论文部分内容阅读
遥感图像具有整体亮度偏暗、对比度较低和目标与背景区分不明显的特点,遥感图像增强技术对于改善图像的对比度、突出某些局部细节等起着积极的作用,图像的多尺度系统已经成功应用在图像处理中。经典的多尺度系统曲波(Curverlet)变换、轮廓波(Contourlet)变换无法将连续性与数字世界进行统一处理,而Shearlet变换是目前多尺度领域内唯一满足这一性质同时还提供对图像的最优稀疏表示的多尺度系统。本文提出一种基于Shearlet变换的遥感图像增强算法,首先将遥感图像进行Shearlet分解,然后对Shearlet变换产生的低频系数进行模糊对比增强,对各尺度各方向的高频系数进行模糊增强。实验结果表明,本文算法增强的结果,在主观上能获得很好的视觉效果,客观指标中图像熵和均值有了大幅度的提升。
Remote sensing images have the characteristics of darker overall brightness, lower contrast and less obvious distinction between target and background. Remote sensing image enhancement technology plays an active role in improving image contrast and highlighting some local details, and the image multi-scale system has been successful Used in image processing. The classical multi-scale system Curverlet transform and Contourlet transform can not unify the continuity with the digital world. The Shearlet transform is the only one in the multi-scale field that satisfies this property and also provides the best image Multi-scale system with excellent sparse representation. In this paper, a remote sensing image enhancement algorithm based on Shearlet transform is proposed. Firstly, the remote sensing image is decomposed by Shearlet, then the contrast of the low frequency coefficients generated by Shearlet transform is enhanced contrastly, and the high frequency coefficients of each scale are fuzzily enhanced. The experimental results show that the enhancement results of this algorithm can obtain good visual effects subjectively, and the image entropy and mean of objective indexes have been greatly improved.