Gait identifying method with robustness to walking gait changes

Abstract

The invention discloses a gait identifying method with robustness to walking gait changes. A distance metric expression among gait feature matrixes of different walking states is established, samples of different walking states of a same individual in a training stage are related and are coupled into a same image space through a machine learning strategy, and after training is completed, projection matrixes of the samples themselves under the different walking states can be obtained. In an identifying stage, when the walking states of the samples to be tested are not uniform with walking states of a registered sample collecting base, projection on projection shafts of the different walking states obtained through training are respectively conducted, and classification is conducted by the adoption of a nearest neighbor classifier. By means of the gait identifying method, prediction and estimation can be conducted without the need of changing one walking state to another walking state, and the gait identifying problem of the walking gait changes can be solved by the direct adoption of the machine learning.

Description

technical field [0001] The invention belongs to the fields of pattern recognition and machine learning, in particular to a gait recognition method robust to changes in walking states. Background technique [0002] The tragic event of "9.11" has caused the whole world to pay special attention to the enhancement of national defense, the security of terrorist attacks, and the automatic protection capabilities after terrorist attacks. Biometric identification technology has been successfully applied in identity verification, access control systems, and may also be applied in the identification of terrorists in airports and other security-sensitive places. The HID (human identification at a distance) research project funded by the U.S. Defense Advanced Research Projects Agency in 2000 is to develop and improve the performance of the current large-scale identification system under long-distance, with high reliability and robust identification. ability. In 2003, the International...

AI Technical Summary

Problems solved by technology

The model method is less affected by external interference, has short features, and can describe the changes of various parts of the body. When the modeling is accurate, the recognition effect is good, but the modeling is complicated and the matching process is complicated.

Images