Research On Fog Digital Image Clearness Algorithm

Haze,fog,or other large particles existing in outdoor environment can cause the obtained image quality degradation and off white,including the visibility degradation,high noise,the degraded contrast and color,blurred scenery profile and the difficulties in extracting image feature,which lead to image distortion finally and influence the visual effect.Thus it is very difficult to realize real-time application for outdoor work systems,such as monitoring,intelligent navigation,unmanned Aerial Vehicle.Furthermore,people's normal work and life are affected and technological development and industrial production are obstructed,hence it is of great significance to research image defogging.Firstly,histogram equalization defogging algorithm and Retinex defogging algorithm are studied.Based on probability theory,histogram equalization defogging algorithm uses histogram transform after converting RGB image into single channel gray image.In this method,image color is enriched and contrast is enhanced,but defogging process is uncontrollable.In view of these deficiencies,adaptive histogram equalization and limit contrast histogram equalization algorithm defogging is proposed.Retinex algorithm is color constancy image enhancement theory on the base of human vision.In order to solve the problem of the restored image color migration resulted from there being only one defogging scale parameter in the Single-scale Retinex algorithm,Multi-scale Retinex algorithm and color restoration Multi-scale Retinex algorithm is put forward.According to the results,it is detected that two methods can restore the fog image effectively,but the restored image is easy to produce some harmful effects such as Halo effect and color migration ect.Secondly,dark channel priori defogging method is researched and dark channel priori defogging algorithm based on segment is proposed.In the algorithm,image segmentation and segmented image defogging are implemented.In order to meet the self-adaptability,maximum entropy segmentation and Otsu segmentation are respectively adopted to divide sky region and non-sky region.The main process of dark channel priori is defogging sky region and non-sky region respectively with corresponding atmospheric light,transmittance and defogging factor.Experimental results show that the method improves effectively color dim in dark channel priori defogging,Halo effect and color deviation in other methods.Finally,using the brightness,contrast and clarity as objective evaluation index,the algorithms including adaptive histogram equalization algorithm,limit contrast histogram equalization algorithm,Multi-scale Retinex algorithm,Multi-scale Retinex algorithm with color restoration,dark channel priori algorithm and dark channel priori defogging algorithm based on segment,are compared and analysed by written image defogging algorithmic synthesis software.The results indicate that Halo effect and color deviation and other problems can be effectively solved by the defogging algorithm of dark channel priori based on segment.Basic research ideas of this paper is as follows:first of all,development status at home and abroad and some common methods of image defogging are introduced,then in order to solve the problems such as Halo effect and color migration,the defogging algorithm of dark channel priori based on segment is proposed combined with dark channel priori.The main content is that atmospheric light and transmittance of divided images are respectively obtained to defog image.Because manual segmentation needs artificial interaction,maximum entropy segmentation and Otsu segmentation are studied to meet the self-adaptability.Finally,objective evaluation on defogging effect is made and the entire defogging module is introduced.