Fusion of SONAR Image using Enhanced Multi-Scale Transform and Sparse Representation Method

Objective: The main goal of this paper is to enhance fusion of sonar image thereby achieve the better entropy, standard deviation and PSNR value. Methods: Multi-Scale Transforms (MST) and Sparse Representation (SR) methods are the two well-known methods used in image and signal representation theory. The novel image fusion framework is proposed in this paper by the combining enhanced MST method and SR based image fusion. The proposed scheme consists of three phase; first de-noised the sonar image using DTCWT with mean filter; second select the obtained pixels or features from sonar image using Novel PCA method; third obtained fusion image using Enhanced MST with SR structure. Findings: It is good at suppressing noise, especially for images with a higher noise level. The advantage of the proposed enhanced MST with SR technique than conventional MST with SR method is different level of decomposition using four popular MST methods; DWT, DTCWT, CVT and NSCT. The proposed method obtained better result in terms of entropy, standard deviation compared to conventional method. Applications: To realize earth surfaces with focus on underwater applications like depth sounding, sea-bed imaging and fish echolocation the SOund Navigation And Ranging (SONAR) technology is used.