Automatic parking space recognition panoramic look-around image real-time splicing method based on look-around

Abstract

The invention discloses an automatic parking space recognition panoramic look-around image real-time splicing method based on look-around. The method comprises the following steps: 1) acquiring imagedata through a fisheye camera; 2) performing distortion correction on the image to obtain a fisheye image after distortion correction; 3) performing inverse perspective transformation on the image after distortion correction to obtain four aerial views of the front, rear, left and right of the automobile body; 4) splicing the obtained aerial views to obtain a 360-degree panoramic look-around imagearound the automobile body; 5) optimizing an image splicing seam by adopting a fusion algorithm to obtain a 360-degree panoramic look-around image around the vehicle body; and 6) inputting the obtained 360-degree panoramic look-around image around the automobile body into a deep learning target detection model for automatic parking space identification and drivable area identified.The method provided by the invention has the advantages of small calculated amount, high real-time performance, high precision, good effect and easiness in implementation.

Description

technical field [0001] The invention relates to the technical field of automatic parking of automobiles, in particular to a real-time mosaic method for panoramic surround-view image recognition of automatic parking spaces based on surround-view. Background technique [0002] The rapid development of deep learning technology has produced a more intelligent automatic parking system. In terms of automatic parking environment perception, most of the current research and products use ultrasonic radar for perception. The environmental information obtained by this distance sensor-based perception scheme is very limited. Applying visual sensors to automatic parking enables 360-degree panoramic view technology. Has been developed rapidly. The system uses 4-way fisheye cameras installed on the front, rear, left and right of the car body to obtain real-time environmental information around the car body, and realizes zero dead-angle detection of the surrounding environment of the car b...

AI Technical Summary

Problems solved by technology

[0004] In the automatic parking system based on deep learning, the use of surround-view vision to identify parking spaces and drivable areas has high real-time requirements, and it is necessary to generate a 360-degree panoramic bird's-eye view around the car body in real time. None of the above methods can meet the real-time requirements. , cannot be directly applied to the automatic parking system based on deep learning

Images