Night vision systems get more and more attention in the field of automotive active safety field. In this area, a number of\nresearchers have proposed far-infrared sensor based night-time vehicle detection algorithm.However, existing algorithms have low\nperformance in some indicators such as the detection rate and processing time. To solve this problem, we propose a far-infrared\nimage vehicle detection algorithm based on visual saliency and deep learning. Firstly,most of the non vehicle pixels will be removed\nwith visual saliency computation. Then, vehicle candidate will be generated by using prior information such as camera parameters\nand vehicle size. Finally, classifier trained with deep belief networks will be applied to verify the candidates generated in last step.\nThe proposed algorithm is tested in around 6000 images and achieves detection rate of 92.3% and processing time of 25Hz which\nis better than existing methods.
Loading....