A vehicle re-identification method and system

Abstract

The present invention provides a vehicle re-identification method and system, based on vehicle component features and set distance metric learning, including the following steps: extracting the global features and local features of the vehicle, and determining the weight of each local feature based on the quality of the local features ; Through ensemble distance metric learning, complete the vehicle re-identification process. The present invention uses ensemble distance metric learning to accelerate the process of feature learning. The ensemble distance metric learning first treats different images of the same car as an ensemble, and optimizes the feature learning process by reducing the image distance within each ensemble while increasing the distance between different ensembles. The invention effectively reduces the computational complexity of training, and at the same time can obtain features with more discriminative power, and can perform vehicle re-identification more accurately.

Description

technical field [0001] The invention relates to the technical field of vehicle identification, in particular to a vehicle re-identification method and system based on vehicle component features and set distance metric learning. Background technique [0002] The vehicle images captured by the monitoring system without overlapping fields of view are the main processing objects used in the vehicle re-identification problem. However, these vehicle images contain problems such as viewing angle changes, resolution, illumination changes, blur, camera settings, complex backgrounds, and occlusions. This makes the vehicle re-identification problem more difficult, and the solutions to these problems are still being studied by many scholars. In the field of vehicle re-identification under the non-overlapping view monitoring system, there are many methods proposed by domestic and foreign researchers. These methods can be roughly divided into two categories, one is the vehicle re-identif...

AI Technical Summary

Problems solved by technology

Using the method of metric learning, it has been successfully applied to the problem of pedestrian re-identification, and has achieved good results, but it is still a relatively new problem to apply it to vehicle re-identification.

Images