Vehicle image accurate retrieval method based on big data

Abstract

The invention provides a vehicle image accurate retrieval method based on big data. The method comprises the steps that (1) the vehicle target and the vehicle front windshield area are detected by using a detection method based on the YOLO V2 algorithm; (2) a vehicle target attribute identification model based on a deep convolutional neural network is constructed and trained so as to identify thevehicle target attributes; (3) the ORB features and the feature descriptors of the front windshield area Ia of the suspected vehicle and the front windshield area Ib of a vehicle to be retrieved are extracted; (4) the Hamming distance between the ORB features of the image Ia and the image Ib is calculated, and the matching point pairs between Ia and Ib are found through the fast nearest neighbor method for violent matching; (5) the wrongly matched feature point pairs of the step (4) are eliminated by using the grid-based motion statistical method GMS; and (6) the ORB feature similarity betweenthe front windshield area of the suspected vehicle and the front windshield area of the vehicle to be retrieved is calculated. The problem of fast and accurate vehicle search in the massive traffic images can be solved on the whole.

Description

technical field [0001] The invention relates to intelligent transportation technology, in particular to a vehicle precise retrieval method based on image big data. Background technique [0002] With the continuous increase of motor vehicles, there are more and more violations of laws and regulations in driving vehicles. In order to meet the needs of urban security prevention and control and traffic management, a large number of surveillance cameras are installed at urban traffic intersections. The vehicle images collected by the camera provide strong evidence support for law enforcement agencies to crack down on vehicle illegal and criminal activities. However, with the continuous increase of urban video surveillance equipment, it is becoming more and more difficult to quickly locate suspected vehicles from massive surveillance data. The traditional method of manually finding vehicles not only wastes a lot of manpower, material resources, and financial resources, but also ...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the above technical problems, the purpose of the present invention is to overcome the shortcomings of low vehicle r...

Images