Daytime Fog Detection and Density Estimation with Entropy Minimisation

The fog disturbs the proper image processing of many outdoor observation tools. For instance, fog reduces the obstacle v isibility in vehicle driving applications. Usually, the estimation of t he amount of fog in the scene image allows to greatly improve t he image processing, and thus to better perform the observation task . One possibility is to restore the visibility of the contras ts in the image from the foggy scene image before to apply the usual image process ing. Several algorithms were proposed in the recent years for defogging. Before to apply the defogging, it is necessary to detect the p resence of fog, not to emphasis the contrasts due to noise. Su rprisingly, only a reduced number of image processing algorithms were propos ed for fog detection and characterization. Most of them are d edicated to static cameras and can not be used when the camera is moving. T he daytime fog is characterized by its extinction coefficien t, which is equivalent to the visibility distance. A visibility-met er can be used for fog detection and characterization, but th is kind of sensor performs an estimation in a relatively small volume of air, a nd is thus subject to heterogeneous fog, and air turbulence w hen the camera moves. In this paper, we propose an original algorithm, based on ent ropy minimization, to detect the fog and estimate its extinc tion coefficient by the processing of stereo pairs. This algorithm is fast, pr ovides accurate results using low cost stereo cameras senso r and, the more important, can work when the cameras are moving. The propose d algorithm is evaluated on synthetic and camera images with ground truth. Results show that the proposed method is accurate, an d, combined with a fast stereo reconstruction algorithm, sh ould provide a solution, close to real time, for fog detection and extincti on coefficient estimation for moving sensors

https://hal.archives-ouvertes.fr/file/index/docid/1068534/filename/doc00019386.pdf