几种典型盲复原算法对光学遥感图像质量提升效果对比分析

Comparison and analysis of several typical blind restoration algorithms for improving the quality of optical remote sensing images

遥感卫星搭载的成像传感器由于受到姿态颤振、传感器自身物理性质等因素的影响,导致遥感图像质量下降,其中辐射质量方面主要表现为图像模糊。本文针对遥感图像模糊的问题,分别采用了基于强度和梯度的L0正则化先验、学习迭代自适应先验、局部最小强度先验的图像盲复原等方法,对遥感图像进行去模糊处理,对比分析了3种算法在遥感图像辐射质量提升上的复原效果。实验结果表明:3种盲复原方法均能在一定程度上较好地提升遥感图像的辐射质量,复原后的遥感图像更加清晰。其中,基于局部最小强度先验的盲复原结果复原细节更多、噪声更少。本文结果可为遥感图像质量提升处理的算法选择提供参考。

The imaging sensor carried by remote sensing satellite is affected by attitude jitter,sensor’s physical properties and other factors,resulting in the decline of remote sensing image quality,in which the radiation quality is mainly reflected in image blur.Aiming at the problem of remote sensing image blur,this paper adopts the image blind restoration methods of L0 regularization prior based on intensity and gradient,local minimal intensity prior and learning iteration-wise generalized shrinkage–thresholding operators to deblur the remote sensing image,and compares and analyzes the restoration effects of the above three algorithms in improving the radiation quality of remote sensing image.The experimental results show that the three blind restoration methods can improve the radiation quality of remote sensing images and the restored remote sensing images are clearer.Among them,the blind restoration results based on local minimum intensity prior have more restoration details and less noise.The results of this paper can provide a reference for the selection of algorithm to improve the quality of remote sensing image.