光学遥感影像DCM-HTM去雾算法研究

针对卫星遥感影像的雾区检测和校正问题,提出了一种暗通道图像DCM与雾厚度图像HTM结合的DCM-HTM遥感影像去雾算法。算法基于暗通道原理,采用mean-shift均值漂移滤波和直方图自动阈值来检测雾区域;同时通过不重叠窗口获取HTM;最后在检测出的雾区内实现去雾校正处理。实验结果表明,算法能较准确地检测出影像上的雾区域,能在影像正常区域不受影响的前提下,有针对性地对受雾影响区域进行校正,校正效果较整幅影像的去雾校正有较大改善。

Optical Remote Sensing Image Defogging Algorithm Based on DCM-HTM

Focusing on fog detection and correction of satellite remote sensing image, a defogging algorithm DCM-HTM for remote sensing image is proposed by the combination of dark channel map ( DCM ) and haze thickness map( HTM) . Based on the principle of dark channel, mean-shift filter and automatic histogram threshold are used to detect fog area, at the same time, HTM is obtained by nonoverlapping window. At last, the correction of fog is produced at the fog area. The results show that the algorithm can accurately detect the fog area, and the areas af-fected by the haze can be effectively corrected in the premise of image normal areas not affected. The results are improved greatly compared with the fog correction of the whole image .