Fast Haze Removal of UAV Images Based on Dark Channel Prior

A fast haze removal algorithm based on dark channel prior is proposed to overcome the color distortion and inefficiency caused by the dark channel prior algorithm in the recovery of UAV images. The quad-tree subdivision of higher efficiency is used for solving the atmospheric light at the first, followed by down sampling and interpolation algorithm to optimize the solution process of the transmission, and fast guided filter is used for thinning transmission. Finally, the transmission can be got by correction of tolerance mechanism. We can get the restoration images by means of the atmospheric scattering model combined with above research. Experiments show that the algorithm can effectively improve the color restoration and distortion in the sky region image, and for the UAV images without the sky area, the dehazing result is also effective; at the same time, the running speed of the algorithm is greatly improved, which is about 34 times of the He method. It can satisfy the real-time requirement of the UAV images to dehaze.