Resonancia estocástica para el mejoramiento del contraste y calidad en imágenes acústicas de sonar de barrido lateral

摘要

This work describes a novel technique of stochastic resonance to improve the contrast and visual quality of side scan sonar acoustic images. The output image is formed by averaging binary images obtained by repeatedly adding white Gaussian noise to the input image and thresholding. The output image contrast and structural quality are quantified by the root mean square contrast and correlation function between output and input images, respectively. The dependence between the contrast and structural quality with respect to the number of binary representations and the intensity of applied noise forming the output image is discussed. A comparison plot between structural quality and the contrast to determine desired characteristics of the output image is considered. The results obtained by the proposed technique are compared with the histogram equalization and stretching processes utilizing the GIMP software (GNU Image Manipulation Program).