Aspects of computational mode and data distribution for parallel range image segmentation (original) (raw)

Parallelization of 3-D Range Image Segmentation on a SIMD Multiprocessor

The paper presents a parallelized algorithm for segmentation of 3-D range images. The segmented image is useful as an input to higher level image processing tasks. The algorithm is implemented on a mesh connected multiprocessor machine, the MasPar series of massively parallel computers. A 2-dimensional hierarchical data distribution scheme is used to allocate the image pixels on the machine. The original sequential segmentation procedure is modi ed to take into account this data distribution. The performance of the new programs is compared to that of the original code, executed on an IBM RS6000 520H workstation. The speed-up is found to be about three for a Mas-Par consisting of an 128 x 128 processor array. This is due to the architectural constraints of the MasPar, in particular the implementation of its interprocessor communications. The utilization of processors on the MasPar is found to be as low as 56 % for this algorithm. Results of executions on machines of various sizes are also shown. Suggestions for better performance are discussed.

Parallelization of 3-D Range Image Segmentation on a SIMD

Segmentation of Range Images

The analyses of three-dimensional (3D) scenes from range data need the segmentation of 3D surfaces into planar patches and quadratic surface regions. In this paper the concept of the Digital Neighbourhood Planes introduced earlier (Pattern Recognition Lett. 11(3), 215-223 (1990)) for 3D binary data is suitably extended for the segmentation of range images. For range images the local neighbourhood of every point is virtually exploded (to 3 × 3 × 5, 3 × 3 x 7, etc.) and Neighbourhood Plane Set (NPS) values, indicative of the orientation of the surface normal at every range pixel, are computed. Subsequently a region growing technique is adopted to cluster points based on the N PS values which, with suitable post-processing, result in the final segments. The algorithm is simple and computationally efficient as it uses set-theoretic operations only. The algorithm is illustrated with the help of several examples. For most of them it produces good segmentation results. Finally, the algorithm has enough potential for parallelization which can be explored in the future.

Complexity analysis of range image segmentation on MasPar MP-1

Proceedings of 36th Midwest Symposium on Circuits and Systems, 1993

Many low level vision tasks that are computationally intensive are easily parallelizable. The lack of parallel processing systems, or their prohibitive costs, have prevented the move of vision processing algorithms from single processor systems to multiprocessor systems. With the recent spurt of parallel processing hardware, there is a need to investigate the feasibility of using such machines for some vision algorithms. Speedup is an important factor in determining the feasibility of migration from single processor systems to parallel processors. In this work, we investigate a particular segmentation algorithm and present theoretical speedup results. Our formula can work out numerical speedups by simply plugging in the parameter values.

An Experimental Comparison of Range Image Segmentation Algorithms

IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996

A methodology for evaluating range image segmentation algorithms is proposed. This methodology involves (1) a common set of 40 laser range finder images and 40 structured light scanner images that have manually specified ground truth and (2) a set of defined performance metrics for instances of correctly segmented, missed, and noise regions, over- and under-segmentation, and accuracy of the recovered geometry. A tool is used to objectively compare a machine generated segmentation against the specified ground truth. Four research groups have contributed to evaluate their own algorithm for segmenting a range image into planar patches.

A new technique for segmentation of range images

[1992] Proceedings of the IEEE International Symposium on Industrial Electronics, 2000

Abet ract This paper presents a new technique for the segmentation of range images. It tends to be a hybrid approach that tends to combine edge detection and region growing techniques toward accomplishing the task of segmentation. Multiple attribute maps were utilised to provide surface characteristics and smoothness information. The emsbreeding approach produces a segmentation map and an edge map. Wes detected by such method possead good localisation property. The incorporation of region growing process eliminates internal micro edges and produces a segmentation map. The edge and segmentation maps when conveyed to the higher level recognition process, may prove valuable for three-dimensional object identification purposes.

An experimental comparison of a hierarchical range image segmentation algorithm

Computer and Robot Vision, …, 2005

This paper describe a new algorithm to segment range images into continuous regions represented by Bézier polynomials. The main problem in many segmentation algorithms is that it is hard to accurately detect at the same time large continuous regions and their boundary location. In this paper, a Bayesian framework is used to determine through a region growing process large continuous regions. Following this process, an exact description of the boundary of each region is computed from the mutual intersection of the extracted parametric polynomials followed by a closure and approximation of this new boundary using a gradient vector flow algorithm. This algorithm is capable of segmenting not only polyhedral objects but also sculptured surfaces by creating a network of closed trimmed Bézier surfaces that are compatible with most CAD systems. Experimental results show that significant improvement of region boundary localization and closure can be achieved. In this paper, a systematic comparison of our algorithm to the most well known algorithms in the literature is presented to highlight its performance.

Multimodal Range Image Segmentation

Vision Systems: Segmentation and Pattern Recognition, 2007

The segmentation process is perhaps the most important step in image analysis since its performance directly affects the performance of the subsequent processing steps in image analysis and it significantly determines the resulting image interpretation. Despite its utmost importance, segmentation still remains as an unsolved problem in the general sense as it lacks a general mathematical theory. The two main difficulties of the segmentation problem are its underconstrained nature and the lack of definition of the "correct" segmentation. Perhaps as a consequence of these shortcomings, a plethora of segmentation algorithms has been proposed in the literature. These algorithms range from simple ad hoc schemes to more sophisticated ones using object and image models. The area of segmentation algorithms typically suffers with the lack of benchmarking results and methodologies. With few rare exceptions in specific narrow applications single segmentation algorithm cannot be ranked and potential user has to experimentally validate several segmentation algorithms for his particular application.

Fast segmentation of range images

Lecture Notes in Computer Science, 1997

A new type of range image segmentation method is introduced. The image segmentation is based on a recursive adaptive regression model prediction for detecting range image step discontinuities which are present at object face borders. Border pixels are detected in two perpendicular directions and detection results are combined together. Two predictors in each direction use identical contextual information from the pixel's neighbourhood and they mutually compete for the most optimal discontinuity detection. The method suggested can be successfully applied also to other image segmentation applications, e.g. panchromatic or multispectral image data, etc.

Range image segmentation by controlled-continuity spline approximation for parallel computation

1992

A classical approach formulates surface reconstruction in term of a variational problem by using two-dimensional surfaces defined by generalized spline functions. We present such an approach in the case of range image segmentation. The particularity of our approach lies in the way the discontinuities are detected. The spline is constrained to stay within a certain maximal distance to the discrete measured data, but is free as far as the maximum distance is not reached. Discontinuity emerges on points where the maximum distance constrains the spline. This method leads to a relaxation algorithm which solves the segmentation iteratively, by locally applying a relation which is close to the diffusion equation in the case of the membrane spline. Being iterative and local, the algorithm is suited for parallelism. We applied the method to range data from laser scanners using two different surface models: the membrane spline, more adequate for polyhedric objects, and the thin plate spline, more adequate for curved objects. The results illustrate the practical performance of this method which is simple, parallel, and controlled by few parameters.

Aspects of computational mode and data distribution for parallel range image segmentation (original) (raw)

Related papers