PDE模型在声纳图像去噪中的应用研究

偏微分方程方法在光学图像去噪中已有很多成功的应用,但用于声纳图像去噪的情况还不多见。针对声纳图像受噪声污染严重的问题,将偏微分方程原理引入到声纳图像去噪中,重点讨论了两种偏微分方程模型:ROF模型和四阶扩散模型。基于这两种模型对声纳图像进行去噪处理,仿真实验证明了偏微分方程去噪算法的有效性,并对比分析了两种模型的去噪性能。ROF模型适用于低信噪比条件下的声纳图像处理,而四阶扩散模型在高信噪比条件下,能够很好地保持图像边缘,但当噪声污染严重时,其去噪后的SNR比ROF模型去噪低了近10 dB,不利于声纳图像处理。

Research on application of PDE model in sonarimage denoising

Many successful applications of partial differential equation(PDE) method in optical image denoising have been mentioned,however the application of PDE in sonar image denoising has been seldom studied before.Aiming at the serious noise pollution problems in sonar image,the principles of PDE used in sonar image denoising are introduced.Two PDE models,ROF model and fourth-order diffusion model are mainly discussed.Sonar images noise can be removed by processing based on these models.Simulation experiments show the effectiveness of PDE method and the property of two models has been compared.ROF model is suitable for sonar image processing under low SNR condition,while the fourth-order diffusion model can maintain image edge well under high SNR condition.But when the noise pollution is serious,its SNR after denoising become nearly 10 dB lower than the ROF model denoising method,so the fourth-order diffusion model is not conducive to sonar image processing.