Research On Tool-wear State Monitoring Based On Digital Image

Tool wear condition monitoring technology is a key technology in automation production.It is one of the main means to reduce the manufacturing cost, reduce environmental hazards, andensure the normal operation of manufacturing system and the quality of products. Tool wearcondition monitoring system laid the basis for modern manufacturing system, automation,flexibility. And its research on the cutting tool state of real-time monitoring has become a hotspot in the study of many countries and become the important technical key which recognized byvarious countries.The tool wear condition monitoring method based on digital image is mainly studied in thispaper. The visual characteristics are extracted from work piece surface texture images in varioustool wear condition and cutting tool edge surface wear image. The connection between the imagefeatures and tool wear state is established, which in order to realize the research of tool wearcondition monitoring based on digital image. This paper respectively research from two aspects,work piece surface texture image analysis and cutting tool edge surface image analysis. Thesurface image texture extraction method and the image segmentation method of tool wear aredetailed study. According to the deficiency of the traditional method, this paper improves thetexture extraction efficiency and accuracy, and the accuracy of tool wear image segmentation.This research contents are as follows:This paper puts forward the texture image preprocessing method to resolve the problem ofimage information redundancy, uneven illumination, noise effect, etc. Automatic cut algorithm isput forward for the first time to cut the work piece surface image. The second order statisticmethod is used to image illumination correction, and median filter method is used to remove theimage noise, in order to complete the work piece surface texture image preprocessing. The workpiece surface image texture analysis method based on Hough transform and stroke lengthstatistics is then put forward in this paper. Edge image is detected based on image texture structure information by using canny edge detection algorithm, which greatly reducing thecalculation workload through remove the irrelevant texture information. Hough transform is thenused to detect straight line information of edge image. The average length and average anglecharacteristics are extracted from edge image by using stroke length statistical algorithm, andwork piece surface image texture feature extraction is effectively and accurately completed.This paper study the processing of the collected cutting tool flank face image, and thenpropose tool angle rotation correction and product dust tumor removal method which based onHough transform, realizing the tool flank face image preprocessing. The tool flank face imageanalysis method based on Markova random field (MRF) is put forward in this paper. Tool wearimage segmentation model is constructed by using Markova random field, and relaxationiteration algorithm is added to improve segmentation effect. The wear area of tool flank faceimage is obtained through accurate segmentation. Then this paper puts forward theeight-connected chain code boundary search algorithm to search the edge of wear area. Imagemean wear VB value is extracted according to the boundary search results. As a result, tool flankface image feature extraction is accurately completed.This paper construct the tool wear condition monitoring system based on digital image andgive the system implementation process, using the work piece surface texture image analysismethod and tool flank face wear image analysis method which proposed in this paper to realizetool wear monitoring. The actual availability of tool wear condition monitoring system based ondigital image is analysis also.