Aiming at the problem that the approximation result is not adaptive to the scale change and depends on the selection of the starting point in the digital curve approximation

an improved contour polygon segmentation approximation method is proposed. Firstly

the contour sharpness is calculated to determine the corner point

and the contour is segmented according to the corner point and set it as the starting point of the polygon

which reduces the dependence of the approximation result on the starting point. On the basis of splitting and merging algorithm

a polygon approximation method with the arc length ratio as the threshold is proposed to improve the robustness of the polygon algorithm to scale changes. The experimental results show that the approximation results are robust to scale changes and have certain significance for feature extraction of contour polygons.