Showing papers on "Range segmentation published in 1999"

Mean shift analysis and applications

Abstract: A nonparametric estimator of density gradient, the mean shift, is employed in the joint, spatial-range (value) domain of gray level and color images for discontinuity preserving filtering and image segmentation. Properties of the mean shift are reviewed and its convergence on lattices is proven. The proposed filtering method associates with each pixel in the image the closest local mode in the density distribution of the joint domain. Segmentation into a piecewise constant structure requires only one more step, fusion of the regions associated with nearby modes. The proposed technique has two parameters controlling the resolution in the spatial and range domains. Since convergence is guaranteed, the technique does not require the intervention of the user to stop the filtering at the desired image quality. Several examples, for gray and color images, show the versatility of the method and compare favorably with results described in the literature for the same images.

Background estimation and segmentation based on range and color

Abstract: Segmentation of background and foreground objects in an image is based upon the joint use of both range and color data. Range-based data is largely independent of color image data, and hence not adversely affected by the limitations associated with color-based segmentation, such as shadows and similarly colored objects. Furthermore, color segmentation is complementary to range measurement in those cases where reliable range data cannot be obtained. These complementary sets of data are used to provide a multidimensional background estimation. The segmentation of a foreground object in a given frame of an image sequence is carried out by comparing the image frames with background statistics relating to range and normalized color, using the sets of statistics in a complementary manner.

Image thresholding by indicator kriging

Abstract: We consider the problem of segmenting a digitized image consisting of two univariate populations. Assume a priori knowledge allows incomplete assignment of voxels in the image, in the sense that a fraction of the voxels can be identified as belonging to population II/sub 0/, a second fraction to II/sub 1/, and the remaining fraction have no a priori identification. Based upon estimates of the short length scale spatial covariance of the image, we develop a method utilizing indicator kriging to complete the image segmentation.

Background estimation and removal based on range and color

Abstract: Background estimation and removal based on the joint use of range and color data produces superior results than can be achieved with either data source alone. This is increasingly relevant as inexpensive, real-time, passive range systems become more accessible through novel hardware and increased CPU processing speeds. Range is a powerful signal for segmentation which is largely independent of color and hence not effected by the classic color segmentation problems of shadows and objects with color similar to the background. However range alone is also not sufficient for the good segmentation: depth measurements are rarely available at all pixels in the scene, and foreground objects may be indistinguishable in depth when they are close to the background. Color segmentation is complementary in these cases. Surprisingly, little work has been done to date on joint range and color segmentation. We describe and demonstrate a background estimation method based on a multidimensional (range and color) clustering at each image pixel. Segmentation of the foreground in a given frame is performed via comparison with background statistics in range and normalized color. Important implementation issues such as treatment of shadows and low confidence measurements are discussed in detail.

Registering two overlapping range images

Abstract: The paper describes a method of automatically performing the registration of two range images that have significant overlap. We first find points of interest in the intensity data that comes with each range image. Then we perform a triangulation of the 3D range points associated with these 2D interest points. All possible pairs of triangles between the two 3D triangulations are then matched. The fact that we have 3D data available makes it possible to efficiently prune matches. We do this pruning by using a simple and effective set of compatibility tests between potentially matching triangles and vertices. The best match is the one that aligns the largest number of interest points between the two range images. The algorithms are demonstrated experimentally on a number of different range image pairs.

Edge detection to guide range image segmentation by clustering techniques

Abstract: Edge detection is an unsolved problem in that, so far, there is no general optimal solution. However, edge detection provides rich information about the scene being observed. This is particularly true in range images, where 3D information is explicit. Many researchers have been taking advantage of edge detection information to improve the segmentation of range images by integrating edge detection with other different segmentation techniques. This paper presents a methodology to perform edge detection in range images in order to provide a reliable and meaningful edge map, which helps to guide and improve range image segmentation by clustering techniques. The obtained edge map leads to three important improvements: (1) the definition of the ideal number of regions to initialize the clustering algorithm; (2) the selection of suitable initial cluster centers; and (3) the successful identification of distinct regions with similar features. Experimental results that substantiate the effectiveness of this work are presented.

Watershed-based hierarchical SAR image segmentation

Abstract: Watershed transformation is a powerful image segmentation tool recently developed in mathematical morphology. In order to segment images initially oversegmented by watershed transformation, two approaches are considered: one is the thresholding of the gradient image proposed by us which is capable of keeping more salient image contours; the other is the well known centroid linkage region growing algorithm which merges regions with certain statistical similarities. By choosing suitable thresholds in the two approaches, hierarchical image segmentation algorithms can be constructed. A Ratio of Averages (ROA) edge detector is proposed to replace the morphological edge detectors prior to watershed transformation when applied to Synthetic Aperture Radar (SAR) images. Applications to SAR agricultural image segmentation with these hierarchical segmentation algorithms are presented. It is demonstrated that the algorithms are efficient in the segmentation of the SARimages and appropriate for land use applications w...

