The invention relates to a chip character recognition method, in particular to a chip character anti-interference recognition method, which comprises the following steps of: (1) setting a camera and a light source; (2) calling a template, acquiring a frame of image by a camera, matching the frame of image with the template, and extracting a chip image only containing a chip; (3) extracting a chip character image only containing a character region; (4) sequentially carrying out noise reduction processing and threshold segmentation on the chip character image to obtain a binary image, and carrying out character segmentation on the binary image to obtain single characters; (5) performing character defect detection to obtain defect characters; (6) performing defect character optimization and character recognition: (6-1) performing defect character optimization; and (6-2) performing artificial neural network character recognition to obtain a final character string. According to the method, part of interference of the environment can be effectively reduced during image acquisition, errors of the placement angle of the chip are prevented, and the method cooperates with other detection equipment, so that the labor cost is remarkably reduced, and the production efficiency is improved.