An Image Processing Algorithm Adapted to Precision Measurement for Tool Profile

摘要

Precision measurement is required of the machining process in order to maintain the quality of products. When tools, such as a drill and an endmill, are measured by image processing, measurement is difficult due to complicated profile, distortion at high magnification, surface condition and reflection of lighting, etc. Therefore, a special image processing algorithm is required to solve these problems. In this paper, an image processing algorithm applied to precision measurement of tool profile is proposed. In this algorithm, a histogram of gray levels and a histogram of values which is differentiated by Sobel operator are combined and neural network decides threshold levels for edge profile. Thus this process is enabled for measurement of images in various conditions flexibly. To confirm validity of this algorithm, edge recognition is performed on different samples. Moreover, in comparison to the method not applying neural network, the superiority of edge detection by this algorithm is proved.