A fast super-resolution algorithm with despeckling for multi-frame sonar images

To obtain a high-resolution sonar image from a set of shifted and decimated low-resolution sonar images, a fast super-resolution algorithm with despeckling is proposed in this paper. According to the sonar imaging model, low-resolution images can be viewed as speckled and downsampled versions of the high-resolution image. Based on L1 norm minimization, original high-resolution sonar image can be obtained by non-iterative data fusion. Then the high-resolution sonar image is despeckled in Curvelet domain to get the final high-resolution image. Without iterative operation, the proposed algorithm is fast. Final simulation results indicate that the proposed algorithm is superior to other super-resolution methods.