investigating the behavior of important experimental parameters such as the probability of correct classification and the accuracy of curvature estimates, measured over variations of significant segmentation variables. Evaluation methods in computer vision are often unstructured and subjective: this paper contributes a useful example of extensive experimental assessment of surface-based range segmentation. >

Refining image segmentation by integration of edge and region data

Abstract: A basic requirement for understanding the dynamics of the Earth's major ecosystems is accurate quantitative information about the distribution and areal extent of the Earth's vegetation formations. Some of this required information can be obtained through the analysis of remotely sensed data. Image segmentation is often one of the first steps of this analysis. This paper focuses on two particular types of segmentation: region-based and edge-based segmentations. Each approach is affected differently by various factors, and both types of segmentations may be improved by taking advantage of their complementary nature. Included among region-based segmentation approaches are region growing methods, which produce hierarchical segmentations of images from finer to coarser resolution. In this hierarchy, an ideal segmentation (ideal for a given application) does not always correspond to one single iteration, but map correspond to several different iterations. This, among other factors, makes it somewhat difficult to choose a stopping criterion for region growing methods. To find the ideal segmentation, the authors develop a stopping criterion for their Iterative Parallel Region Growing (IPRG) algorithm using additional information from edge features, and the Hausdorff distance metric. They integrate information from regions and edges at the symbol level, taking advantage of the hierarchical structure of the region segmentation results. Also, to demonstrate the feasibility of this approach in processing the massive amount of data that will be generated by future Earth remote sensing missions, such as the Earth Observing System (EOS), all the different steps of this algorithm have been implemented on a massively parallel processor. >

Spatio-temporal segmentation based on motion and static segmentation

Abstract: The problem of segmenting an image sequence in terms of regions characterized by a coherent motion is among the most challenging in image sequence analysis. This paper proposes a new technique which sequentially refines the segmentation and the motion estimation by combining static segmentation and motion information. The motion is robustly computed by a global estimation which remove the camera motion, followed by a local estimation using a matching technique and a robust estimator. Simulation results show the efficiency of the proposed technique.

A new methodology for gray-scale character segmentation and recognition

Abstract: Generally speaking, through the binarization of gray-scale images, useful information for the segmentation of touching or overlapping characters may be lost. If we analyze gray-scale images, however, specific topographic features and the variation of intensity can be observed in the character boundaries. We believe that such kinds of clues obtained from gray-scale images should be useful for efficient character segmentation. In this paper, we propose a new methodology for character segmentation and recognition which makes the best use of the characteristics of gray-scale images. In the proposed methodology, the character segmentation regions are determined by using projection profiles and topographic features extracted form gray-scale images. Then the nonlinear character segmentation path in each character segmentation region is found by using multistage graph search algorithm. Finally, in order to confirm the character segmentation paths and recognition results, recognition based segmentation method is adopted.

Adaptive image segmentation using genetic and hybrid search methods

Abstract: This paper describes an adaptive approach for the important image processing problem of image segmentation that relies on learning from experience to adapt and improve the segmentation performance. The adaptive image segmentation system incorporates a feedback loop consisting of a machine learning subsystem, an image segmentation algorithm, and an evaluation component which determines segmentation quality. The machine learning component is based on genetic adaptation and uses (separately) a pure genetic algorithm (GA) and a hybrid of GA and hill climbing (HC). When the learning subsystem is based on pure genetics, the corresponding evaluation component is based on a vector of evaluation criteria. For the hybrid case, the system employs a scalar evaluation measure which is a weighted combination of the different criteria. Experimental results for pure genetic and hybrid search methods are presented using a representative database of outdoor TV imagery. The multiobjective optimization demonstrates the ability of the adaptive image segmentation system to provide high quality segmentation results in a minimal number of generations. >

Segmentation of numeric strings

Abstract: This paper presents a complete procedure for the segmentation of handwritten numeric strings. The procedure uses an hypothesis-then-verification strategy in which multiple segmentation algorithms based on contiguous row partition work sequentially on the binary image until an acceptable segmentation is obtained. At this purpose a new set of algorithms simulating a "drop falling" process is introduced. The experimental tests demonstrate the effectiveness of the new algorithms in obtaining high-confidence segmentation hypotheses.

Segmentation of magnetic resonance brain image: integrating region growing and edge detection

Abstract: The authors present a method that combines region growing and edge detection for magnetic resonance (MR) brain image segmentation. Starting with a simple region growing algorithm which produces an over segmented image, the authors apply a sophisticated region merging method which is capable of handling complex image structures. Edge information is then integrated to verify and, where necessary, to correct region boundaries. The results show that this method is reliable and efficient for MR brain image segmentation.

