Feature Extraction Algorithm Based on Otsu's Method and Hybrid Hierarchical Grid-Quadtree Partitioning

Aiming at the sensation to changes in lighting conditions within the traditional ORB fixed threshold strategy in visual SLAM, a feature extraction ORB algorithm based on Otsu's method combined uniform distribution with hierarchical grid-quadtree hybrid algorithm (OU-ORB) is proposed. The thresholds of ORB feature extraction algorithm are calculated based on Otsu's method to improve the feature extraction rate of the algorithm on different images. Uniform distribution with hierarchical grid-quadtree hybrid algorithm is used to select feature points. Experimental results show that compared with the feature extraction method in ORB-SLAM2, the OU-ORB algorithm proposed in this paper improves the uniformity of feature points by 9.35% while ensuring the feature extraction rate.