Abstract: Optimization methods of edge linking and linear fitting are researched to improve the efficiency and accuracy of the industrial computed tomography (CT) image processing system. For edge linking, the linking mode is decided by the size of the gap between edges. The smaller gap is linked with a line and the bigger gap is linked with an optimal path calculated by fuzzy theory. For linear fitting, feature points of the contour are obtained, and then an improved method of set intersection operation for line recognition is adopted to fit the feature points. The application proves that the optimization methods are effective and feasible.