![]() | Only 14 pages are availabe for public view |
Abstract This thesis introduces new algorithms for edge detection since edge detection is considered as one of the most fundamental steps in computer vision. The new algorithmsbelong to ”Local Structure Information” class. This class is based oil statistical analysis of the local structure properties of the neighborhood of pixels to identify significant edges. First, a full survey of computer vision process is introduced with an emphasis on the related topics. Then, a proposed computer vision system is described based upon scale-space concept. Three classes of edge detection techniques have been introduced. Each class is represented by various techniques to clarify the major properties of the class. The main issues that are related to the research points are presented in details. Tile new algorithms are based oil ”Hierarchical Structures”, ”Scale Space Concept”, and ”Local Structure information”. Hierarchical structure concept introduces the application of resolution pyramids, while scale-space concept is implemented through Gaussian filter. The first two proposed algorithms are based upon the application of hierarchical pyramids on local structure information techniques, while the other four algorithms introduce the concept of scale-space concept to give the new Gaussian pyramid. III the simulation stage, algorithms that represent the recently published techniques have been implemented and results are summarized. The ideas concluded from the traditional classes are presented. The proposed algorithms have been introduced and implemented. Experiments have been established with different conditions such as various noise levels and wide ranges for the parameters of the algorithms. The thesis is concluded by the simulation results with practical recommendations about the use of the newly proposed algorithms. |