The basic principle of SONAR is to use sound to detect or locate objects, typically in the ocean. During acquisition, some pixels of SONAR images could be lost. The recovery of lost or damaged data can be done by using an inpainting algorithm. But, SONAR images are perturbed by speckle noise, so speckle reduction filters should be used to improve the results of the inpainting algorithm. This paper presents a novel and efficient algorithm that combines denoising and inpainting methods applied on real SONAR images. Some simulation results and comparisons show the performance of the new algorithm. The results are analyzed in terms of the Peak Signal to Noise Ratio (PSNR).