Panoramic Image Stitching for Distributed Infrared Sensors

Aiming at the real-time and stability requirements of infrared distributed sensor panoramic image stitching in airborne missile warning systems, a four-channel infrared image panoramic stitching method based on the Accelerated-KAZE(AKAZE) algorithm and multi-band fusion is proposed. This method first extracts the region of interest (ROI) from the images captured by four infrared sensors and performs horizontal flip preprocessing. Then, the cylindrical projection model is adopted to eliminate image distortion, and the AKAZE algorithm is used for feature detection and matching. Finally, the Laplacian pyramid multi-band fusion technique is employed to achieve seamless stitching of overlapping regions. Considering the particularity of 360° circular stitching, a closed-loop verification mechanism for the first and last images is designed, and an adjustable left image display ratio parameter is introduced to optimize the fusion effect. Experimental results demonstrate that the proposed method can effectively achieve high-quality panoramic stitching of four-channel infrared images, with natural transitions and complete detail preservation, meeting the application requirements of real-time panoramic monitoring for airborne missile warning systems.