基于NSCT循环抽样的声纳图像去噪方法

针对声纳图像对比度差、不易判别边缘及细节的特点，提出一种有效的声纳图像去噪方法。对含噪声纳图像循环抽样后得到多幅声纳图像，分别执行NSCT变换，选取适当的阈值对变换的系数进行取值，筛选后的系数进行NSCT逆变换；将得到的多幅去噪后的图像空域平均后输出去噪后的图像。通过实验数据比较，此改进算法好于其他经典方法，提高了处理结果的信噪比及边缘保持效果。由于充分利用了NSCT变换的平移不变性，使去噪后的细节保护能力增强。

Sonar image denoising method based on NSCT circle sample

Through studying the characteristics of sonar image which include poor contrast and unrecognizable edges and detail, an effective sonar image denoising method was proposed. Firstly, obtained many sonar images by circles sample for noise sonar image then carried out NSCT respectively, selected appropriate threshold to choose value of transformed coefficient, NSCT inverse transformed to select coefficient, obtained denoised images by image airspace average finally. Compared to experimental data, this method is better than other classical methods and improves PSNR of processing result and edges keeping of the sonar image effect. This method takes full use of translation invariant of NSCT and enhances detail keeping ability.