Algorithm improvement of vision-based incomplete parking space recognition in complex scenes
In order to enhance the accuracy of incomplete parking space recognition in complex scenes,an im-proved vision-based parking space recognition algorithm was proposed.This recognition algorithm adopts a con-trast-limited adaptive histogram equalization method to enhance image details.By setting regions of interest and using the global threshold segmentation method,the parking spaces and backgrounds are distinguished,and the impact of ambient noise around parking spaces on parking recognition is reduced.Morphology and connec-ted component analysis are utilized to eliminate other small noises and residual noises on the parking space.Combined with Canny edge detection,the edge points of the parking line contour are marked,and the accurate parking line contour is drawn on the original image.Experiments were conducted on incomplete parking space recognition in different complex environments,and results show that compared with traditional parking space recognition algorithms,the proposed improved algorithm can identify incomplete parking spaces more accurately and has better adaptability and real-time performance.
complex scenesvisionparking space recognitionconnected component analysis
NETL
NSTL
万方数据
