Realtime Registration for Phased-Array Submarine Sonar Image

Image registration is the basis of subsequent moving recognition and image stitching. Because underwater environment intermingles with strong noise and the sonar transducer is in the moving state, the two adjacent sonar images have only partial overlap. So it is necessary to solve the problem that the registration effect is deteriorated due to the improper selection of the initial parameter and the mismatch of the control point pairs in the nonoverlapping region. In the paper a realtime registration method of phased-array sonar image is proposed. The ship hull translation and rotation matrix are obtained by the differential GPS locator and the attitude instrument as the initial iteration parameters to accelerate the convergence speed of the iteration. The mismatch of the control point pairs is solved in the registration process of adjacent sonar frames by reverse projection. The nearest point search method is used to search effective control point pairs in the overlapping region of two adjacent frames to avoid improper selection of control point pairs which results in worsening of registration effect. Experimental results show that the proposed method can achieve effective and rapid registration of the submarine sonar image.