Abstract:
Parking guidance system, as one of the main ways to improve urban traffic and improve people’s travel efficiency, has become a hot research topic at present. Sufficient i...Show MoreMetadata
Abstract:
Parking guidance system, as one of the main ways to improve urban traffic and improve people’s travel efficiency, has become a hot research topic at present. Sufficient information on the number of empty parking spaces has become a necessary condition for relevant research, which is mainly obtained by sensors at the entrance and exit of the car park. However, because of the economic cost, installation and construction, it is impossible to deploy the sensor in every parking lot in the city, so there will be a large area of parking data missing in the city.This paper presents a new technology to repair the parking lot data with DCGAN(Deep Convolutional Generative Adversarial Networks). First, the parking lot is classified by the high-dimensional clustering in the geographical space and the parking data of the existing parking lot is mapped to the empty car rate curve image. The same kind of car parking rate image has been set up in DCGAN to generate the empty car rate image of this type of parking lot. Finally, the 2-D empty car rate image is mapped to one-dimensional time series. Experiments show that DCGAN has achieved good results in data repairing of parking lot.
Date of Conference: 11-13 August 2018
Date Added to IEEE Xplore: 26 August 2019
ISBN Information: