Resolution adaptive feature extracting and fusing for pedestrian re-identification method

Abstract

The invention discloses a resolution adaptive feature extracting and fusing forpedestrian re-identification method. According to the invention, biological features and appearance features are fused, and a method fusing face features and experience features is used for identifying pedestrians, so feature diversity is improved; according to performance of the features in different image scales, different pedestrianfeatures are compared in different scales; a screening mechanism is adopted in which fusion features obtained by use of color features and profile features are firstly screened, then the face features are used for complementing screening results and at last texture features are extracted from the screened pedestrians, so color region features and weighing color features are greatly stabilized; and by using overall features and local features obtained via fusion and extraction by use of the self-adaptive weighing method, the fusion features are obtained. In this way, via self-adaptive fusion of the biological features and the appearance features, accuracy of the method can be improved; and by extracting the appearance features in a low scale, via the screening mechanism, complexity can be reduced.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a pedestrian re-identification method for resolution adaptive feature extraction and fusion. Background technique [0002] Person re-identification is a very important problem in video surveillance. Although many studies have focused on this field, person re-identification still faces many challenges. 1. At present, the resolution of pedestrian pictures captured in surveillance sites is low, and it is very difficult to extract clear biometric features for pedestrian re-identification. Second, the time when different pedestrians appear in the two camera scenes varies greatly, and it is very difficult to use time and space constraints to help pedestrians re-identify. Third, relying solely on appearance features to identify pedestrians is not stable, because pedestrian appearance features will be affected by posture, illumination, background and occlusion. Fourth, the l...

AI Technical Summary

Problems solved by technology

1. At present, the resolution of pedestrian pictures captured in surveillance sites is low, and it is very difficult to extract clear biometric features for pedestrian re-identification

2. The time when different pedestrians appear in the two camera scenes varies greatly, and it is very difficult to use time and space constraints to help pedestrians re-identify

Third, relying solely on appearance features to identify pedestrians is not stable, because pedestrian appearance features will be affected by posture, lighting, background and occlusion

4. The low calculation efficiency of pedestrian re-identification in large-scale video surveillance networks also brings great challenges to the application of pedestrian re-identification

The metric learning method will greatly improve the comparison effect by performing calibration training on a specific scene, but it is not universal and needs to be retrained for new scenarios. The training and calibration process is more complicated, and the space complexity of the method is High, it is still difficult to apply to the actual system

Images