Detection And Application Of Empty Parking Space In Outdoor Parking

In recent years,a difficult"parking difficult"problem has begun to haunt urban travel citizens.Real-time detection of available spaces in parking lots through computer vision technology and pushing drivers in need of parking are important issues and directions in the study of statistics and artificial intelligence.By detecting the empty parking space and providing it to the parking driver in time,the phenomenon of parking time and regional congestion can be alleviated to a certain extent.Parking can be divided into indoor parking and outdoor parking two types,this paper focuses on outdoor parking based on statistical pattern recognition,deep learning and computer vision technology of empty parking space detection problems.Aiming at the problem of hollow parking space detection in outdoor parking lot,this paper makes an in-depth study of two methods of technical line based on statistical pattern recognition and deep learning.Through data acquisition,model establishment,platform construction,simulation and other processes to complete the corresponding work.The main summaries are as follows:(1)Improved AdaBoost pattern Recognition method for outdoor empty space detection.Because the traditional method of detecting parking status of parking space is high maintenance cost and easy to be disturbed by external factors,it may affect the accuracy and robustness of the test results.In this paper,the empty parking space is detected according to the small outdoor sample scene.In the small sample scene,this paper presents a method for detecting outdoor empty parking spaces based on Haar-Like features and enhanced AdaBoost.In the original image,the method is first segmented by Image segmentation,after that the sample is divided into two categories of empty parking space sample and possession parking space sample,then the information of parking space status is extracted by Haar-Like feature,and finally,the identification detection is carried out by enhancing the AdaBoost model.Because the noise data that the image may exist in the enhanced AdaBoost model is detected and eliminated,the sensitivity of the AdaBoost model to the noise data is reduced,and the accuracy and robustness of the detection are improved to a certain extent.Second,An improved deep-learning method for outdoor parking space detection.(2)Improved a deep learning method for outdoor empty space detection.This part mainly studies the application of FR~2Net to large sample data scene.First,build a deep learning environment;Then,set and adjust the parameters of the Faster R-CNN target detection model;and finally,use ResNet-101 to continuously deepen the network.This method enhances the quality of the training set and makes the test results achieve better results.Through the non-restrictive test in the complex scene,the accuracy of the model to the air parking space detection is more than 98%.Intuitive display technology makes it easy for parking users to quickly and accurately find their nearest empty parking space after entering a large car park,saving time to find an empty parking space in the parking lot.For the parking lot Management department,the intelligent empty parking space detection can optimize the resource allocation of parking space,improve the utilization rate of parking lot,reduce the management cost.Intelligent parking system based on intelligent air space detection will improve the intelligent management level and automatic operation efficiency of parking lot.The method proposed in this paper can be applied to intelligent parking system to produce good social value and economic value.