基于视觉感受野的夜间彩色图像自适应增强算法

针对夜间环境造成的图像对比度下降和增强过程中带来的噪声、过曝和光晕效应等问题,提出了一种基于视觉感受野的夜间彩色图像自适应增强算法。首先将图像转换到HSV空间,鉴于视觉感知的对比度敏感特性,利用中心兴奋区及外周抑制区组成的拮抗式同心圆双高斯差(DOG)感受野模型对图像亮度空间V进行对比度自适应调节,并对模型参数σ1、σ2和局部对比度都进行量化水平为4的均匀量化,局部对比度较高的量化区间采用较小模型参数,局部对比度较低的量化区间采用较大的模型参数,模型参数A2与局部对比度成反比进行微调,最后将图像转换到回RGB空间。通过与流行算法MSRCR、NPE及LIME的对比实验表明,该算法获得了较好的主观视觉效果,且在对比度、亮度提高的同时避免了暗区域噪声过大,并更好地增强了轮廓信息。

Adaptive night color image enhancement based on visual receptive fields

Aiming at problems such as contrast reduction,noise,overexposure and halo effect of image contract deduction and enhancement under nighttime environment,an adaptive night color image enhancement based on visual receptive fields is proposed.Firstly,the algorithm converts the image into HSV space.Due to the contrast-sensitive characteristics of visual perception,contrast adaptive adjustment of the image brightness space V is carried out by antagonistic concentric circular double Gaussian differential(DOG)receptive fields model composed of the central facilitation zone and the peripheral zone of suppression.The model parameter and local contrast are uniformly quantized with a quantization level of 4.The quantization zone with high local contrast needs smaller model parameters to suppress,and the model parameter is inversely proportional to the local contrast for fine-tuning.Finally,the image is converted into RGB space.Compared with popular algorithms MSRCR,NPE and LIME,the results show that the contrast and brightness are enhanced while avoiding excessive dark area noise,which has better enhanced contour information.