@misc{oai:tokyo-metro-u.repo.nii.ac.jp:00007030,
 author = {テイ, ラクラク and 丁, 楽楽},
 month = {Mar},
 note = {During the past decade, gait recognition technologies have been attracted more and more attention in the field of biometrics. Some identification systems, for instance, the face, retina, and fingerprint have been widely used already. However, they are also expensive, require the cooperation of the users and so on. In contrast to traditional identification technologies, gait recognition can be used from a distance and even do not need users’ direct cooperation. Although a lot of effects has been spent on developing practical gait recognition systems, none of this system developed was perfect and all were far away from ready to be used in commercial because many challenging problems need be solved, which mostly concentrate on feature extraction. In recent years, studies on gait recognition were mostly based on image processing technology. However, specialized computer science knowledge was necessary. Moreover, human gait was the movement of three-dimensional, image processing technology can only make use of the feature in two-dimensional, and will miss more effective characteristics. With the development of RGB-D sensor, it can be used as marker-less motion capture system to capture and record human movement without attaching markers to the subjects. Although it was a cheap and convenient equipment, it also has the disadvantage of the precision. Ground reaction force has been concentrated for a long time as an important feature in the field of biomechanics. According to the difference of weight and personal gait, there are also some individual features in ground reaction force. Furthermore, it could be measured by force plate that placed on the floor easily. The purpose of this study was to develop a new gait recognition system by combining RGB-D sensors with force plate. RGB-D sensors are a specific type of depth sensing devices that work in association with a RGB camera. By constructing the subjects’ database and using the support vector machine as a pattern recognition tool to identify subjects. There 5 chapters in this paper. To begin with this paper, the introduction of background, related study, purpose were described. At the end of chapter 1, one of the most representative RGB-D sensors (KINECT) was introduced. In chapter 2, in an attempt to evaluate the precision of RGB-D sensor, we carried out a pilot experiment. The same motion of subject was recorded by motion capture system and RGB-D sensor. By comparing the angular of the left knee in different systems, the accuracy of RGB-D sensor was confirmed. In chapter 3, a normal gait recognition experiment was carried out, recognition rate on the condition of normal gait was tested. In section 3.4, gait feature extraction and data processing methods were described in details. The basic principle of support vector machine (SVM) and an open source library of SVM (LIBSVM) were introduced in this section. In chapter 4, walking conditions have been changed by using the orthosis, the measurement frequency of force plate has been declined, and recognition rate on the condition of abnormal gait was tested. With the difference from normal gait, abnormal gait has many changes in walking movement. Some new gait features have been extracted to improve the recognition rate in this chapter. In the last chapter 5, we summarized this paper and make some conclusions about this paper. At the last of this chapter, the future study was elucidated., 首都大学東京, 2018-03-25, 修士（工学）},
 title = {Development of Gait Recognition System Using RGB-D Sensor and Force Plate},
 year = {2018}
}