摘要
为了改善低对比度、富含噪声以及非均匀亮度的水下图像的分割效果,本文提出一种基于图像灰度波动特征的分割方法,该方法采用了非对称的自适应中值滤波的方法来改善图像的前景目标与背景之间的边缘特性,分别从行和列的方向上检测出大尺度的波峰和波谷,再通过局部阈值分割来确定图像中的前景目标边缘,最后将行与列的分割结果取交集,得到图像分割的最终结果。仿真结果表明:与基于模糊熵或者模糊聚类的图像分割方法相比,本文的算法性能更优,分割效果更好。
A novel image segmentation method is proposed based on grayscale wave to improve the segmentation results of underwater images with low contrast,considerable noise,and non-uniform brightness.An asymmetrical adaptive median filter method was adopted to improve the edge characteristic between the foreground object and image background.Large-scale wave crests and troughs were detected in the row and column directions,respectively.The edges of foreground objects in the image were determined through local threshold segmentation.Finally,the row and column segmentation results were intersected to obtain the final results of image segmentation.Simulation results show that the proposed algorithm in this paper has better performances and segmentation results compared with the algorithms based on fuzzy entropy or fuzzy clustering.
作者
颜明重
黄冰逸
朱大奇
YAN Mingzhong;HUANG Bingyi;ZHU Daqi(Underwater Vehicle and Intelligent System Laboratory, Shanghai Maritime University, Shanghai 201306, China)
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2020年第9期1268-1273,共6页
Journal of Harbin Engineering University
基金
国家自然科学基金项目(51575336)
上海市自然科学基金项目(17ZR1412400).
关键词
图像分割
灰度波动
水下图像
中值滤波
非均匀光照
模糊熵
模糊聚类
主元分析
image segmentation
grayscale wave
underwater image
median filter
non-uniform illumination
fuzzy entropy
fuzzy clustering
principal component analysis
