基于暗通道先验的海边雾天图像去雾算法研究

发布时间：2018-03-20 22:48

本文选题：暗通道先验　切入点：K-means聚类　出处：《青岛大学》2017年硕士论文　论文类型：学位论文

【摘要】：雾天海边环境中,大气中的悬浮粒子中水滴含量较高,大气光的散射作用更强,雾的浓度更大,获取的图像质量退化,影响了计算机对图像的分析、识别与处理。由此可见,对雾天图像进行有效的去雾处理,增加图像的清晰度,提高图像的质量,有着现实和理论的迫切需要。目前的图像去雾方法的研究,大多是基于暗通道先验算法理论,图像去雾效果比较显著。但是,这种方法仍然存在天空色彩失真、图像处理速度慢的缺点。本文针对这些不足,结合海边雾天图像的特征,提出一种基于暗通道先验的海边雾天图像去雾算法。针对暗通道算法在处理天空区域时色彩失真的现象,本算法对大气光强度的估计方法进行了改进。对含天空区域较多的海边雾天图像,运用改进的K-means聚类方法分割原始图像,得到保留比较完整图像信息的天空区域。整幅图像的大气光强度估计值是由对分割的图像分别估计天空区域和非天空区域的大气光强度值并进行加权处理得到。针对软抠图算法用时过长问题,本算法采用导向滤波器对透射率进行优化,不但缩短时间,还能实现边缘平滑、增强细节,有效消除方块效应,得到相对准确的透射率估计值。由于海边雾天图像较暗,色彩不明显,图像对比度低,对于复原的图像进行色调重映射,提高了图像的对比度。本文提出的基于暗通道先验的海边雾天图像去雾算法借助MATLAB实验平台进行仿真,并与中值滤波和双边滤波两种算法进行结果比对,实验证明本算法能有效去除雾天对海边图像的影响,使图像更自然明亮,对比度更高,可视性更好,实现雾天图像的快速、有效处理。
[Abstract]:In the foggy seaside environment, the droplets in the suspended particles in the atmosphere are higher, the scattering of atmospheric light is stronger, the concentration of fog is greater, the quality of the obtained images is degraded, and the analysis, recognition and processing of the images are affected by the computer. There is a realistic and theoretical urgent need for effective defog processing to increase the sharpness and improve the quality of fog images. Most of the current researches on image de-fogging methods are based on the theory of dark channel priori algorithm. The effect of image de-fogging is remarkable. However, this method still has the disadvantages of color distortion of sky and slow image processing. A priori defog algorithm for seaside fog images based on dark channel is proposed. The color distortion of dark channel algorithm in dealing with the sky region is discussed. The algorithm improves the estimation method of atmospheric light intensity. The improved K-means clustering method is used to segment the original images for seaside fog images with more sky regions. The atmospheric light intensity estimation of the whole image is obtained by estimating the atmospheric light intensity of the segmented image in the sky region and the non-sky region, respectively. Aiming at the problem that soft matting algorithm takes too long time, In this algorithm, the transmissivity is optimized by using a guide filter, which can not only shorten the time, but also realize the edge smoothing, enhance the detail, eliminate the square effect effectively, and get the relatively accurate transmittance estimation value. The color is not obvious and the contrast of the image is low, so the reconstructed image is remapped to improve the contrast of the image. In this paper, the image de-fogging algorithm based on the dark channel priori is simulated with the help of the MATLAB experimental platform. The experimental results show that the algorithm can effectively remove the influence of fog on the seaside image, make the image more natural and bright, the contrast is higher, the visibility is better, and the image of fog can be quickly realized, and the results are compared with the results of median filter and bilateral filtering, and the experimental results show that the proposed algorithm can effectively remove the influence of fog on the seaside image. Effective handling.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】