Parking Lot Occupancy Detection Using Image Overlay And Intersection Technique With Harris Corner Detector

Abstract

Parking lots are very important to cities with large number of populations especially for countries that do not have many public transports which lead to the increase in number of vehicles on the road. Locating for an empty parking lot can be time consuming and may even lead serious problems such as traffic congestion. Even though there a lot of hardware-based parking monitoring systems have been achieved success in the market, the maintenance cost is still expensive. Hence, many researchers have started to explore other alternatives such as vision-based approach. Current methodology using Harris Corner Detector for parking lot detection generates the corner data from vehicle as well as any other unwanted surrounding corner data such as partially shaded shadows which are non-vehicle data that will contribute as noise and impact the classification accuracy. By implementing image overlay with Harris Corner Detector, the strongest corner data around the vehicle region will be generated since the image overlay shape is added using the grayscale intensity at range zero which is black and thus, other unwanted objects beyond the image overlay can be excluded easily by analyzing the strongest corner data only. The datasets used in this research is the video source which consists of full day outdoor parking lot images in time-lapsed produced by Cambridge Consultants in year 2017. Image overlay method improved the existing Harris Corner Detector implementation accuracy by 19.79% and proven better accuracy compared to image thresholding method in partially shaded outdoor parking lot environment by 1.04% and hence, both can be used as a hybrid system.