Vehicle Re-Identification Method Based on Keypoint Detection and Local Feature Alignment

Abstract

The invention belongs to the field of computer vision and pattern recognition, and specifically relates to a vehicle re-identification method based on key point detection and local feature alignment, aiming to solve the problem that the existing vehicle re-identification method has poor feature consistency when eliminating vehicle viewing angle changes. This leads to the problem of poor robustness of vehicle re-identification. The method includes acquiring an image of a vehicle to be identified as an input image; performing key point detection on the input image, obtaining key points of the vehicle to be identified and their corresponding confidence levels, and dividing the vehicle to be identified in the input image into N parts , as a local image; extract the features of the input image and each local image as global features and local features, and splicing each local feature with the global feature as a comprehensive feature; calculate the relationship between the comprehensive feature and the corresponding features of each image in the vehicle image library The distances are sorted, and the sorted results are output as re-identification results. The invention improves the robustness of vehicle re-identification.

Description

technical field [0001] The invention belongs to the field of computer vision and pattern recognition, and in particular relates to a vehicle re-identification method, system and equipment based on key point detection and local feature alignment. Background technique [0002] Vehicle re-identification is a subproblem in the field of image retrieval. Given a query vehicle image, the vehicle re-identification task aims to find images of the same vehicle in other scenes. The essence of its task is to learn a vehicle feature vector generation method to represent the unique characteristics of a vehicle, which is not only robust to various external changes of the same vehicle (such as viewing angle, occlusion, lighting, license plate replacement), but also able to Discriminative enough for two cars that look alike. Among them, the essence of changes such as occlusion and perspective is that the spatial position of the same semantic region in different images of the same car is no...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the existing vehicle re-identification method has poor feature consistency when eliminating vehicle viewing angle changes, which leads to poor vehicle re-identification robustness, the first aspect of the present invention , a vehicle re-identification method based on keypoint detection and local feature alignment is proposed, which includes:

Images