基于均值滤波的遥感模糊图像对比度增强方法  

Adaptive Multi⁃Threshold Image Segmentation Based on Deep Learning and Potential Function Clustering

作　　者：张艳晓 ZHANG Yanxiao(College of Electronic and Information Engineering,Xi'an Siyuan University,Xi'an 710038,China)

机构地区：[1]西安思源学院电子信息工程学院,西安710038

出　　处：《吉林大学学报(信息科学版)》2024年第6期1058-1065,共8页Journal of Jilin University(Information Science Edition)

基　　金：陕西省教育厅科研计划基金资助项目(21JK0300)。

摘　　要：为提高遥感模糊图像对比度增强效果,增加清晰度,提出基于均值滤波的遥感模糊图像对比度增强方法。首先,采用快速中值自适应均值滤波算法对遥感模糊图像整体进行去噪处理;其次,结合遥感图像边缘的分形自相似特征以及灰度梯度变化实现对图像边缘点的提取,并将图像的整体区域划分为明亮和暗淡区域;最后,采用细节保留映射算法和感知对比度映射方法分别增强两个区域的对比度,完成对遥感模糊图像整体的对比度增强,实现对该图像的色彩还原。实验结果表明,该方法增强图像能有效去噪,绝对均值差小于0.85,在图像对比度和清晰度增强方面表现出良好的性能。In order to improve the contrast enhancement effect of remote sensing blurred images and increase clarity,a method based on mean filtering for remote sensing blurred image contrast enhancement is proposed.Firstly,a fast median adaptive mean filtering algorithm is used to denoise the entire remote sensing blurred image.Secondly,combining the fractal self-similarity feature of remote sensing image edge and the change of gray scale gradient,the edge points of the image are extracted.On this basis,the whole area of the image is divided into bright areas and dim areas.Finally,the detail preserving mapping algorithm and perceptual contrast mapping method are used to enhance the contrast of the two regions,respectively,and the overall contrast of the remote sensing blurred image achieving color restoration of the image.The experimental results show that the proposed method can effectively denoise images,with an absolute mean diference of less than O.85,and exhibits good performance in enhancing image contrast and clarity.

关 键 词：均值滤波 遥感图像 对比度增强 去噪 边缘提取 色彩还原 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]