B-spline contour reconstruction algorithm based on image contour concavity and convexity

Image contour reconstruction is an important method for reverse construction of target geometric model. The key idea of such method is to obtain appropriate control points from image contour. Feature point detection-based method is currently the main approach that extract the control points from the image, which will encounter two primary issues: the complexity of the methodology is substantial, and the control points obtained struggle to accurately represent the local details of the contour. In this paper, a method of extracting control points using the concavity and convexity properties of image contours is proposed. Convex and concave pixels that can reflect the local concavity and convexity properties of the contour were considered as control points, and the pixel distribution characteristics of the eight surrounding domain was used to pre-extract the control points. Additionally, constraint conditions based on non-control points were established to further extract the optimal set of control points. Simulated and real examples demonstrate that the contour reconstructed based on the control points extracted by the method has characteristics of high accuracy and good smoothness, indicating that the proposed method is effective and robustness.