A rapid vehicle type identification method

Abstract

The invention provides a rapid vehicle type identification method, which mainly solves the problems of accuracy and real-time performance in vehicle type identification, and comprises the following steps: firstly, combining color space conversion with a multi-channel HOG feature extraction algorithm, and extracting vehicle front face features while reducing the influence of illumination environment; using PCA dimension reduction operation to reduce the sample feature dimension, and reducing the calculation complexity; Performing sparse representation and nonlinear mapping on the sample features to reduce the correlation between the features; And finally, establishing a relation between the sample characteristics and the sample labels, and solving a weight coefficient between the sample characteristics and the sample labels to realize a rapid vehicle model identification effect. Experimental results on the BIT-Vehile database show that the identification precision of the method is 96.69%, the identification speed is 70.3 fps, and the real-time performance is ensured while the vehicle type identification precision is improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a method for quickly identifying vehicle types. Background technique [0002] In recent years, with the popularization of traffic monitoring equipment and the rapid development of computer vision, some computer vision technologies based on traffic video images have been applied in modern intelligent transportation systems. As an important part of intelligent transportation system, real-time vehicle recognition technology has a wide range of applications, such as highway toll system, traffic flow statistics, urban traffic monitoring and assisting criminal investigation [Document 1] (Zhang F, Wilkie D, Zheng Y, et al.Sensing the Pulse of Urban Refueling Behavior[C] / / Proceedings of the 2013ACM International Joint Conference on Pervasive and Ubiquitous Computing.New York:ACM,2013:13-22.) and other aspects. [0003] At present, research in the field of vehicle ide...

AI Technical Summary

Problems solved by technology

The matching principle of the method based on the 3D model is simple, but the modeling process is complex, the robustness is poor, and the recognition accuracy is low; the method based on the deep network model has strong fault tolerance and high recognition accuracy, but this type of method requires a large number of training samples, the computational complexity is high, time-consuming, and it is difficult to meet the real-time requirements; the feature extraction-based method is faster than the deep network model due to its fixed feature extraction method, but its recognition accuracy is low

Images