Study And Implementation Of Fiber Ingredients Automatic Detection System Based On Microscopical Images

Automatic fiber identification deals with many research fields, such as image processing, pattern recognition, computer vision and so on. The traditional fiber identification method has many shortcomings. With the development of computer image processing technology, the research of fiber composition automatically identification has made great progress. However, the research about the automation fiber identification by using computer technology is still a considerable difficulty. The application and research of the field is still relatively few at home and abroad, so there are still many problems to be studied and solved. This paper is a part of the research sponsored by the Foundation of National Excellent Doctoral Dissertation of China and Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry. The research is also sponsored by the Shanghai Entry-Exit Inspection and Quarantine Bureau of PR China. The research mainly deals with the algorithms of automatic segmentation and adherence separation of the fiber image and the subsequent recognition of fiber image.Automatic fiber identification system consists of three steps. First, the system need to preprocess the microscopic fiber cross sectional images obtained before, in order to counteract the effects of lighting or to remove it altogether. Second, achieve single fiber target contours by using image segmentation and adherence separation algorithm. And then, the system extract features by using SIFT, and train classifier with the support vector machine(SVM).Accurate separation and extraction of fiber’s microscopic image is required prerequisites for the feature analysis of the fiber. An edge detection algorithm based on level set combined with clustering idea is proposed as there were no significant differences between target area and background, and each cotton fiber has one lumen. Firstly, the small region is obtained by binarization algorithm, and the outer contour is got by level set algorithm based on it. Then combined the fiber’s outer contour with OSTU, and the rough edge of the fiber is appeared after the burr and broken edge are eliminated. And the seed area of the fiber is emerged by using flooding algorithm and dilation algorithm. Finally, the adherence separation algorithm found on clustering idea is applied to gain the separated fibers. After that, the system can extract feature by using feature extraction algorithm and identify the fiber by training the classifier. The experimental results show that this algorithm cannot only ensure the integrity and the continuity of the fiber’s edge, but also segment each fiber rapidly and reliably.Proven by the experimental results, the proposed image segmentation and adherence separation algorithm has no bias occurred to the separated fibers compared to the conventional algorithms that separate objects based upon the contour calculation solely. Satisfactory segmentation result and identification rate are obtained for the application of mass overlapped fiber separations.