Research On The Technology Of Flaw Detection In Castings Of Automobile Wheels Based On Aim Extraction And Match

It is important to realize the automated flaw detection of automobile wheels, in order to improve the efficiency of producing and to reduce the cost. In this paper, the features of X-ray images of wheels are analyzed, then 3 steps are used to realize the automated flaw detection: extraction of aim-region of the flaws; seed-fill-thresholding; fuzzy pattern recognition. A system for automated flaw detection of automobile wheels is developed, which is tested by 100 images. The results showed that this algorithm does not need image preprocessing, many flaw types can be detected, and the images with poor contrast, high noise level, and complex structures can be detected as the same, reference images are not needed, the efficiency of the method is high, which meets the remand of industry.The main research contents of the paper are as follows:Chapter 1: The origins, developments, and the applications of the Image Processing and Pattern Recognition are introduced. It also outlines the importance and development status of the technology of automated flaw detection of automobile wheels, the disadvantages of the methods in existence. In the end, the details of research contents and technology scheme are given.Chapter 2: Analyzing the changes of gray in the pixels caused by various factors in the X-ray images of automobile wheels. The concept and principle of peak locating are presented. And a method for extracting aim-regions of flaws based on peak locating and morphology are presented.Chapter 3: It is analyzed that thresholding and ROI extracting methods are helpless in extracting the information of flaws in wheels. Feasibility of seed-fill-thresholding in it is analyzed. A new seed-fill-thresholding based on flaw pieces and edge detecting in aim-regions of flaws is presented. Two difficulties ofseed-fill-thresholding in extracting the information of flaws in wheels--Creatingseeds and Principle of stop, are solved.Chapter 4: On the basis of having extracted the information of flaws, researching the classification of flaws. The advantages and disadvantages and applications of various methods of pattern recognition are analyzed. Feasibility of the application of fuzzy pattern recognition in classification of flaws is researched. And a new fuzzy pattern recognition based on step-up is presented, which is tested in the end.Chapter 5: All of the method researched--the method for extracting aim-regions of flaws based on peak locating and morphology, the new seed-fill-thresholding based on flaw pieces and edge detecting in aim-regions of flaws, the new fuzzy pattern recognition based on step-up are all applied into practice by developing a system for automated flaw detection of automobile wheels, which consists of two sub-systems, three function modules. In the end, the system is tested.Chapter 6: The summary of the paper, and also giving prospects for future researches.