An Adaptive Corner Detection Method Based on Deep Learning

Corner detection is one of the most important techniques in image processing. But traditional corner detection algorithms are susceptible to complex background, and the threshold needs to be set manually. We propose an adaptive corner detection method based on deep learning in this paper. First, 6900 photos containing the target are taken for training. Then, we detect the target through the Faster R-CNN model to obtain the coordinates of the target area. Finally, we perform an adaptive Harris algorithm on this area to detect corner points. According to the experiment results, the proposed method can not only avoid the influence of complex background, but also improve the efficiency of corner detection.