低质遥感图像压缩域细节特征高效增强仿真

Low-Quality Remote Sensing Image Compression Domain Detail Features Efficient Simulation

针对现行压缩域图像增强方法,在提高图像对比度时,存在难以有效增强压缩域细节及保持图像整体亮度、清晰度的局限性,提出基于Retinex理论的低质遥感图像压缩域细节特征高效增强方法。将Retinex理论引入到图像离散余弦变换域,将离散余弦变换系数划分为照度分量和反射分量。以构造的强光抑制函数对照度分量进行调整,压缩遥感图像动态区间;设定新的光谱含量比对反射分量进行调整,增强图像局部对比度。并设计一种平滑滤波器抑制块状效应,实现对压缩域细节特征的增强。实验结果表明,与现行压缩域图像增强方法相比,所提方法具有更好的细节特征增强及整体亮度保持效果,有效地抑制了块状效应,提高了图像清晰度。

A method to effectively enhance detail features of compressed domain in low-quality remote sensing im- ages based on Retinex theory was proposed. Retinex theory was introduced into the discrete cosine transform domain of image, and the discrete cosine transform coefficients were divided into illumination component and reflecting com- ponent. Then, the constructed strong light restraining function was used to adjust the illumination component and compress the dynamic range of remote sensing image. Meanwhile, compared with the reflection component, a new spectral content was adjusted to enhance the local contrast of image. Finally, a smoothing filter was designed to sup- press blocking effect. Thus, the detail characteristic of compression domain was enhanced. Simulation results show that, compared with existing enhancement methods of compression domain image, the proposed method has better effect for enhancing detail feature and keeping overall brightness, which effectively suppresses blocking effect and im- proves image resolution.