Piecewise fitting of distorted image edge contour

作者：Li Tan, Qiongqiong Wang, Haonan Zheng, Yi Si · 年份：2022 · DOI：10.1145/3584748.3584756 · 研究领域：Image Processing Techniques and Applications、Image and Object Detection Techniques、Advanced Measurement and Detection Methods

Subpixel edge localization, a key technique for visual measurement, has been widely studied in recent years. In this paper, gaussian fitting algorithm and Zernike orthogonal moment based edge detection technology are used to extract the edge contour of the image, BP neural network is used to determine the specific form of image geometric distortion, and the target image corrected by product image is obtained by backward gray interpolation method. Differential and fitting are combined, and matlab is used to calculate the length of each contour. The final curve equation is obtained by iteratively comparing the fit error and the threshold value to obtain all the segment points of the profile and fitting the least squares to the coordinates within the segment interval.