Experimental Assessment of Image Enhancement and Object Detection Methods Applied to Mechanically Scanned Sonar Systems

In this paper, some efficient methods for acoustic image enhancement and automatic object detection are presented and assessed on a specific class of real data. The aim is to increase the potentiality of mechanically scanned fan-shaped-beam and pencilbeam sonar systems employed during underwater operations. Concerning the fan-shaped-beam sonar system, a set of methods has been assessed that is devoted to increase the visual quality of the generated images. The first step is a scan conversion procedure, to move from a polar coordinate system as that of the sonar to a dense, regular matrix of pixels. An efficient interpolation step, working on the same polar coordinates of the sonar system, has been developed, that performs the computation of the value to assign to every pixel on the basis of the weighted average of the neighbouring acoustic samples. Moreover, a dynamic pixel brightness assignment law, evaluated image by image, is carried out to optimally exploit the available brightness range. The joint application of the interpolation method and the optimised brightness assignment allows one to improve image quality thus providing a simple and effective tool to generate particularly refined acoustic images. Finally, further processing steps allow to reduce typical problems blurring this kind of imaging systems, like speckle noise, multiple echoes and low