Parking space data expansion method

The invention relates to the technical field of automotive electronics, and provides a parking space data expansion method, which comprises the following steps of: extracting a parking space line image from an acquired live-action parking space image to generate a foreground database based on a scene requirement of a deep learning algorithm in an image recognition technology, and extracting parking space background images except parking space lines to generate a background database; through separation and combination of the parking space lines and the background images, the number of simulated live-action parking spaces can be effectively increased. A reference parking space line is generated by acquiring parking space line features of any parking space line image, a target parking space set is generated according to the reference parking space line and a preset distribution rule, and the number of simulated live-action parking spaces is further expanded by forming parking space images with different preset d