Techniques adopted in the post processing of active sonar data from Royapuram site-off Chennai

摘要

A buried object detection SONAR has been developed by the marine sensors and systems group of National Institute of Ocean Technology and the analysis of the data from a specific site is reported in the paper. Handling the unpredictable noise is a major concern in sonar signal processing, especially in buried object detection sonar systems. To improve the signal to noise ratio and also to preserve the boundaries of targets, special post processing techniques are to be applied. Signal averaging is found to be a useful technique in this regard and this paper compares and analyzes various averaging techniques including moving averaging, exponential averaging, and median filter. The exponential averaging with median filter is found to be one of the best suitable methods for noise reduction in detecting buried objects in shallow waters, since it significantly improves the signal to noise ratio by preserving the boundaries of targets. It is observed that the original sonar image with 6% noise level is improved to 0.03 to 0.04 noise variance using the combination of exponential moving average and median filter and the same trend is observed up to 35% noise level when corrupted by Gaussian noise. Performance evaluation of the techniques has been carried out and is quantitatively verified with the data collected during the sea trials.