Parking Space Recognition Method Based On Feature QR Code In The Scene Of Autonomous Valet Parking

Autonomous valet parking is an unmanned driving technology that can realize autonomous driving and automatic parking in a certain area.It can effectively alleviate the problem of parking difficulties and improve the utilization rate of parking spaces and parking efficiency of drivers.The existing autonomous valet parking system is mainly realized through three methods: infrastructure-based,vehicle-based,and infrastructure-based combined with vehicle-based.As a key technology,parking space recognition is crucial to the operation of the autonomous valet parking system.Relying on the infrastructure-based technology to realize the positioning and identification of the allocated parking spaces has a high cost for the renovation of the parking lot,and it is difficult to promote on a large scale.Relying on vehicle-based technology to realize the active recognition and confirmation of the parking space,its effect is too dependent on the reliability of the sensor,the computing power of the hardware,and the degree of intelligence of the algorithm.Through in-depth research,this article finds that the root cause of the problem lies in the lack of means and methods for direct information interaction between parking spaces and vehicles.Aiming at the problem of low reliability of parking space recognition and confirmation in the scene of autonomous valet parking,this paper proposes a parking space recognition and confirmation method based on feature QR Code.Firstly,in view of the lack of direct information interaction between parking spaces and vehicles,the features of typical parking spaces in the scene of autonomous valet parking are analyzed and their features are classified.The QR Code technology is introduced to graphically characterize the extracted parking space features,and then study the encoding and decoding methods of the parking space feature QR Code.A graphical description method of parking space feature information based on QR Code is proposed.Secondly,based on the constructed parking space feature QR Code,combined with the physical structure of the parking space and the installation position of the onboard camera,the present method of the parking space feature QR Code in the parking space is studied.The feature QR Code function is analyzed.The collection effect of objects at different positions on the ground under the field of view of the vehicle camera is compared and studied.And a matching method of feature two-dimensional code and parking space is proposed.According to the characteristics of the vehicle-mounted fisheye camera,this paper introduces the camera imaging principle,fisheye distortion model and camera calibration method for the image acquisition of feature QR Code.The image processing technology of feature QR Code is studied by combining with the characteristics of parking space scene.Then,in view of the typical problems of occupied parking spaces and defaced feature QR Code,the parking space recognition method based on feature QR Code is studied.The main recognition methods of the current parking spaces are analyzed,and then a parking space recognition method based on the feature QR Code is proposed.The processed image is analyzed,the identification method of occupied parking spaces is put forward.Based on the template matching algorithm,the similarity calculation formula based on the QR Code is designed,and the enhanced recognition method of the defaced QR Code is proposed.By combining the parking space feature information obtained by the decoding algorithm and the positioning information of the QR Code obtained by the positioning algorithm,and further combining the relative position relationship between the feature QR Code and the parking space,the position information of the parking space is derived,and the accurate confirmation and recognition of the parking space is realized.Finally,the experimental system architecture is designed,and typical experimental scenarios are built to carry out parking space recognition experiments in different driving conditions,such as the recognition of occupied parking spaces,the enhanced recognition of defaced feature QR Code,and standard parking spaces.The experimental results show that the method proposed in this paper can realize the confirmation and identification of empty parking spaces even when the lighting conditions are complex and the adjacent vehicles partially occupy the parking spaces.The direct information interaction between the vehicles and the parking spaces is realized,and the recognition efficiency and reliability of the parking space in the scene of autonomous valet parking are improved.