关键词: 图像去雾, 人类视觉机制, 单色大气散射模型, 视网膜皮质模型, 大气传输函数

Abstract: Aiming at the problem that the existing image dehaze algorithm can not effectively enhance
degraded images in complex atmospheric environment, animage dehaze algorithm based on visual physical
model (VPM) is proposed bycombiningmonochrome atmospheric scattering model (MASM), atmospheric
transmission function (ATF) and retinex. The model can describethe image degradation under complex
atmospheric conditions such as non-uniform light degradation, haze degradation, and noise degradation.
Environmental light degradation is first obtained though variational method. And then, markov random
field is introduced to convert the problem of scene reflection rate to the maximum posterior probability.
Finally, the brightness of scene reflectance is adjustedwith adaptive histogram equalization, and this is the
final dehazed image. Experimental results show that VPM can enhance degraded images under complex
atmospheric environment and has better physical fidelity and visual pleasure compared.

Key words: image dehaze, human visual system, monochrome atmospheric scattering model, retinex;
atmospheric transmission function