Automatic Matching of Infrared Image Sequences Based on Rotation Invariant

Infrared image sequences matching is a key technique in the field of video object orientation, object orientation, recognition etc.. This technique is a process of searching corresponding relationships between pixels in infrared images, which influences the precision of infrared image registrations directly. In order to accelerate the search speed, the Pyramid Strategy is introduced. In view of such characteristics as low S/N ratio and great distortion of infrared image sequences, wavelet threshold arithmetic is also introduced when filtering the infrared images in preprocessing. Considering the characteristics that the rotational angles between the Unmanned Aerial Vehicle infrared images are usually large, SIFT is introduced in image sequences matching, which combines method of correlative coefficient controls to get rid of the wrong matching points. From this way, it achieves a good effect which wonpsilat be influenced by the large-angles of the sensor flat in marching.