This article establishes a system of anti-bud injury in sugarcane cutting based on computer vision using MATLAB software as a development platform. The seedcane cutter box cuts twice each turn. Match the accurate shutter trigger time interval according to the rotation speed of the cutter , the transportation speed of sugarcane, in order to make the position of collected image of the target happen to be the next cutting site. Image processing is based on digital image processing tools of Matlab, and the image is processed from two aspects of the cane edge of the curve smoothness and internode surface color using Matlab by which sort out the image suitable for the identification of sugarcane internode, and then deal with the collected images by filtering denoising and binarization to determine whether the cutter cuts sugarcane internode. When the recognition threshold is 3500, the recognition accuracy of this system is 99%. After modifying the recognition threshold to 4700, its recognition accuracy is 100%.