Adaptive motion-compensated temporal filtering of sector scan sonar image sequences

Abstract

We present an adaptive weighted temporal averaging filter with implicit motion-compensation for effective object enhancement in sector scan sonar image sequences. Visual blurring artifacts introduced by the temporal filtering process due to motion of the sonar platform are minimized by accurate motion estimation and compensation. An algorithm is proposed to perform object boundary extraction for better motion estimation. Motion estimation is performed directly on polar image sequences using cross-correlation followed by a Minimum Mean Square Error (MMSE) method. Each pixel of the filtered image is computed as the weighted average of the image pixel values over successive frames after motion compensation. The performance of the proposed filter is tested using real sector scan sonar image sequences and the results are compared with those obtained using the temporal averaging and motion compensated temporal averaging filters.