Font Size: a A A

Study On Adaptive Edge Detection Of Coronary Artery CT Images

Posted on:2019-03-02Degree:MasterType:Thesis
Country:ChinaCandidate:S M ZhangFull Text:PDF
GTID:2428330548458936Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Image,as the main method of human perception and understanding of the world,is closely related to our life.And extracting important and useful information to assist human studies is the value of the image,so the digital image processing technology has become a key technology in computer field.Edge detection and extraction is the base for further work such as pattern recognition and matching.At present,there are many methods of edge detection.This paper makes a further study on the edge detection methods of natural images and medical coronary CT images.First,we introduce the background and significance of edge detection and medical image research,and briefly summarize the research status of edge detection technology and CT technology at home and abroad.Then several classical boundary detection strategies are explored,and the experimental results are compared on natural images.Based on the experimental results,the advantages and disadvantages of these detection strategies are analyzed.Then introduces the mathematical model and the basic principle of ant colony algorithm for traveling salesman problem,analyzes the advantages and disadvantages of the classical ant colony algorithm,and introduces the principle of boundary detection algorithm based on ant colony algorithm.For the basic ant colony algorithm,it is easy to get the discontinuous boundary,and it is difficult to explore t the weak edge of the image.This paper proposes several measures such as using Otsu algorithm on the original image and the pheromone asynchronous updating increases to prevent premature stagnation.We also analyze and compare the results of the detection of natural images.Finally,because medical images often contain noise,image preprocessing for the coronary artery CT images becomes very important.Then ant random initialization will make each experiment to get different results.And there is the problem of reducing the accuracy of excessive ants in the non-boundary region,so a gradient based ant distribution method is given.When initializing,the ant individual is more distributed in the area with boundary,and the efficiency of ant colony algorithm is improved.At the same time,the initialization mode of heuristic matrix is modified,the parameter adaptive update is modified and the pheromone update strategy is changed,and the threshold of pheromone on each path is controlled to prevent premature stagnation and find more weak edges.Then the analysis of multiple parameters in ant colony algorithm and the comparison and analysis of coronary artery CT images are carried out.
Keywords/Search Tags:image processing, edge detection, ant colony algorithm, CT image
PDF Full Text Request
Related items