摘要
针对航拍图像大场景下通用图像去雾算法出现去雾效果差、细节丢失等问题,提出一种基于区域方差保边的航拍图像去雾算法。首先,依据区域得分法估计大气光值,并采用平滑-锐化滤波方法融合线性透射率与块透射率以求取准确透射率,再求解大气散射模型,获得初始去雾图像。其次,在区域方差调整中引入透射率,有效保留图像细节中的远近特性。最后,将两个模型的输出进行线性综合,得到锐化后的图像。实验结果表明,该算法主观上在去雾效果和细节保留方面都有较好的效果,客观上在信息熵、平均梯度、对比度和自然度上分别优化了1.9、4.6、14.4和3.4个百分点。尤其在实际图像目标检测应用中,目标检测的精度和数量都有提升。
In response to the issues of poor dehazing effects and detail loss in general image dehazing algorithms for large-scale aerial images,a dehazing algorithm for aerial images based on regional variance edge preservation is proposed.Firstly,the atmospheric light value is estimated using the regional score method,and a smoothing-sharpening filter method is used to fuse the linear transmittance and block transmittance to obtain accurate transmittance,which is then used to solve the atmospheric scattering model and obtain the initial dehazed image.Secondly,the transmittance is introduced into the regional variance adjustment to effectively preserve the far and near characteristics in the details of the image.Finally,the outputs of the two models are linearly combined to obtain a sharpened image.The test results show that entropy can be increased by 1.9%,the average gradient can be improve 4.6%,the contrast can be increased by 14.4% and natural image quality evaluator can be decreased by 3.4%.Especially in the application of image object detection,both the accuracy and quantity of object detection have been improved.
作者
黄绿娥
肖文祥
鄢化彪
邓亚峰
HUANG Lve;XIAO Wenxiang;YAN Huabiao;DENG Yafeng(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou,Jiangxi 341000,China;School of Science,Jiangxi University of Science and Technology,Ganzhou,Jiangxi 341000,China)
出处
《遥感信息》
北大核心
2025年第2期52-59,共8页
Remote Sensing Information
基金
国家自然科学基金(62363013)
江西省自然科学基金(20224BAB202036)
江西省教育厅科学技术重点研究项目(GJJ2200805)。
关键词
去雾
大气散射模型
大气光值
透射率
区域方差
dehazing
atmospheric scattering model
atmospheric light value
transmittance
regional variance
