基于低秩矩阵的自适应边缘检测算法

An Adaptive Edge Detection Algorithm Based on Low-rank Matrix

针对传统边缘检测算法处理含噪图像存在边缘模糊、缺失等问题,为提高算法的抗噪性与边缘完整性,提出了一种基于低秩矩阵的自适应边缘检测算法。首先使用低秩矩阵理论将图像中稀疏噪声矩阵分离,在去除噪声的同时能完好保留图像边缘,再根据图像灰度采用迭代循环的方法进行阈值分割选取,最终提取出连续完整的边缘。实验结果表明,使用该算法在混合噪声情况下具有较好的抗噪声能力与算法鲁棒性,提高了边缘信息的完整性,在视觉测量的应用取得了较好的效果。

In order to improve the noise immunity and edge integrity of traditional edge detection algorithm which exists problems of blurring edge and edge loss in processing noise images, an adaptive edge detection algorithm based on low-rank matrix was proposed. First, our algorithm separates the sparse noise matrix of an image with low-rank matrix theory, so it can remove noise and also keep the integrity of image edges. Then, the proposed algorithm uses gray values to calculate threshold by segmentation and selection. At a result, we extract a complete and continuous edge. The experimental results show that our proposed algorithm has better noise immunity and robustness in the case of mixed noise. This algorithm improve the integrity of the edge information and get the better result in vision measurement.