Underwater Image Stitching Method Based on AUV-guided Light Source Optimization

Abstract

Background: Image stitching is the procedure of combining two or more pictures having the same feature point location information into a panoramic image, which is useful in target discovery, seabed research, and military applications. Most of existing underwater picture stitching technology makes use of well-lit photos, however natural light gradually diminishes during propagation of underwater autonomous submersibles plunge further into the sea. When the image focuses on the regions of lighting concentration and the larger dark, image details are lost and feature points are not matched, the perspective transformation matrix obtained does not reflect the mapping relationship of the entire image, resulting in a poor stitching effect and making it difficult to meet practical application requirements. Objective: This study aims to obtain underwater images with a good enhancement effect and improve feature point matching image. Methods: An adaptive image enhancement method based on adaptive light source optimization is proposed in this paper, underwater photos are preprocessed to enhance images for the feature registration. Results: The experimental consequence indicate that the improvement algorithm can improve picture standard with better detail performance and color recovery by preprocessing submerged photographs. Conclusion: It is accomplished that high exactness for images stitching by adding the feature points of the enhanced image for feature alignment.