In order to reconstruct the digital contour curve with arc and straight line, a method was proposed to integrate the segmentation of digital contour curve based on the slope difference judgement method and the dynamic segment merge method. First, the slope difference between the front and back directions of each point on the contour curve was calculated using straight line fitting. The process of obtaining the slope difference used the adaptive length fitting window to balance the relationship between the precision and the velocity. Then the adaptive smoothing method was applied to the direction slope difference curve. Based on the smooth slope of the slope curve, the intermittent point was extracted as the selected section segmentation point. Finally, a dynamic merge algorithm based on the perceptual error was used to select a segmentation scheme with the smallest perceptual error from the selected segments as the final result. The simulation results show that the error of this segmentation method is less than 1%. The experimental results of the contours of bearing oil groove as the actual test cases show the feasibility of this adaptive contour segmentation method in practical application.