Human gait is one of the unobtrusive behavioral biometrics that has been extensively studied for various commercial and government applications. Biometric security, medical rehabilitation, virtual reality, and autonomous driving cars are some of the fields of study that rely on accurate gait recognition. While majority of studies have been focused on achieving very high recognition performance on a specific dataset, different issues arise in the real-world applications of this technology. This research is one of the first to evaluate the effects of changing walking speeds and directions on gait recognition rates under various walking conditions. Dataset was collected using the KINECT sensor. To draw an overall conclusion about the effects of walking speed and direction to the sensor, we define distance features and angle features. Furthermore, we propose two feature fusion methods for person recognition. Results of the study provide insights into how walking speeds and walking directions to the KINECT sensor influence the accuracy of gait recognition.
You may also start an advanced similarity search for this article.