使用像素质量函数和边界约束实现图像去雾

Image dehazing using pixel quality function and boundary constraint

针对优化算法估算传输率复杂度高和去雾图像亮度不足的问题,提出一种基于像素质量函数和边界约束估算传输率的新算法。首先根据像素去雾度、像素信噪比构建像素质量函数,通过求解该函数最大值获取最优传输率,再结合复原图像灰度边界约束形成粗糙传输率;然后运用导向滤波获取精细传输率;最后使用大气散射模型对雾图以及去雾图像分别进行去雾和亮度增强处理。视觉效果显示,提出的算法复原的图像明亮自然;客观数据表明,算法处理结果的平均梯度比、信噪比和运行时间等指标值大幅提升。相对于其他测试算法,其平均梯度比均值至少提升1.34倍,信噪比均值至少提升1.07倍,运行速度均值至少加速1.28倍。该算法有效提高了雾图清晰度和明亮度,并且具有较好的实时性。

Aiming at the high computational complexity of optimization algorithm for estimating transmission and the problem of dim brightness of dehazed image,this paper proposed a new algorithm based on pixel quality function and boundary constraint to estimate the transmission.Firstly,the algorithm estimated optimal transmission by maximizing pixel quality function which included pixel hazy removal rate and pixel signal-to-noise ratio indicators.The boundary constraint of scene radiance modified the optimal transmission to produce rough transmission.The guided filter smooth the rough transmission to obtain fine transmission.Finally,the algorithm improved the visibility of hazy image and brightness of dehazed image using atmospheric scattering model.The visual effect shows that the image became more natural and bright after two-stage dehazing.The objective data shows that the average gradient ratio,signal-to-noise ratio and running time of the proposed algorithm are greatly improved.The average value of the average gradient ratio of proposed algorithm is at least 1.34 times higher than that of benchmark algorithms,the signal-to-noise ratio is at least 1.07 times higher,and the running speed is at least 1.28 times faster.Therefore,the proposed algorithm is effective in enhancing the clarity and brightness of hazy images,and is suitable for real-time projects.