Image segmentation for coding

Chowdhury, Md. Mahbubul Islam (2000) Image segmentation for coding. Masters thesis, Memorial University of Newfoundland.

[English] PDF (Migrated (PDF/A Conversion) from original format: (application/pdf)) - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (12MB) [English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. (Original Version)

Abstract

The segmentation of regions is an important first step for a variety of image analysis and visualization tasks. There is a wide range of image segmentation techniques in the literature. Conventional segmentation techniques for monochromatic images can be categorized into two distinct approaches. One is region based, which relies on the homogeneity of spatially localized features, whereas the other is based on boundary finding, using discontinuity measures. Based on one or both of these properties, diverse approaches to image segmentation exhibiting different characteristics have been suggested in the literature. -- The research of this thesis was aimed at combining region growing and edge detection methods to provide better segmentation results. Existing schemes that use region-based processing provide unambiguous segmentation, but they often divide regions that are not clearly separated, while merging regions across a break in an otherwise strong edge. Edge-based schemes are subject to noise and global variation in the picture (e.g. illumination), but do reliably identify strong boundaries. The proposed combined algorithm begins by using region growing to produce an over-segmented image. This phase is fast (order N, where N is the number of pels in the image). The over-segmented output of the region growing is then modified using edge criteria such as edge strength, edge straightness, edge smoothness and edge continuity. Two techniques - line-segment subtraction and line-segment addition - have been investigated. In the subtraction technique, the weakest edge (based on a weighted combination of the criteria) is removed at each step. Every time that a weakest edge is removed, the combined edge strengths of the remaining edges are recalculated. In the addition technique, the strongest edge (based on the weighted combination of all criteria) of all the edges is calculated first. It is used to seed a multi-segment line that grows out from it at both ends. At each end of the strongest edge, a binary tree containing four branches is investigated. The adjoining edge that has the highest edge strength is appended to the seed. This process of appending continues until a closed loop or a boundary is reached. The overall procedure for both techniques for segmentation has been developed. -- In order to investigate the performance of the proposed segmentation techniques, a segmentation evaluation method is demonstrated. Since a human is the ultimate judge, a subjective evaluation method is developed. Segmentation produced by a human is compared to segmentation produced by the algorithm and correlation is calculated between the human method and the algorithms. Subjective tests performed on the algorithms and the results confirm that the proposed algorithms can be used to produce better image segmentation than the segmentation produced by existing region-based techniques.

Actions (login required)

View Item