一种改进的基于Canny算子边缘检测算法

 基于已改进的Canny算子, 用半像素插值方法将已改进的3×3模板扩展为5×5模板, 并利用归一化自相关最大值的方法判断边缘方向, 缓解了传统Canny算子由于噪声影响使检测效果有断边缘或假边缘的现象. 实验采用接受者操作特性曲线评价提出的算法性能. 结果表明, 与已有算法相比, 改进算法的检测效果有所提高, 可以精确边缘方向, 同时减小假边缘点出现的概率, 对线条边缘丰富的图像效果尤其显著.

An Improved Edge Detection Algorithm Based on Canny Operator

Based on the improved Canny operator,the modified templates of size 3×3 were extended to the ones of size 5×5 via half-pixel interpolation method,and the method of maximum value of normalized autocorrelation was used to determine the edge direction so as to mitigate the broken edges or false edges due to the effects of noise.In the experiment,we used receiver operating characteristic curves to evaluate the performance of the algorithm.The results show that the modified Canny operator improves the effect of the detection,for it can accurate the edge direction while reducing the probability of false edge points,especially for that of the images rich in line edge.