Enhancement For Fog-degraded Images

Image haze removal is aiming at the problem of lower contrast and the thinnercolor characteristics of fog-degraded images captured by outdoor surveillance system.From the perspective of image enhancement and image restoration, this paper studiesthe methods of foggy image enhancement in order to improve the image contrast andobject definition. The study on recovering a haze image has a great significance inmartial defense, target recognition, traffic navigation, remote sensing.Firstly, conventional haze remove methods from the perspective of imageenhancement, including histogram equalization, partially overlapped sub-blockhistogram equalization and the Retinex algorithm, are analyzed. Dark channel priormethod from the idea of image restoration is used to get the depths of the scene from asingle input image to remove haze. Comparing the experimental results, advantagesand disadvantages of all the algorithms are discussed. Secondly, in order to improvecolor distortion from the sky region processed by dark channel prior method, refinetransmission map method is proposed, in which added histogram equalization canadjust the bright region transmission map to enhance the brightness and contrast ofrestored image. Simulation results show that the method is effective in fog-degradedimages. Finally, comprehensive evaluation, including objective evaluation criteria andexecution time, is introduced to analyze the haze remove methods. The results showthat the proposed method is useful in enhancing fog-degraded image clearness.