基于两次双边滤波的煤矿监控图像去雾研究

发布时间：2018-05-10 07:09

本文选题：雾尘图像复原 + 两次双边滤波　；参考：《图学学报》2017年03期

【摘要】：煤矿井下智能视频监控系统在各大煤矿得到广泛的应用,然而由于井下条件恶劣,视频监控中经常遇到伴有各种随机噪声的尘雾图像。针对降质严重、视觉效果差的监控图像,提出一种基于两次双边滤波的快速图像去雾方法(FDA-DBA)。首先利用四叉树方法获取全局环境光亮度,然后对最小通道图采用双边滤波器获取粗略的大气散射图并进一步优化;其次利用容差机制进行透射率修正,解决明亮区域处理后颜色失真问题;最后利用大气散射模型复原雾尘图像。实验表明,该算法能较准确地恢复场景的色彩和清晰度,可获得较真实的清晰无雾图像,具有较高的准确性和鲁棒性,并且算法的时间复杂度与图像像素数呈线性函数,适合于煤矿智能视频监控环境。
[Abstract]:Intelligent video surveillance system in coal mine has been widely used in various coal mines. However, due to the poor conditions in the underground, the video surveillance often meets the dust fog image with various random noises. A fast image de-fogging method based on two-stage bilateral filtering is proposed for monitoring images with severe degradation and poor visual effect. First, the global ambient luminance is obtained by quadtree method, then the rough atmospheric scattering image is obtained by using a two-sided filter for the minimum channel map, and then the transmissivity correction is carried out by using the tolerance mechanism. The problem of color distortion after bright region processing is solved, and the fog dust image is reconstructed by atmospheric scattering model. The experiments show that the algorithm can restore the color and sharpness of the scene accurately, and can obtain a clear and fog free image with high accuracy and robustness. Moreover, the time complexity of the algorithm and the number of pixels of the image are linear functions. It is suitable for coal mine intelligent video surveillance environment.
【作者单位】：辽宁工程技术大学矿业技术学院;辽宁工程技术大学矿业学院;
【基金】：国家自然科学基金项目(50774041)
【分类号】：TD76

【参考文献】