Vehicle model identification method under traffic monitoring

Abstract

The invention provides a vehicle model identification method under traffic monitoring. After image preprocessing and positioning and identification of a vehicle plate, the steps of identifying the vehicle model are as follows: with the position of the vehicle plate as the reference, acquiring a candidate rectangular area; analyzing a candidate area histogram, and identifying the candidate area histogram to describe a vehicle head or a vehicle tail; positioning an air inlet hole in the vehicle head or a vehicle lamp in the vehicle tail; acquiring the edge direction of each pixel point, and calculating an edge histogram descriptor; matching a database according to the edge histogram descriptor and acquiring a corresponding vehicle model type. The vehicle model identification method disclosed by the invention has the advantages that firstly, the problems caused by difficulty in positioning of a vehicle logo and too-little vehicle logo information are solved; secondly, based on the uniqueness of the air inlet hole in the vehicle head and the vehicle lamp in the vehicle tail, according to the method, a vehicle manufacturer (the vehicle logo) can be identified, and the vehicle model and the age limit can be identified; thirdly, the identification can be realized through the vehicle head and the vehicle tail.

Description

technical field [0001] The invention relates to a method for automatic vehicle model identification according to traffic monitoring image data, which belongs to the technical field of intelligent transportation. Background technique [0002] With the maturity of license plate recognition technology, the importance of vehicle logo recognition in intelligent transportation is becoming more and more obvious. This technology can be used for: (1) While identifying the license plate number, it is necessary to determine the consistency between the license plate and the model, that is, to identify whether it is a set plate vehicle. (2) The data volume of surveillance camera information is increasing exponentially, and the search for suspected vehicles has become a major problem for public security and traffic control departments. (3) According to the visual experience of the human eye, it is often more sensitive to the car series, model, and body color, and the identification of ca...

AI Technical Summary

Problems solved by technology

(2) The data volume of surveillance camera information is increasing exponentially, and the search for suspected vehicles has become a major problem for public security and traffic control departments

[0005] Vehicle logo positioning is the bottleneck of this method. The accuracy of positioning directly affects the accuracy of vehicle logo recognition. However, in view of the uncertainty of the location of the car logo and the impact of environmental noise around the car logo, there is no ideal positioning method. In addition, in high-definition surveillance cameras, the car logo has fewer pixels and less information, which also has a certain impact on the recognition rate.

Images