Abstract: In order to solve the problem that it is difficult to measure the wear of small tools in complex machining environments, an on-machine tool wear detection method based on machine vision is proposed. First, a method of adaptive segmentation calculation for knife edge images is designed. The original image is segmented by using the improved Gaussian mixture model segmentation algorithm. Subsequently, the noise is reduced by using the improved nonlinear filter, effectively retaining edge details while reducing error during fuzzy noise reduction process, thus achieving automatic segmentation, denoising, and measurement of tool image. The single camera tool image acquisition mechanism is further designed, and cutting experiments are carried out. The results show that the errors in the tool side, bottom wear area, and the maximum wear bandwidth are 3.88%, 5.41%, and 6.26%, respectively, compared with the traditional electron microscope measurements, achieving micron-level accuracy. Compared with traditional wear measurement methods, this system and algorithm enables faster image denoising with clearer edges. Irregular wear areas are calculated by pixel points, which effectively solves the problem that small tool blades are small and difficult to observe and measure, and can meet the requirements of on-machine detection of small tools.

Key words: machining tool, machine vision, image denoising, image segmentation, wear measurement