一种基于L0范数平滑滤波的遥感影像底图匀色算法

A method for color consistency in remote sensing image mosaics based on an L0 gradient minimization filter

色彩一致化是遥感影像镶嵌中的关键步骤。一种常用算法是使用基于底图的遥感影像匀色算法,通过将底图色彩映射到目标片上,实现各目标片间色彩的统一。处理过程中,高、低频信息提取方式与信息融合方法会对处理质量产生影响,导致对比度较差,甚至细节损失等问题。本文提出一种基于L0范数平滑滤波(L0 gradient minimization filter)的遥感影像底图匀色算法,首先提取底图低频信息和目标片高频信息,然后通过改进的IHS(intensity,hue,saturation)算法将两部分信息进行融合,最后得到色彩均衡的处理结果。利用本文所提方法与常用的Wallis匀色方法进行实验对比,经主、客观评价,本文所提方法可以得到更优的匀色效果,具有一定优越性。

Color consistency is a key issue in the Mosaic of remote sensing images.One of the common methods is tone image based method.By mapping colors in tone images to target images,color consistency can be achieved.In the process,the high and low frequency information extraction and information fusion will affect the image quality,such as poor contrast and detail loss.In this paper,an L0 gradient minimization filter based method is proposed to achieve consistent color for remote sensing image mosaics.Firstly,the low frequency information of the base image and the high frequency information of the target image are achieved.Then,the improved IHS(intensity,hue,saturation)method fuses the information of the two parts to obtain a color balance.The method proposed in this paper is compared with the commonly used Wallis and Color method by an experimental study.The proposed method can obtain better color consistency and has some advantage according to both subjective and objective evaluations.