Research On Contrast Enhancement Of Fog-Degraded Image

Because of the atmosphere scattering phenomenon, weather degraded imagescaptured in fog suffer from poor contrast and low visibility. This significantlyaffects the traffic systems and outdoor visual systems. Hence, it is imperative toenhance the low-contrast images degraded by fog. At present, researches on theenhancement of fog-degraded images are still under discussion. A few researchesare concentrate on the analysis of the complex physical model. Due to thecomplexity of modeling situations, many of these methods are restricted inpractical applications.In order to remove the reliance on the scene structure information, this papertake the enhancement procedure as an image-processing problem. Based on theanalysis of the mechanism of imaging in fog, this paper conducts a fullinvestigation into the enhancement of fog-degraded images. Our work includes:(1) Aiming at the characteristic of scene depth variations on fog images,adaptive histogram equalization is used to enhance the local information ofimages. Methods including local HE, POSHE, interpolating HE, generalized HEare deeply analyzed in detail, and their applications on fog images.(2) Color fog images are processed to restore contrast. Based on the huepreservation principle, images are transformed both in the RGB and CMY colorspaces;the enhancement methods in gray scale are generalized to the color space.(3) The applications of multi-scale contrast enhancement methods arediscussed. Multi-scale Retinex, wavelet multi-scale gradient, and curvelettransform methods for contrast enhancement are introduced in this paper.Theoretical analysis and experimental results demonstrate that, methods usedin this paper are effective for contrast enhancement of fog-degraded images, andcan be applied to the practical situations. This research will be a solid base for thefurther investigations on contrast enhancement of fog-degraded images.