An Improved Sobel Face Gray Image Edge Detection Algorithm

In this paper, an improved Sobel edge detection algorithm is proposed to overcome the shortcomings of traditional Sobel edge detection operators, such as the limitation of detection direction in horizontal and vertical directions, and the need to set detection threshold artificially. Firstly, the detection direction is improved, based on the horizontal and vertical detection directions, two directions of 45 degree and 135 degree are added, which can detect the edge information of multiple gradient directions of the image. Secondly, considering the overall and local gray level of the input image, an edge judgment threshold is adaptively generated to make the detected image edge more complete. Finally, the multi-directional detection and adaptive threshold generation are combined. The experimental results show that the improved Sobel edge detection algorithm can extract more direction edge information, and the edge boundary is clear, which has better robustness to noise interference.