Image segmentation based on region growing and edge detection

Abstract: We present a method that incorporates region-growing and edge-detection techniques for performing image segmentation tasks. We first apply edge detection techniques to obtain a difference in strength map. We then employ the region growing technique to work on the map. The idea is this: when a region grows to such extent that it touches an edge pixel, a criterion proposed in this paper is used to determine whether the growing process should be terminated in this direction. In this manner, one can obtain more accurate boundary of region by clearly separating regions from edges. After all pixels of the map have been processed, a map of primitive region is generated. Finally, a merge process based on a similarly measurement (i.e., connectivity degree and averaged gray level between regions) is employed to aggregate regions to obtain the final segmentation result.

Adaptive pixel management using object type identification

Abstract: The present invention is directed to a method for differentially processing image types within a document image to enhance the quality of on an image on a receiving medium. In one aspect of the invention, the method begins by receiving document image data including one or more image regions. The method determines for each image region whether the region includes any image data corresponding to a first image type. The method then prints regions including image data corresponding to the first image type according to a first pixel management process and prints regions not including image data corresponding to the first image type according to a second pixel management process.

Method for segmenting a digital image

Abstract: Briefly, in accordance with one embodiment of the invention, a method of segmenting an initial digital image includes the following. The initial digital image is processed to produce a first digital image with defined edges corresponding to the initial digital image and to produce a second digital image with at least two dominant contiguous regions corresponding to the initial digital image. Distinct non-overlapping regions of the first digital image formed by the defined edges are identified. The distinct non-overlapping regions of the first digital are combined based, at least in part, on a correspondence with the at least two dominant contiguous regions in the second digital image. Based, at least in part, on the remaining regions after combining the distinct non-overlapping regions of the first digital image, the initial digital image is segmented.

Coupled Geodesic Active Regions for Image Segmentation

Abstract: This paper presents a novel variational method for image segmentation which is obtained by unifying boundary and region-based information sources under the Geodesic Active Region framework. A statistical analysis over the observed density function (image histogram) using a mixture of Gaussian elements, indicates the number of the different regions and their intensity properties. Then, the boundary information is determined using a probabilistic edge detector, while the region information is given directly from the observed image using the conditional probability density functions of the mixture model. The defined objective function is minimized using a gradient-descent method where a level set approach is used to implement the resulting PDE system. According to the motion equations [PDE], the set of initial curves is propagated towards the segmentation result under the influence of boundary and region-based segmentation forces, and being constrained by a regularity force. The changes of topology are naturally handled thanks to the level set implementation, while a coupled multi-phase propagation is adopted that increases the robustness and the convergence rate by introducing a coupled system of equations for the different level set functions. Besides, to reduce the required computational cost and to decrease the risk of convergence to a local minimum, a multi-scale approach is also considered. The performance of our method is demonstrated on a variety of synthetic and real images.

Color image segmentation method using watershed algorithm and contour information

Abstract: In this paper, we propose a segmentation method of color images with being rid of oversegmentation. Many conventional segmentation methods assign one region to each gathering of pixels which have similar property. But, these methods tend to oversegment by the optical influences. Besides, they do not use the edge information, therefore the edge is not always selected to the boundary. In this study, we propose a segmentation method of color images with being rid of oversegmentation. Our method uses not only the pixel's property but also the information of regions' contour. It consists of the first segmentation process and the region combining process. First, the given image is segmented in consideration of the edge information. Next, the obtained regions are combined according to pixels' property and regions' contour.

Method of and apparatus for optical imaging inspection of multi-material objects and the like

Abstract: A novel method all of and apparatus for categorizing different material regions of an object-to-be inspected, by optical scanning of the object to produce a pixel image thereof, discriminating different pixel regions of the image corresponding to possibly different materials on the basis of color, and texture brightness measurements of the pixel regions and assigning preliminary likelihoods to such discrimination through with ambiguities; and comparing the measurements of the pixel regions with their local neighborhood surroundings in the image to assist in resolving said ambiguities and determining the material categorizations of the pixel regions with a high likelihood.

Binary and ternary flows for image segmentation

Abstract: A novel region-based approach to snakes is introduced in this paper for the segmentation of images composed of two or three types of regions where each region may be distinguished by a given statistic. The basic idea behind this technique is to formulate curve evolutions which separate two or more values of a predetermined set of statistics computed over geometrically determined subsets of the image data. Our methodology provides a natural framework for incorporating both global and local image information in the active contour motion while avoiding the use of image derivatives. As such, this technique possesses a robustness to noise which is noncharacteristic of most edge-based snake algorithms.

