摘要： 分析仅以梯度模值作为依据的Canny边缘检测算法的不足，提出一种以曲率加权梯度模值为判据的改进算法以满足基于边缘信息的图像编码压缩对边缘提取的严格要求。实验结果表明，改进的Canny算法能更有效地抑制对于表征图像对象无重要意义的杂乱边缘，更好地反映图像的结构特征，性能优于现行的Canny算法。

关键词: 边缘检测, Canny算法, 图像编码压缩, 曲率加权梯度模值

Abstract: This paper analyzes the shortage of the current Canny edge detection algorithm which only uses the Gradient Module(GM) to characterize edges and proposes an improved algorithm based on Curvature-Weighted Gradient Module(CWGM), which meets the special need of edge extraction for image coding compression. Experimental results show that the improved Canny edge detection algorithm can compress the stray edges which are non-significant to object representation more effectively, and its performance is better than the current Canny algorithm.

Key words: edge detection, Canny algorithm, image coding compression, Curvature-Weighted Gradient Module(CWGM)

中图分类号: 

• TP391.4