Abstract: The vision-based automatic parking system is susceptible to uneven illumination, and when image information is complicated, it is often difficult to make an accurate judgment on parking spaces. In view of this, a parking space detection method is proposed based on background illumination removal and connected regions. Fish-eye cameras are used to capture the images of the surroundings around the vehicle, a panoramic bird's eye view is obtained by processing the captured images, which is then input into parking space recognition system. The removal of background illumination from initial image effectively solves the problem of uneven illumination and complex background. A parking space extraction method based on connected regions is designed, which simplifies the extraction of parking spaces and reduce the computation efforts in parking space recognition. The results of experiment show that the methods proposed can achieve the effective recognition and accurate location of parking spaces

Key words: parking space, binarization, connected region, Hough transform