基于LEP多尺度滤波分解的图像增强算法

Image Enhancement Algorithm Based on Multi-scale Decomposition of LEP Filtering

为了解决传统图像增强方法不能在多个尺度上处理图像细节的问题,采用一种基于局部边缘保持(Local Edge-Preserving,LEP)滤波器的图像多尺度分解方法对图像进行增强。利用LEP滤波器对图像进行多尺度分解,分解成一个基本层和两个细节层;根据分解后各层的特点和增强的要求,通过保留基本层,对细节层进行S形曲线增强,实现图像细节的增强;最终合成并恢复颜色,得到最终增强图像。采用主观评价和对比度增益、信息熵来评价图像增强效果。实验结果表明,该方法能够有效增强图像细节,提升对比度,避免光晕现象,所得增强图像具有良好的视觉效果。

In order to solve the problem that traditional image enhancement methods cannot manipulate the image details on multi-scale, an image multi-scale decomposition method based on the local edge-preserving filter is proposed. Firstly, as use of LEP filter for multi-scale decomposition, an image decomposes into a base layer and two detail layers. Secondly, according to the characteristics of the decomposed layers and the enhanced requirements, an S type curve is used to detail layers and keep the base layer unchanged. Finally, an enhanced image is obtained by restructuring the image and restoring the color. The subjective aspect, contrast gain and entropy are used to evaluate the efficacy of the image enhancement. The experi-mental results show that the proposed method can enhance the image details, improve the contrast ef-fectively, it can also avoid halo. The enhanced image has good visual effect.