Adaptive FAST Corner Detection Algorithm Based on Regional Grayscale Change

According to the FAST corner detection algorithm applies only on a single type of corner and a single threshold is not suitable to the whole image, an adaptive FAST corner detection algorithm based on regional grayscale change is proposed. In this algorithm, Gauss filtering of images; then based on gray difference selected corner candidate; Besides, by introducing an adaptive threshold, candidate corners are classified according to adaptive gray level threshold. Finally, the best match corners are selected by using different templates in accordance with the classification results. Experimental results show that: the proposed improved algorithm can not only effectively overcome the threshold selection inappropriate resulting in loss and redundancy some corners, extract more efficient corners, but also has good robustness to noise.