As a medical device for diagnosis and treatment, the distortion of medical rigid endoscopes directly affects the accuracy of the surgeons judgment of the surgical position. Therefore, the measurement of endoscopic relative percent distortion is an urgent problem to be solved. A relative percent distortion detection system based on image processing was designed, and the images of the calibration target plate were acquired by a CCD camera. After vertical calibration of endoscope visual axis and target plate, the distortion target plate was replaced to collect images. The computer filtered the obtained image, extracted the region of interest and carried out the size statistics, and the relative percent distortion was finally obtained. Using a 30° laparoscope as the measurement model, the experiments show that the relative percent distortion value measured by the designed calibration system at 70% of the full field of view of the endoscope to be measured is better than that of the system without vertical calibration by 2.2%. The results show that the system has higher consistency of detection results and higher accuracy than the system without vertical calibration.