首页|结合Lee滤波的NL-Means声呐图像滤波方法

结合Lee滤波的NL-Means声呐图像滤波方法

扫码查看
声呐图像中存在的散斑噪声不仅极大地影响了图像质量,还对图像的后续分割、增强、边缘检测等增加了很多负面影响.针对这一问题,提出了结合Lee滤波的NL-Means算法:先用Lee滤波算法对声呐图像进行一次滤波,再用滤波后的图像计算NL-Means算法中的权重部分的权值,最后用NL-Means算法结合该权重进行滤波.改进的滤波方法避免了权重计算过程中由于噪声的影响而使导致权重偏小的问题,精确了 NL-Means算法中的权重分配.实验证明,改进方法对声呐图像的散斑噪声不仅有更好的抑制效果,还提升了图像的质量.
NL-Means sonar image filtering algorithm combined with Lee filtering
Scatter noise is prominent in sonar images,and the presence of scatter noise greatly affects the quality of the image,adding a lot of negative impact on the subsequent segmentation,enhancement,edge detection and other processing of the image.To address this problem,an NL-Means algorithm incorporating Lee filtering is proposed,which first uses Lee filtering algorithm to filter the sonar image once,and then uses the filtered image to calculate the weights of the weights in the NL-Means algorithm,and then finally combines this weight with the NL-Means algorithm to perform filtering.This method avoids the problem of small weights due to the influence of noise in the weight calculation process and accurately assigns the weights in the NL-Means algorithm.Experiments have proved that the method has better suppression effect on the scattering noise of sonar image and improves the quality of the image.

sonar image filteringLee filteringNL-Means algorithmevaluation indexes

田原嫄、雷玉峰、郭海涛

展开 >

东北电力大学机械工程学院,吉林吉林 132012

海南热带海洋学院海洋科学技术学院海南三亚 572022

声呐图像滤波 Lee滤波 NL-Means算法 评价指标

国家自然科学基金海南省自然科学基金中国三亚市科技创新专项

61661038420CXTD4392022KJCX83

2024

河南科技学院学报(自然科学版)
河南科技学院

河南科技学院学报(自然科学版)

影响因子:0.557
ISSN:1673-6060
年,卷(期):2024.52(5)