Design Of Barrel Bore Surface Flaw Detection System Based On Pattern Recognition

Barrel is an important component of the artillery, and the quality of this part is directly related to such as the accuracy of the target, the flight speed of artillery shells and the safety of operators in the military operations. Therefore, the detection of artillery barrel bore is very meaningful, and have got more and more people’s attentions. At present, what is the mainly problems about the flaw detection of barrel bore surface are sampling and the slow speed of defect image recognition. To solve these problems, this topic has designed the system of barrel bore surface flaw detection which using the image processing as the background, combined with pattern recognition technology. Compared with the traditional flaw detection system, this topic has abandoned the method of artificial detection. At the same time, it not only saves resources, but also makes the sampling process more intelligent and the results of the test seems more accurate.The part of image processing uses wavelet transform, the gray histogram of image preprocessing and threshold method. Using the method of exhaustive and branch threshold algorithm of image has completed the feature selection, and based on the decision tree classifier design has completed the feature extraction. In the part of pattern recognition, through the comparison of several common pattern recognition methods considering there are so many flaw types of barrel bore surface, it adopts decision theory method for pattern recognition which has higher efficiency. Classification of flaw uses the theory of decision theory, according to the characteristics of the defect, it uses the ratio of area and the diameter as the parameter. Through the threshold of extraction, the detected object in the image pixel extracts coverage area and the total image pixel ratio. Through the actual total area of the image accurately calculates the defect area of the object which has been tested. And it has completed the design of the flaw detection system of barrel bore surface.Using the barrel bore surface flaw as the subject of this paper which combined with pattern recognition and image processing has completed the design of the flaw detection system of barrel bore surface. Through the way of the combination of hardware and software, the system has achieved the precise identification of bore surface rust, defect, iron stains, scratches and so on. And the result of detecting system has reached the expected purpose of the experiment.