布谷鸟搜索的非均匀弱光图像自适应增强算法

CS-based Adaptive Enhancement Algorithm for Non-Uniform Low Light Image

为了解决现有图像增强算法容易产生光晕伪影以及细节丢失的问题,文中针对非均匀弱光条件下的图像提出一种新的增强算法。通过色彩空间变换提取原始彩色图像的亮度Y分量,采用引导滤波增强图像细节;对细节增强后的亮度图像采用改进的Sigmoid函数(S型曲线)进行自适应线性增强,其中改进的Sigmoid函数最佳参数的获取是以布谷鸟搜索算法寻得的使图像对比度最大时的参数作为最佳参数;通过线性色彩转化得到增强后的彩色图像。实验结果表明,该算法对非均匀弱光图像有明显的增强效果,在峰值信噪比、结构相似度、对比度质量指数及绝对平均亮度误差四个评价指标中优于其他增强算法,各指标平均提高约41.6%、26.0%、61.3%、63.3%,可有效地改善非均匀弱光图像在增强过程中所存在的增强不足以及光晕伪影现象。

In order to solve the problems that the existing image enhancement algorithms tend to produce halo artifacts and detail loss, this paper presents a new image enhancement algorithm.The brightness component Y of the original color image is extracted by color space transformation, and the details of the image are enhanced by guided filtering.Then, the brightness image with enhanced details is adaptively and linearly enhanced by the improved sigmoid function(S-shaped curve).The parameters found by cuckoo search to maximize the contrast of the image are taken as the best parameters of the improved Sigmoid function.Finally, the enhanced color image is obtained by linear color transformation.The experimental results show that this algorithm has obvious enhancement effect on non-uniform low-light images, and it is superior to other enhancement algorithms in four evaluation indexes: PSNR,SSIM,PCQI and AMBE,which are improved by about 41.6%,26.0%,61.3% and 63.3% on average.It can effectively enhance non-uniform low-light images, improving the lack of enhancement and halo artifacts.