提出了一种新的图像边缘检测算法, 首先研究分析了典型边缘检测算法的边缘检测能力、存在的不足及原因; 然后在综合Sobel边缘检测算子优势的基础上, 提出了新的边缘检测算子, 将其大小设置为5×5, 参与检测的算子元素增多, 提高了边缘检测的灵敏度, 同时算子考虑图像低通滤波特点, 有效滤除了噪声; 最后对本算法有效性进行了实验验证。实验结果表明, 该方法在一定程度上可以解决边缘断裂和双边响应问题, 可尽可能完整的将属于同一目标对象的边缘轮廓提取出来, 为后续应用提供更完整和更精确的边缘数据。

A new image edge detection method was proposed. Firstly, full research and analysis of the ability of the typical edge detection methods were given, including shortcomings and the reasons. Then, based on the advantages of the Sobel edge detection operator, a new edge detection operator was proposed, whose size was set to 5×5, and with the number of elements in the detection operator increasing, the proposed method improved edge detection sensitivity. While the low image filter feature of the detection operator was taken account in, the method could filter out the noise effectively. Finally, the effectiveness and superiority of the method were verified by experiments. Experimental results indicate that the method can solve edge fracture and bilateral response problems, to some extent. The edge of the same object can be extracted as complete as possible, providing more complete and more accurate edge data for subsequent applications.