Extraction de caractéristiques, segmentation d'image et morphologie mathématique

Abstract: The purpose of this thesis is to investigate new morphological methods for extracting the characteristics of an image regions. These methods are destined to be applied to image segmentation. Two different approaches are first presented : the first one is based on granulometries (classical sieving process), the second one considers the image extrema. We then focus on the latter and on the study of dynamics. This transformation associates with each image extremum the contrast of the region it marks. We show that it can be considered as a contrast sieving process and that it is closely linked to granulometric approaches. We then propose a generalization of the dynamics concept using connected morpholog- ical operators. These operators act on grey level images by simply propagating their flat zones. By computing more and more selective connected filters, image structures progres- sively disappear. The level for which one given structure is totally eliminated characterizes the structure in terms of the filtering criterion : in terms of contrast, size, shape... This leads us to introduce a new class of morphological transformations, the extinction func- tions, which associate with each image extremum a characteristic of the region it marks. An extinction function key concept is that of merging tree of image extrema which corre- sponds to a hierarchical description of the image regions. Extinction functions can be used for selecting significant regions in an image. They are therefore of great interest in filtering and segmentation applications (for solving the well known marking step before computing watershed transform). We illustrate the lat- ter point with many segmentation examples. The results obtained by this method are compared with the results deduced from more classical approaches. The most significant contribution of extinction functions in segmentation applications is to simplify the adjust- ment of segmentation algorithms based on the watershed transform. In particular, they allow to produce efficient hierarchical interactive segmentation algorithms.

Strategies in character segmentation: a survey

Abstract: This paper provides a review of advances in character segmentation. Segmentation methods are listed under four main headings. The operation of attempting to decompose the image into classifiable units on the basis of general image features is called "dissection". The second class of methods avoids dissection, and segments the image either explicitly, by classification of specified windows, or implicitly by classification of subsets of spatial features collected from the image as a whole. The third strategy is a hybrid of the first two, employing dissection together with recombination rules to define potential segments, but using classification to select from the range of admissible segmentation possibilities offered by these subimages. Finally, holistic approaches that avoid segmentation by recognizing entire character strings as units are described.

Fuzzy connectedness and object definition

Abstract: Approaches to object information extraction from images should attempt to use the fact that images are fuzzy. In past image segmentation research, the notion of `hanging togetherness' of image elements specified by their fuzzy connectedness has been lacking. We present a theory of fuzzy objects for n-dimensional digital spaces based on a notion of fuzzy connectedness of image elements. Although our definitions lead to problems of enormous combinatorial complexity, the theoretical results allow us to reduce this dramatically. We demonstrate the utility of the theory and algorithms in image segmentation based on several practical examples.

Genetic algorithms for object recognition in a complex scene

Abstract: A realworld computer vision module must deal with a wide variety of environmental parameters. Object recognition, one of the major tasks of this vision module, typically requires a preprocessing step to locate objects in the scenes that ought to be recognized. Genetic algorithms are a search technique for dealing with a very large search space, such as the one encountered in image segmentation or object recognition. The article describes a technique for using genetic algorithms to combine the image segmentation and object recognition steps for a complex scene. The results show that this approach is a viable method for successfully combining the image segmentation and object recognition steps for a computer vision module.

Foreground/background segmentation of color images by integration of multiple cues

Abstract: This research addresses the issue of automatically segmenting color images into foreground (F) and back-ground (B) regions with the assumption that background regions are relatively smooth but may have gradually varying colors or be slightly textured. A multi-level segmentation scheme is used that involves color clustering, unsupervised segmentation using MDL (minimum description length) principle, edge-based F/B separation, and integrated F/B segmentation. The approach has been tested on more than 100 images. Some of the experimental results are presented.

Spatial and Feature Space Clustering: Applications in Image Analysis

Abstract: We propose a novel approach to image segmentation, called feature and spatial domain clustering. The method is devised to group pixel data by taking into account simultaneously both their feature space similarity and spatial coherence. The FSD algorithm is practically application independent. It has been successfully tested on a wide range of image segmentation problems, including grey and colour image segmentation, edge and line detection, range data and motion segmentation. In comparison with existing segmentation approaches, the method can resolve image features even if their distributions significantly overlap in the feature space. It can distinguish between noisy regions and genuine fine texture. Moreover, if required, FSD clustering can produce partial segmentation by identifying salient regions only.

A Supervised Approach to the Evaluation of Image Segmentation Methods

Abstract: Evaluation is an important step in developing a segmentation algorithm for an image analysis system. We first give a review of segmentation evaluation methods, and then demonstrate how a supervised evaluation method based on shape features is used in the development of a segmentation algorithm for fluorescence images of white blood cells.

Texture segmentation using hierarchical wavelet decomposition

Abstract: Texture segmentation deals with identification of regions where distinct textures exist. In this paper, a new scheme for texture segmentation using hierarchical wavelet decomposition is proposed. In the first step, using Daubechies' 4-tap filter, an original image is decomposed into three detailed images and one approximate image. The decomposition can be recursively applied to the approximate image to generate a lower resolution of the pyramid. The segmentation starts at the lowest resolution using the K-means clustering scheme and the result is propagated through the pyramid to a higher one with continuously improving segmentation.

