Contour matching using ant colony optimization and curve evolution

• Main Author: Saadi, Younes
• Format: Thesis
• Language: English; English; English
• Published: 2013
• Subjects: TA1501-1820 Applied optics. Photonics
• Online Access: http://eprints.uthm.edu.my/2166/1/24p%20YOUNES%20SAADI.pdf; http://eprints.uthm.edu.my/2166/2/YOUNES%20SAADI%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/2166/3/YOUNES%20SAADI%20WATERMARK.pdf; http://eprints.uthm.edu.my/2166/
• Institution: Universiti Tun Hussein Onn Malaysia

Shape retrieval is a very important topic in computer vision. Image retrieval consists of selecting images that fulfil specific criteria from a collection of images. This thesis concentrates on contour-based image retrieval, in which we only explore the information located on the shape contour. There are many different kinds of shape retrieval methods. Most of the research in this field has till now concentrated on matching methods and how to achieve a meaningful correspondence. The matching process consist of finding correspondence between the points located on the designed contours. However, the huge number of incorporated points in the correspondence makes the matching process more complex. Furthermore, this scheme does not support computation of the correspondence intuitively without considering noise effect and distortions. Hence, heuristics methods are convoked to find acceptable solution. Moreover, some researches focus on improving polygonal modelling methods of a contour in such a way that the resulted contour is a good approximation of the original contour, which can be used to reduce the number of incorporated points in the matching. In this thesis, a novel approach for Ant Colony Optimization (ACO) contour matching that can be used to find an acceptable matching between contour shapes is developed. A polygonal evolution method proposed previously is selected to simplify the extracted contour. The main reason behind selecting this method is due to the use of a stopping criterion which must be predetermined. The match process is formulated as a Quadratic Assignment Problem (QAP) and resolved by using ACO. An approximated similarity is computed using original shape context descriptor and the Euclidean metric. The experimental results justify that the proposed approach is invariant to noise and distortions, and it is more robust to noise and distortion compared to the previously introduced Dominant Point (DP) Approach.