A Method of Infrared Image Mosaic Based on Keyframe Matching

Infrared images usually have a narrow field of view, requiring splicing of multiple image sequences to meet the application requirements of wide field of view and high resolution. However, due to the large overlap between adjacent frame images in infrared image stitching, it can lead to accumulated stitching errors and increased computational costs. An infrared image mosaic method based on keyframe matching is proposed. Firstly, key frames are selected as the stitching object based on the similarity of image statistical vectors. Then, the SURF algorithm is used for feature extraction and image registration. Finally, three typical image fusion methods are used to complete panoramic image stitching. The experiments results based on on infrared image datasets show that this method can significantly reduce computational complexity while ensuring the quality of stitching.