An improved three-dimensional block matching image de-noise algorithm and its application in a 3D imaging sonar

There are urgent needs for sonar image processing technology since sonar images generally have characteristics of low resolution and a weak target shielded in a strong reverberation background. A three-dimensional block matching image de-noise algorithm was used to suppress the speckle noise in SAS imaging processing. It usually adopts adjacent sub-blocks of the image for 3D transformation, and attains image de-noise by processing in the transform domain. Furthermore, we now face the challenge of three-dimensional image processing in a kind of 3D sonar, and there are already abundant azimuth sections of three-dimensional images available to be studied. There are more similar characteristics amongst different azimuth sections of 3D images than there are amongst adjacent sub-blocks of conventional image. Thus, an improved three-dimensional block matching image de-noise algorithm is proposed for 3D sonar image processing, in which the effectiveness of image de-noise are produced by ensemble average and transform domain processing. Its application in the 3D imaging sonar is also described in this paper. And because of the greater quantity of elements and greater similarity amongst sub-blocks, the image processing results show that the background noise is being properly suppressed.