Unstructured Road Image Enhancement Method in Amblyopia Environment Based on Improved Dark Channel Prior*
Pengyu Xue, Yang Chen, Yuejun Cheng, Ping Feng, Hongliang Wang, Dawei Pi · 2024
Perception and recognition of roadable areas is an important part of intelligent driving. Most of the existing researches on road perception methods are limited to structured roads. However, in the complex field environment, the difficulty of identifying roadable area surges, so it is necessary to develop a road perception method that adapts to the field environment. Based on existing sensor equipment, through the information fusion of multi-sensor data, driving area in free space can be detected and identified, and the driving efficiency and safety of vehicles in amblyopia environment can be improved. This paper takes intelligent vehicle as the object to carry out the research of environmental perception enhancement. An image enhancement algorithm for amblyopia environment was designed, and the transmission accuracy fitting experiment was carried out by improving the dark channel prior theory. The fitting results were applied to the smoke image to be processed by combining haze image degradation model, and the smoke image reconstruction was carried out to ensure the high definition and fidelity of the image in the amblyopia environment.