Parking space state identification method and device based on video streaming

Abstract

The invention discloses a parking space state identification method and device based on video streaming. The method comprises the steps that firstly, four corner point coordinates of a parking space are acquired, and through perspective transformation, a parking space image is converted into a standard rectangular image; the standard rectangular image is input to an identification model, and the duty probability of the parking space of the frame is output; the duty probability of the parking space of the frame is compared with a preset threshold value, and whether or not a vehicle exists underthe state of the parking space of the frame is judged; finally, in combination with the state of the parking space of the current frame and preset frames before the current frame, a final state of the parking space is judged. According to the scheme, a parking lot is subjected to parking space state identification based on a camera mode, and the defects can be greatly overcome that in a traditional mode, the cost is high, maintenance is troublesome, and the device is easily damaged. According to the technical scheme, the state of the parking space is comprehensively judged through multiple frames of pictures, a small number of mistaken results caused by interference factors such as pedestrian and vehicle movement are filtered out, the anti-interference capacity is high, and the identification accuracy is high.

Description

technical field [0001] The invention relates to the technical field of image intelligent recognition, in particular to a video stream-based parking space state recognition method and device. Background technique [0002] With the continuous increase of the number of cars in our country, the shortage of parking spaces has become a major problem. At present, an effective way to alleviate the shortage of parking spaces is to monitor the status of each parking space in the parking lot, and provide support for subsequent parking space counting, parking space guidance and intelligent car search, thereby improving the utilization rate of parking spaces. [0003] The methods for monitoring the status of parking spaces include: burying induction coils under the ground of the parking spaces, and judging whether there is a car in the parking spaces by detecting changes in the magnetic field; or tracking moving objects within the effective range through ultrasonic detection technology, ...

AI Technical Summary

Problems solved by technology

[0003] The methods for monitoring the status of parking spaces include: burying induction coils under the ground of the parking spaces, and judging whether there is a car in the parking spaces by detecting changes in the magnetic field; or tracking moving objects within the effective range through ultrasonic detection technology, and identifying parking spaces based on the detection results. These two parking space sensors all need to be constructed and installed above or below the parking space, and then the situation of each parking space is transmitted to the upper end through a dedicated network. The above methods all have the following disadvantages: the installation and construction are complicated, and it is necessary to open the road, dig grooves, etc. The construction cost and maintenance cost are high. Because it is installed on the ground, it is easy to be accidentally bumped by the driver, man-made damage, etc., and the later maintenance cost is high. Each sensor can only monitor one or two parking spaces, especially not suitable for outdoor open parking lots. Lay out a lot of induction equipment

However, there are many problems in this method in practical application: First, different vehicles have large differences in size, shape, and color, and the deformation caused by the distance and angle in the field of view is different, which increases the complexity of algorithm design. Large; second, vehicles and pedestrians frequently enter and exit in the open parking lot, and the walking route is unknown, causing great interference to the parking space, and the ability of the monitoring algorithm to eliminate interference is poor

Images