Study On Image Boundary Detection Based On The Genetic Algorithms

Boundary is one of the most basic image features and often carries a lot of image information that is of great importance in analyzing, describing and, understanding an image in a high level.Boundary detection plays an important role in image processing and analyzing. Research on boundary detection has last for the past decades. During this period, a lot of excellent papers have been published and the boundary detection algorithms have progressively developed. However, boundary detection algorithm still ranks as a difficult problem in the image processing. The traditional boundary detection algorithms are on the basis of gradient and have more or less limitations. On the other hand, boundary has different definitions according to different applications. So it is not easy to find a universal boundary detection method. With the development of science and technology, many new boundary detection algorithms based on neural network, genetic algorithm, wavelet, and morphology are presented.In this thesis, research work focuses on how to use genetic algorithms to solve boundary detection problems. The main researches are as follows:1. The traditional boundary detection algorithms based on masks are summarized and its limitations are also analyzed. Furthermore, the issue of how to design the best edge detection mask is discussed and a novel mask design method based on genetic algorithms is researched. The simulation and experimental results show that the given method can effectively extract the boundaries from the homologous images.2. The basic strategy and convergence of genetic algorithms are analyzed and its evaluation functions are researched. The chaos genetic algorithm (CGA) which can effectively overcome the premature convergence is also discussed.3. Geometric primitive extraction is a classical problem in the image processing arena. Hough transform, which is often used in boundary detection, has a serious limitation for its consumption of time and space. What is more, Hough transform can only apply on binary images. To solve this problem, one method based on genetic algorithms is presented to extract geometric primitives directly from the gray images and also tested on the iris localization. Simulation results show that this method is encouraging and...