基于改进PM扩散方程的NSST域声呐图像融合  

NSST Domain Sonar Image Fusion Optimization Algorithm Based on Improved PM Diffusion Equation

作　　者：冯时宇 张钟铮[1] 赵振贺 周凯 FENG Shiyu;ZHANG Zhongzheng;ZHAO Zhenhe;ZHOU Kai(North Automatic Control Technology Institute,Taiyuan 030006,China)

机构地区：[1]北方自动控制技术研究所,太原030006

出　　处：《火力与指挥控制》2024年第6期200-207,共8页Fire Control & Command Control

摘　　要：针对声呐图像边缘特征信息模糊丢失等问题,为使图像尽可能包含更多边缘和纹理特征等信息,提出一种基于改进PM扩散方程非下采样剪切波变换声呐图像融合改进算法。采用非线性扩散滤波对NSST多尺度分解进行改进,将多幅源图像分解为高频系数和低频系数,并且结合局部能量及PCNN方法对高低频系数进行融合,经过逆变换重构为融合图像。经过实验验证与其他算法相比,所提方法能够较好地保留源图像中的边缘特征信息,在含噪声声呐图像上表现较为明显。To address the issue of blurring and loss of edge feature information in sonar images,an improved algorithm for sonar image fusion based on improved PM diffusion equation according to Non-subsampled Shearlet Transform is proposed to make the images contain as much edge and texture features,others as possible.The multi-scale decomposition of NSST with nonlinear diffusion filtering is improved,multiple source images are decomposed into high-frequency and low-frequency coefficients.The local energy and PCNN method are combined to fuse high and low frequency coefficients.A fused image is reconstructed through inverse transformation.After experimental verification and comparison with other algorithms,this proposed method can better preserve the edge feature information in the source images and performs more obviously in noisy sonar images.

关 键 词：非下采样剪切波变换 声呐图像 图像融合 PM扩散方程 边缘提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]