Self-adaptive overlapped image rotation method

The invention discloses a self-adaptive overlapped image rotation method, which comprises the following steps: on the basis of using an A-KAZE feature extraction algorithm and a bidirectional KNN feature matching algorithm as image registration, judging whether relative rotation exists in an overlapped area of two images by detecting whether feature matching pairs of the two images are crossed or not; rotating the images rotated by limiting registration areas of the images, and meanwhile, shortening the registration time of unnecessary areas; and adjusting the rotation angle of the images to obtain ordered images with parallel overlapping areas. The right image is selected as a to-be-rotated image, the left image is selected as a reference image, the to-be-rotated image can be rotated to the position where the overlapping area of the to-be-rotated image is parallel to the reference image, and self-adaptive rotation of the images at any shooting angle is achieved. 本发明公开了一种自适应的重叠图像旋转方法，在使用A-KAZE特征提取算法和双向KNN特征匹配