Research On Single Image Defogging Based On Physical Imaging Model

Image defogging technology,an important part of the field of image restoration,has important practical significance in the field of automatic driving,which draws much attention of many scholars.Many scholars are attracted and do a lot of research,and the new relate research results keep coming up.This thesis mainly focuses on the single image defogging method based on image restoration.This method combines the prior knowledge of the image and some characteristics of the image with the imaging model to estimate the fog-free image;although this type of method shows a good defogging effect,it still has the problems of unsatisfactory edge-preserving effect and long time-consuming.This thesis starts with the restoration of the visibility model of a single image,and tries to seek an algorithm for faster defogging and a more complete image edge preservation effect after defogging.The main tasks of this article are as follows:1)In order to optimize the problem of the unsatisfactory edge-preserving effect of the single image visibility model of the rapid restoration,a defogging algorithm based on propagation filtering is proposed.Firstly,the atmospheric light value can be estimated due to the characteristic that the fog is white after reaching image white balance.Secondly,in order to solve the problem of the unsatisfactory edge-preserving effect of the single image visibility model of the rapid restoration,the author uses propagated filtering to estimate the atmospheric light component to achieve defogging.The comparison of simulation experiments shows that the edge-preserving effect of this method is better than the initial method and the speed is faster.2)In order to solve the problem that the strong edge defogging is not ideal for the single image visibility model of the rapid restoration,an improved defogging algorithm based on single image visibility of the restoration is proposed.Firstly,under the condition of image white balance,atmospheric light composition is estimated by the characteristics of fog.Secondly,aiming at the problem of undesirable strong edge defogging in the restoration of the single image visibility model,the author proposes to separate the transmittance from the atmospheric light component and then optimize it.This makes the edge information of the transmittance map as consistent with the original picture as possible.The fog-free image obtained is dark at this time.Finally,it is necessary to perform gamma transformation on the image to improve the image brightness.Simulation experiments show that the method has better defogging effecton strong edges and faster running speed.