Abstract: The management of public security protection in forest areas has always been a problem that people lack attention to. It reveals the weakness of public security protection management in forest areas that committing crimes, illegal felling of trees and poaching animals. Our group has been committed to the application of machine vision in forest areas. In order to enhance the ability of forest area to prevent illegal cutting, this paper presents a novel algorithm of license plate detection which combines image defogging processing. This method uses the technique of contrast limited adaptive histogram equalization to remove the interference of image blurring in foggy, rainy and snowy days, and combines the features of the license plate such as color, aspect ratio, etc. to detect the license plate in fog, rain and snow weather. Compared with the ordinary license plate detection method, the result of this method has been greatly improved. The accuracy of detecting license plate in fog, rain and snow weather can reach 88.00%, while the accuracy of ordinary license plate detection can only reach 82.67%. This paper preliminarily proves the feasibility of the proposed algorithm in fog, rain and snow weather of forest areas.