Research Of Sonar Image Fusion Technique

During investigation in turbid water, high-resolution sonar can obtain multiple backscattered images. However, in signal processing procedure, single image is processed usually. This paper proposes sonar image fusion algorithms. The algorithms can enhance the image quality based on complementary information, while improve the image reliability based on redundant information from different images. The detection capacity is thus boosted.This paper based on the framework of digital image fusion techniques, studies the matching of features between different images and multi-scale fusion of sonar images.Invariant points are first extracted via two algorithms, SIFT and SURF. Feature matching and coordinates transformation are then applied. Performances are compared between the above two algorithms. And the results of different multi-scale image fusion framework are analyzed by evaluating the image fusion performances, from both directions of multi-scale geometric transforms and fusion strategy. Finally, a sonar image fusion framework based on morphology strategy is established.Numerical experiments are conducted between different sonar images, transformations and strategies. The experiment results demonstrate the performance of the above sonar image fusion framework based on morphology strategy.