Research On Haze Image Enhancement Processing Technology

In recent years,haze has occurred most frequently in northern China,especially during the winter heating period,this situation is particularly serious.Under the condition of haze,due to air pollutants caused a sharp decline in visibility,makes image collected by outdoor image acquisition system is degraded and fuzzy,resolution is decreased and a lot of the important feature information is lost,which brings great difficulties to the image information extraction,also cause restrictions and effects for object tracking and outdoor work.Therefore,it is of great value and practical significance to study the image enhancement technology of haze days.In recent years,with the rapid development of image processing technology,many ways of dealing with the smog image has been derived.Which is roughly divided into two categories: one is processed by digital image processing method.The degradation of the image is generally bright with low contrast,we do not consider the image degradation reasons,enhance the image from the perspective of the image brightness and contrast when we use image enhancement method.The other is to establish mathematical models by understanding the causes of the degradation of the haze image and to implement the treatment of the image.The traditional image enhancement method also loses the detail of the image while enhancing the image,enhances the noise signal contained in the image,and causes the information entropy to decrease.In this paper,the basic framework of the haze image enhancement algorithm is discussed from the perspective of the fog image enhancement,and the key technologies in the algorithm are studied deeply,and the shortcomings of the traditional image enhancement algorithm are improved.In the image enhancement method based on Retinex algorithm,based on the traditional multi-scale Retinex algorithm(MSR),the original basic algorithm is improved,and then the results and the algorithm rate of the proposed algorithm werecompared with the existing algorithm.In the method of image enhancement based on wavelet transform,combined with wavelet transform multi-scale,multi-resolution features and with the characteristics of contrast-constrained adaptive histogram equalization,An image enhancement algorithm based on discrete wavelet transform and contrast-constrained adaptive histogram equalization is proposed.Experimental results show that these two methods can effectively enhance the image,relatively speaking,the distortion is relatively small,to achieve the haze sky image enhancement.