Aiming at the problems of low spatial resolution and narrow field of view of infrared images, which leads to low image registration rate and poor realtime performance, an infrared panoramic mosaic algorithm based on the region of interest (ROI) of image is proposed. The method is built to run as follows: firstly, the ROI of the two adjacent infrared images is calculated according to their position relationship. Then, the SIFT feature points are extracted in the ROI and used as moving targets, thus the feature points in the infrared image to be registered are determined by combining the KLT tracking algorithm. Secondly, the RANSAC algorithm is used to eliminate mismatches. Finally, the pixel level fusion method is used to eliminate the stitching traces and an infrared panoramic image with stable resolution and wide field of view is synthesized. The experimental results show that compared with the traditional SIFT algorithm, the registration rate and the running time of the proposed algorithm is improved by 3.491% and nearly 50%, respectively, which can realize the seamless splicing of multiframe infrared images accurately and effectively.