To solve the problem of uncertain position of camera and rotary milling cutter blade and improve the timeliness of image processing, a milling cutter wear detection method based on machine vision is proposed. According to the structure similarity index, the image quality of the tool was judged, and the image acquisition interval angle coefficient was introduced, and the image acquisition interval angle and spindle speed were determined. Features from accelerated segment test (FAST) algorithm was used to achieve fast and accurate adaptive cutting of tool wear area. Based on FAST feature points, an adaptive threshold segmentation method was proposed to effectively extract the edge of the wear region. Hough transform and minimum external rectangle method were used to correct the inclination angle of the main cutting edge, and then the average width of the wear zone B was extracted. Finally, the milling test was carried out. In 16 groups of tests, the maximum, minimum and average errors between the calculated value and the real value were 4.76%, 0.91% and 3.63% respectively. The experimental results show that the proposed method can obtain high-quality images of all milling cutter wear regions when the spindle is rotating, and then extract wear parameters efficiently and accurately.