A new approach for Latin/Arabic character segmentation

Abstract: In this paper, we propose two methods of character segmentation for Arabic handwritten characters and cursive Latin characters. Classical horizontal and vertical projections detect the lowercase writing area in lines. The problem of overlapping lower or upper strokes is resolved with a contour-following algorithm which starts in the lowercase writing area and labels the detected contours. In the first method, the junction segments connecting the characters to each other are detected by taking into account the writing line thickness. The second method detects the upper contour of each word. The strokes are detected in order to find primary segmentation points (PSP). These points are analysed with an automaton that considers the shape of the word for the determination of definitive segmentation points (DSP). The two methods are compared and the results are discussed.

A New Plant Cell Image Segmentation Algorithm

Abstract: A new plant cell image segmentation algorithm is presented in this paper. It is based on the morphological analysis of the cell shapes and on gradient information. The difficulty of the segmentation of plant cells lies in the complex shapes of the cells and their overlapping, often present due to recent cellular division. The algorithm presented tries to imitate the human procedure for segmenting overlapping and touching particles. It analyzes concavities in the shape of a group of cells as well as the existence of a coherent surface for the segmentation. The algorithm can be divided in two main parts, firstly a simple method for finding dominant concave points in shapes is introduced and secondly several parameters between concave points are calculated as a criterion for segmentation. It has been also shown that the algorithm produces good results when the output is applied as a marker for the morphological watershed algorithm. Results will be presented in real images of a cell suspension culture to show the validity of the chosen approach.

Image segmentation towards new image representation methods

Abstract: Very low bit-rate video coding has recently become one of the most important areas of image communication and a large variety of applications have already been identified. Since conventional approaches are reaching a saturation point, in terms of coding efficiency, a new generation of video coding techniques, aiming at a deeper “understanding” of the image, is being studied. In this context, image analysis, particularly the identification of objects or regions in images (segmentation), is a very important step. This paper describes a segmentation algorithm based on split and merge. Images are first simplified using mathematical morphology operators, which eliminate perceptually less relevant details. The simplified image is then split according to a quad tree structure and the resulting regions are finally merged in three steps: merge, elimination of small regions and control of the number of regions.

Knowledge-Based Segmentation for Automatic Map Interpretation

Abstract: In this paper, a knowledge-based framework for the top-down interpretation and segmentation of maps is presented. The interpretation is based on a priori knowledge about map objects, their mutual spatial relationships and potential segmentation problems. To reduce computational costs, a global segmentation is used when possible, but an applicable top-down segmentation strategy is chosen when errors in the global segmentation are detected. The interpretation system has been tested on utility maps and the experiments show that when a top-down resegmentation strategy is used to correct errors in the global segmentation, the recognition performance is improved significantly. © Springer-Verlag Berlin Heidelberg 1996.

Using explicit segmentation to improve HMM phone recognition

Abstract: We show that many of the errors in a context-dependent phone recognition system are due to poor segmentation. We then suggest a method to incorporate explicit segmentation information directly into the HMM paradigm. The utility of explicit segmentation information is illustrated with experiments involving five types of segmentation information and three methods of smoothing.

Ground-truthing and benchmarking document page segmentation

Abstract: We describe a new approach for evaluating page segmentation algorithms. Unlike techniques that rely on OCR output, our method is region-based: the segmentation output, described as a set of regions together with their types, output order etc., is matched against the pre-stored set of ground-truth regions. Misclassifications, splitting, and merging of regions are among the errors that are detected by the system. Each error is weighted individually for a particular application and a global estimate of segmentation quality is derived. The system can be customized to benchmark specific aspects of segmentation (e.g., headline detection) and according to the type of error correction that might follow (e.g., re-typing). Segmentation ground-truth files are quickly and easily generated and edited using GroundsKeeper, an X-Window based tool that allows one to view a document, manually draw regions (arbitrary polygons) on it, and specify information about each region (e.g., type, parent).

Probabilistic multiscale image segmentation by the hyperstack

Abstract: In the medical community, images from different modalities have found their way to a variety of medical disciplines. Multidimensional images have become indispensable in clinical diagnosis, therapy planning and evaluation. Image segmentation -- dividing an image into meaningful objects -- is a subfield of image processing that is of crucial importance for quantitative analysis and, in the case of 3D images, volume visualization. We present an approach to segmentation of multidimensional images called the hyperstack. This method is based on the scale space theory, where an image is considered at multiple levels of scale (resolution) simultaneously. The original image represents the highest scale (detail information), while succesively low-pass filtering of the image produces the larger scales (global information). By establishing linkages between these levels of scale, the global information can efectively be used to guide the segmentation of the pixels in the original image. In particular, the research has focused on the segmentation of partial volume voxels. It is shown that this leads to an improvement of both quantitative analysis and 3D volume renderings. Furthermore, it has been investigated if the segmentations could be further improved by using nonlinear filtering techniques to generate the scale space. The hyperstack has basically been applied to neurological MR T1 images, but is not restricted to a particular type of image or modality.