Non-standard parking behavior recognition method and device

Abstract

The invention discloses a method for realizing non-standard parking behavior identification by using a monitoring video. A high-point camera is combined with a GIS and a computer vision technology torealize rapid, accurate and real-time identification of various non-standard parking behaviors according to parking postures. The method mainly comprises the steps of constructing a non-standard parking behavior rule base, extracting spatialized parking space poses of a video scene based on an improved Vibe algorithm, an SSD convolutional neural network model and a CSRT tracking algorithm, and finally realizing accurate detection of non-standard parking behaviors in combination with a geometrical relationship between parking spaces and parking features. The invention further provides a devicebased on the method, and the invention is high in recognition precision and high in real-time performance.

Description

technical field [0001] The invention relates to an image recognition method, in particular to an irregular parking behavior recognition method and device. Background technique [0002] The phenomenon of irregular parking occurs from time to time, resulting in a series of phenomena such as vehicle scratches and traffic jams. At present, there are mainly two ways to detect irregular parking behaviors: manual identification and sensor identification. Manual identification is mainly through on-site law enforcement, which consumes a lot of manpower and material resources. Sensor detection methods can be divided into two types: intrusive detection and non-invasive detection. The magnetic induction coil method is the most widely used in intrusion detection. It uses coils and capacitive devices laid underground to generate an oscillating circuit. When a vehicle passes by, the metal parts will cause the circuit to change and thus be detected (Kong Q J.An Approach to Urban Traffic S...

AI Technical Summary

Problems solved by technology

Lv Fenghua (Lv Fenghua. Research on Vehicle Classification Method Based on SVM Considering Tilt Angle [D]. Kunming: Kunming University of Science and Technology, 2011) used Freeman linked list to track and extract the vehicle outline, and then obtained the center of the front and rear wheels of the vehicle through the circle detection method, and calculated the vehicle The tilt angle between the image and the horizontal direction, this method only uses the external geometric features of the vehicle to identify the vehicle, and the classification recognition rate is easily affected by image shadows and mutual occlusion between vehicles; ZhangZ (Zhang Z et al.Three-dimensional deformable-model-based localization and recognition of road vehicles[J].IEEE Transactions on Image Processing,2012,21(1):1-13) proposes to use the wireframe model for vehicle pose estimation. This method optimizes the matching of the wireframe model and the vehicle outline in the image Error to obtain the final pose, the method is relatively simple, but the pose estimation accuracy is not high; The mapping method extracts the vehicle pose parameters to analyze the three driving states of the vehicle forward, left, and right. This method requires the use of distributed targets, and the pose calculation accuracy is affected by the target extraction effect.

[0009] However, most of the existing methods only detect the general direction of the vehicle, and do not fully consider various complex and irregular parking behaviors. The detection method for the vehicle pose is simple and has poor robustness. Low and susceptible to image noise

Images