Quadtree-Based Genetic Algorithm And Its Applications To Computer Vision.pdf

Pattern Recognition 37 (2004) 1723–1733
ate/patcog
Quadtree-based ic algorithm and its applications to
computer vision
Minglun Gong, Yee-Hong Yang∗
Department puting Science, Computer Graphics Research Group, University of Alberta, Edmonton, AB, Canada, T6G 2E8
Received 9 September 2003; accepted 12 February 2004
Abstract
puter vision problems can be formulated as optimization problems. Presented in this paper is a new framework
based on the quadtree-based ic algorithm that can be applied to solve many of these problems. The proposed algorithm
incorporates the quadtree structure into the conventional ic algorithm. The solutions of image-related problems are
encoded through encoding the corresponding quadtrees, and therefore, the 2D locality within a solution can be preserved.
Examples addressed using the proposed framework include image segmentation, stereo vision, and motion estimation. In all
cases, encouraging results are obtained.
? 2004 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.
Keywords: ic algorithm; Ill-posed problems; Image segmentation; Quad-tree; Optimization; Stereo vision; Motion estimation
1. Introduction to solve the above-mentioned optimization problem. How-
ever, being a essful global optimization technique, the
Many problems in puter vision area are ill-posed ic algorithm [5] is rarely used. This is mainly because
problems, . there is insu9cient information available to ic algorithms encode the solutions using 1D strings,
solve the problem. Regularization is a method that employs while solutions of these image-related problems have inher-
additional constraints so that a solution can be determined. ent 2D structures. Therefore, if we simply use pixels as genes
Among the many proposed constraints, the smoothness con- directly, it is unlikely to get good results since the straight-
straint [1] is widely used. Using this constraint, the solving forward encoding method cannot preserve