Real-Time Color Fidelity Dehazing Algorithm for Power Grid Surveillance Images

Aiming at the problems of sky bias, pseudo contour, dark result and lack of real-time performance of the traditional dark-channel dehazing algorithm for Power Grid Surveillance Images in haze days, we propose an algorithm for color fidelity dehazing of power grid surveillance images. By improving the atmospheric light value calculation method to achieve the purpose of brightening, by amending the physical model of the transmission map to achieve the engineering purpose of color fidelity, through the corrosion operation combined with the upper and lower sampling to achieve real-time processing effect to meet the engineering needs. Experiments show that the color fidelity dehazing algorithm for power grid Surveillance images eliminates off-color and pseudo-contours, the average increase in variance is 667.77, the information entropy increases by 0.885, the percentage of saturated pixels decreases by 1.334%, and the processing time decreases from 4 seconds to 0.5 seconds, which meets real-time requirements.