With the change of fish belly contour, accurately controlling the flexible cutting of laparotomy cutter is one of the key technologies to be broken of fish killing machine, in order to improve the intelligent level of fish killing machine, a visual inspection method was designed to control the laparotomy tool automatically. The end-cutting point under the hip fin was extracted by corner detection and screening, the ROI region of interest of the tool entry point is segmented by using the gray median value,and the start-cutting point was extracted by three characteristic parameters: the long side slope, pixel area and ratio of long side to short side of the smallest circumscribed rectangle in each connected region; integrated the calibration factors and the image coordinate of points between start-cutting point and end-cutting point, the world coordinate of the abdominal contour was obtained, which was used as the controlling input information for the real-time flexible adjustment. The results show that the experiment verified that the positions of the start-cutting point and end-cutting point selected in the actual machining are consistent with the positions of the two points obtained in this study. At the same time, it is verified that the abdominal contour obtained after camera calibration coincides with the real fish belly contour. The research showed that this method can solve the cutting length and depth problems in traditional fish killing machines, and it can provide data support for the accurate control of laparotomy tools in fish killing machines.