Research Of Enhancement And Recovery Algrithm Of Fog Degraded Image

As atmospheric scattering of the fog, the contrast and color of the image taken infog conditions will be degraded, so the processing of the fog images is of greatsignificance. Current fog image processing include image enhancement and imagerestoration these two directions; image enhancement based on the perspective ofdigital images and enhance image contrast in order to achieve a clear picture of theeffect; image restoration based on atmospheric physics-based models from thephysical point of view of atmospheric scattering model to recover clear fog-freeimage. This paper analysis image defog processing methods widely used at home andabroad, and make improvement based on the of existing algorithms to get better suitedfor the poor quality image with low contrast which is the degraded by fog.Several common image enhancement processing method for fog image are:histogram equalization, Retinex and wavelet transform.Retinex method has obviousadvantages in the enhancement of color images,but it can not play its strengths indealing with grayscale images; wavelet transform is of the high computationalcomplexity, and its running time and availability is not satisfactory;In contrast, thehistogram equalization algorithm computational has a relativity low complexity andhigh practicability, and its enhancement effect is obvious toward gray-scale image.Therefore, after anglicising a variety histogram equalization algorithms derived fromtraditional histogram equalization algorithm and considering the characteristics ofhigh variety of scene depth information, this paper proposed histogram equalizationalgorithm based on the separation of the sky area, and establishment a new iterativemethod based on normal distribution to fit a normal distribution to extract the skyregion segmentation threshold.Currently the basic idea of the main image restoration processing method for fogimage is to establish a atmospheric scattering model of the atmosphere based on the physical model of atmospheric scattering model and use the model and existingrestoration algorithm for image restoration fog. Based on the different roles of highfrequency and low frequency components in image and combined with thecharacteristics of most foggy part of fog image occupies most of low-frequencycomponents, this paper treat high and low frequency parts of th image respectively,combine La PlataSri Lanka sharpening and color image restoration together. With thehelp of this combination of image enhancement and restoration algorithms, the colorimage have not only attained improvement in contrast but also maintain good colorsaturation and visual effects.In order to prove an effectiveness of the algorithm proposed in this paper, weshould run each algorithm in Matlab environment after illustrating the algorithmtheory and procedure. We process the real shot fog grayscale images and color images,and estimate every processed image by comparing their contrast, signal to noise ratioand information entropy in order to verify the algorithm validity and practicability.Through the testify of the experiment, the the algorithm proposed in this paper canlargely improve the contrast, SNR and of images and sustain the color saturation inthe meanwhile to pretend color distortion, which make the quality and visual affectionboth get advanced.