Multiscale sigma filter and active contour for image segmentation

Abstract: Image segmentation is essential for image recognition. To achieve image segmentation, it is often desirable to process the image into piecewise smooth regions while preserving or even enhancing important edges. In this paper, we first design a multiscale sigma filter to achieve the above objective. To locate the object more accurately, we then follow the sigma filter processing step and a multiscale active contour method with much simpler implementation than the originally proposed active contour algorithm.

A robust bubble delineation algorithm for froth images

Abstract: Describes a robust segmentation algorithm for froth images from flotation cells in mineral processing. The size, shape, texture and color of bubbles in a froth image is very important information for optimizing flotation. To determine these parameters, the bubbles in a froth image have to be delineated first. Due to the special characteristics of froth images and a large variation of froth image patterns and quality, it is difficult to use classical segmentation algorithms. Therefore, a new segmentation algorithm was developed to delineate every individual bubble in a froth image. A new segmentation algorithm based on valley-edge detection and edge tracing has been developed. In order to detect bubble edges clearly and disregard the edges of the white spots, the algorithm just detects valley-edges between bubbles in the first step. It detects each image pixel to find if it is the lowest valley point in a certain direction. If it is, the pixel is marked as an edge candidate. Before this procedure, to alleviate noise edges, an image enhancement procedure was added to filter out the noise pixels. After valley-edge detection, the majority of edges are marked at one time, but some small gaps between edges, and noise still exist in the image. To reduce the noise, a clean up procedure was developed. To fill the gaps, an edge tracing algorithm was applied, in which, edges are smoothed into one pixel width. Endpoints and their directions are detected, and edge tracing starts from the detected endpoints. When a new valley-edge pixel is found, the algorithm uses it as a new endpoint, and the valley-edge tracing procedure continues until a contour of a bubble is closed. The segmentation algorithm has been tested on images from Pyhasalmi mine in Finland and Garpenberg mine in Sweden. The processing speed of the algorithm is much faster than for normal morphological segmentation algorithms. The processing accuracy is better than that of manual segmentation result.

Segmentation of human faces in color images using connected operators

Abstract: In this paper we present a face segmentation algorithm for color images based on connected operators. Using a skin color model, we construct a skin probability image that indicates the probability of each pixel representing skin. Morphological filters are applied to this probability images instead of applying them to the original image. A hierarchy of operators with geometrical criteria (size, compactness, orientation) is employed to simplify the skin probability image and a gray level criterion based on principal components analysis is used for final classification. Using connected operators, regions with different probabilities of being skin are analyzed which leads to a more robust segmentation compared to a single threshold classification.

Geometric-Variational Approach for Color Image Enhancement and Segmentation

Abstract: We merge techniques developed in the Beltrami framework to deal with multi-channel, i.e. color images, and the Mumford-Shah functional for segmentation. The result is a color image enhancement and segmentation algorithm. The generalization of the Mumford-Shah idea includes a higher dimension and codimension and a novel smoothing measure for the color components and for the segmenting function which is introduced via the Γ-convergence approach. We use the Γ-convergence technique to derive, through the gradient descent method, a system of coupled PDEs for the color coordinates and for the segmenting function.

Differential feature distribution maps for image segmentation and region queries in image databases

Abstract: We present a novel region segmentation framework, dedicated to region queries in content based image retrieval. Some of the features that are considered for image indexing are used for segmenting the image in a few regions of interest. The novelty of our technique comes from the unification of the feature space and the image space segmentation in a common framework. The method uses no prior modeling of the image, focusing on local feature distributions and their spatial stability in a multi feature, multi resolution approach. Several experiments are presented on real world imagery, demonstrating the power of the method for segmentation and region queries in image databases.

High precision image alignment detection

Abstract: High precision image alignment detection uses an iterative part pixel shift detection algorithm to accurately determine the displacement of a received image with respect to a reference image. An alignment pattern is inserted into the original image and a portion of the alignment pattern, such as a single line, also is stored as a reference image. Using cross-correlation the received image is compared with the reference image to locate the alignment pattern, and selected portions of the received image alignment pattern are then used in conjunction with the reference image to determine the total pixel shift of the received image. An integer pixel shift is determined by cross-correlation of the received alignment pattern with the reference image. Using the integer pixel shift to identify a starting point, data is extracted from the received alignment pattern about a specific feature and a part pixel shift is measured. The received alignment pattern is then shifted by the part pixel shift, the data is again extracted and an additional part pixel shift is measured. These steps are iterated, using the sum of all prior part pixel shifts for each subsequent shift. At completion the total of the integer pixel shift value and all the part pixel shift values determines the pixel shift required for registration of the received image vis a vis the original image.

