Research And Application Of Outdoor Empty Parking Space Detection Algorithm

With the increase in car ownership year by year,parking difficulties has become a problem hindering urban development.How to quickly obtain parking status information in cities so that car owners can more easily find empty parking spaces to park has become an important step in the smart parking system.This paper uses computer image processing technology to detect the state of parking spaces in two scenarios,road parking spaces and outdoor parking lots,to obtain the free state and location of the parking spaces,and complete the task of collecting front-end parking state information of the smart parking system.The main research contents of this paper are as follows:(1)A road empty space detection algorithm based on improved Faster RCNN is proposed.To solve the problem of road empty parking space detection,first,use K-means clustering to cluster the original data candidates;then,add a convolutional layer to the Faster RCNN regional proposal network to extract data features;then use Soft NMS(Soft Non-Maximum Suppression)the algorithm filters the candidate frames and obtains the final detection result.A road parking space data set was established using road video surveillance images,and the proposed road empty parking space detection algorithm is tested on this data set.The accuracy of the proposed algorithm is improved by 2% compared with the original algorithm.(2)An algorithm for detecting empty spaces in outdoor parking lots based on improved SSD(Single Shot Multi Box Detector)is proposed.Aiming at the problem of vacant parking space detection in outdoor parking lots,first,the residual network ResNet50 is used as the backbone network for feature extraction;then,the extracted features are input into the classifier,and the non-maximum suppression algorithm is used to eliminate the detection targets with small probability.The improved SSD outdoor parking lot empty space detection algorithm is experimentally verified on the internationally public PKlot data set.Experimental results show that the improved SSD algorithm has an accuracy of 96% and 98% in rainy and sunny weather,respectively.It has good generalization ability and robustness.