Research On Surface Crack Identification Algorithm Based On Image Processing

The surface crack detection of parts is an indispensable part of the quality testing.We can often see a variety of safety accidents due to parts in the work process by the alternating load end due to load stress during loading or not is due to stress concentration resulting in sudden fracture of parts,harm serious or even cause the operation of the safety of the lives.Therefore,it is necessary to check the situation of each component before the machine work,so as to avoid the occurrence of safety accidents.However,due to the fact that there are no mature reliable detection methods at present,most of the people at home and abroad use manual detection to detect cracks in various parts.This method of testing has many drawbacks,and does not fundamentally improve the efficiency of the work.Moreover,the accuracy of the artificial recognition is not high,which is not conducive to the machining of the parts of the parts.In this paper,an image processing function based on MATLAB image processing toolbox is used to extract the related features of the cracks on the surface of the collected parts,so as to facilitate subsequent use of the machining robot to provide track parameters for processing the cracks on the parts.So which suitable algorithm is used to identify the crack image? In this paper,the surface cracks of the parts are extracted from the following steps.First,the surface crack image of the parts is properly pretreated.First,the gray distribution histogram of the crack image is drawn by using the related functions of the image toolbox in the MATLAB software.Due to the lack of light and the texture of the metal itself,the gray value of the surface crack usually concentrates on the lower part of the gray value.Therefore,it is necessary to stretch the gray value to a certain extent,and improve the contrast of the whole grayscale image.Digital image is always interfered by some external signals,no matter in image acquisition or in some other steps.For noise signal interference,we need to take some reasonable measures to filter out some external interference signals.In this paper,we compare all kinds of noise removal algorithms and their corresponding results,and then modify some image processing algorithms based on the results of image processing,and get a new median filtering algorithm.Then the effect of different algorithms is compared again,and it is found that the effect of median filter is better than that of other algorithms.The pre processed digital image must be followed by the segmentation of the digital image,that is,to turn the image into a "0-1" image.To turn the image into a "0-1" image,it is necessary to find the optimal threshold to segment the image.After comparison,it is found that the K-Means clustering segmentation algorithm is more satisfactory than the other segmentation algorithms.After the digital image is divided into "0-1" image,there will still be burr noise,it must adopt the way to filter the morphology of "0-1" in the image edge burr interference information,and finally get the image binarization of a relatively smooth edge.The final purpose of digital image processing for surface cracks is to obtain various geometric parameters of the crack,such as the length of the crack,the circle degree and other geometric parameters.On this basis,the mathematical function expression of the crack image curve is obtained by fitting the trend of the image by the least square method,which provides the track basis for the following robot processing.