Block selection-based image processing

Abstract: An image processing system includes input of image data, performance of block selection processing on the input image data to determine types of pixel data within the image data, a first determining step of determining, based on the block selection processing, if subject pixel data represents a text pixel, a second determining step of determining if the subject pixel data represents an edge pixel, performance of a first processing on the subject pixel data in a case that the subject pixel data is determined to represent a text pixel and an edge pixel, and performance of a second processing on the subject pixel data in a case that the subject pixel data is not determined to represent a text pixel and is not determined to represent an edge pixel

An approach to image segmentation using multiresolution analysis of wavelets

Abstract: Local changes or variations of the intensity of an image (such as edges and corners), are important information for image processing and pattern recognition. Wavelet analysis is one of the most popular techniques that can be used to detect local intensity variation. In this paper, multiresolution analysis of wavelets is used to decompose images into pyramid images. Edges and peaks are extracted from pyramid images. A coarse-to-fine image segmentation method is proposed in this paper.

Segmentation of range and intensity image sequences by clustering

Abstract: Presents a method for segmenting temporal sequences of range and intensity images. The paper addresses two problems: fusion of intensity and range data for image segmentation, and visual tracking of segments over time. Our method is based on clustering in a 4D feature space which contains intensity and geometric features. The problem of tracking segments over time is solved by adaptive image sequence clustering. The main idea is to use the cluster centers of the previous image to initialize clustering for the current image. This link between consecutive clustering steps allows one to track clusters over time without explicit correspondence analysis. First experiments show that our method can successfully segment and track objects independent of their shapes and motions.

Texture analysis and segmentation of images using fractals

Abstract: Many objects in images of natural scenes are so complex that describing them by traditional techniques is inadequate. This paper presents a family of techniques suitable for texture analysis and segmentation of objects in aerial images. Texture has been one of the most important but difficult properties for image coding and compression. It is important because it describes the entire area of a region and provides the essential structure information in regions of an image. Our goal here is to decompose an image to texturally homogenous regions. An efficient technique for computing the fractal dimension of images is used. Three different techniques; the Hurst transform, the Sobel operator and the variance are applied to two images and the results are compared. It is shown that variance dimension converts the original image to one whose texture information permits simple thresholding for texture analysis and segmentation.

Image processing method and apparatus and image forming method and apparatus using the same

Abstract: An image can be outputted with a high sharpness without deteriorating a resolution. In order to output an image of a high picture quality while suppressing the occurrence of a moire, an exposure amount is modulated in accordance with pixel density information of the image divided into pixels of a predetermined size by an exposure amount modulating unit in a light scanning unit, thereby expressing an image dark/light state. In this case, in a highlight density region in which a pixel density is equal to or less than ⅓ of the maximum image density, the density data of two adjoining pixels is modulated by one pixel and the other pixel is not recorded. In a density region in which the image density lies within a range from ⅓ to ½ of the maximum image density, a part of the density data of one of the two adjoining pixels is transposed to the other pixel. In a density region in which the image density is equal to or larger than ½ of the maximum density, the pixel transposition is not performed.

Image memory for extended field of view ultrasonic diagnostic imaging

Abstract: An extended field of view image memory has x-y coordinates corresponding to the sight area of an extended field of view image, and a depth corresponding to the maximum number of elemental image pixels that can be compounded to form a pixel of the extended field of view image. Each elemental image is entered into the memory in registration with a previously acquired elemental image, and pushes down any previously entered images it overlaps. The finite depth of the memory thus forms a FIFO register at each pixel location, which eliminates the oldest pixel data when a newly entered image causes the maximum pixel depth at a location to be exceeded.

A new technique for the segmentation of lumen from endoscopic images by differential region growing

Abstract: A new algorithm for the segmentation of endoscopic images is presented. The proposed technique consists of a dual-step methodology viz. segmentation of quasi Region of Interest (qROI) using a global thresholding technique and segmentation of actual lumen using differential region growing. The proposed scheme is faster than the conventional gradient based region growing technique. The accuracy and high speed response of the proposed technique is validated with several endoscopic images and the results are presented.

A genetic algorithm based image segmentation for image analysis

Abstract: A new genetic algorithm (GA) based image segmentation method is proposed for image analysis. This method using a mean square error (MSE) based criterion can segment an image into some regions, while estimating a suitable region representation. The criterion is defined as MSE caused by interpolating each region of an observed image with a parametric model. Since the criterion is expressed with not only the parameters of the model but also shape and location of the regions, the criterion can not be easily minimized by the usual optimization methods, the proposed method minimizes the criterion by a GA. The proposed method also includes a processor to eliminate fragile regions with the Markov random field (MRF) model. Though the thresholds of the existent methods negatively affect image segmentation results; since no thresholds are required in the proposed method, it segments images more accurately than the existent methods.