基于归一化标量权重映射与融合金字塔的彩图对比度增强算法

Image contrast enhancement algorithm based on normalized scalar weight map coupled fusion pyramid

为了解决当前对比度增强算法在图像纹理或高信号活动区域易出现颜色改变以及过度增强,且难以兼顾图像全局内容增强与局部细节增强等不足,提出了基于归一化标量权重映射与融合金字塔的图像对比度增强算法。基于2D方向偏导,定义对比度与亮度的度量模型,并以此构造标量权重映射模型;引入拉普拉斯金字塔分解机制得到图像的层次结构,再计算其权重映射的高斯金字塔;定义这两个金字塔的融合规则,得到融合金字塔;在图像度量模型与权重映射模型的引导下,利用融合金字塔完成图像重构。实验结果表明,与当前对比度增强算法相比,该算法的对比度增强质量最佳,失真度较小,没有影响颜色平衡,消除了过渡增强与人工饱和度的引入。该算法能够较好地增强彩图对比度。

In order to solve these defects such as color change and over enhancement in the image texture or high signal activity area,as well as can not be both global content enhancement and local details enhancement by the current contrast enhancement algorithm,this paper proposed the image contrast enhancement algorithm based on normalized scalar weight map coupled fusion pyramid. It used the 2D direction of partial derivative,and defined the contrast and brightness metric model; and constructed the scalar weight map by these two models. Then it introduced the Laplacian pyramid decomposition mechanism to obtain the hierarchical structure of image; and embedded Gauss pyramid decomposition to produce the Gaussian pyramid of weight mapping. By defining fusion rules of the two pyramids,it got the fusion pyramid. Finally,by the fusion pyramid under the guide of image measurement and weight mapping model,it finished the image reconstruction. Experiments results show that the contrast enhancement quality of this algorithm is best to eliminate the introduction of transition enhancement and artificial saturation,as well as not affect the color balance comparison with current contrast enhancement algorithm.