Date: 2006.2.24 (Fri)
Time: 10:45-11:15
Place: Network Conference Room, Information Building, AIST, Tsukuba
Society: Technical Committee on Pattern Recognition and Media Understanding (PRMU), Institute of Electronics, Information and Communication Engineers (IEICE)

Description: HG_ Precise Team members of Otsu Laboratory present their research results at PRMU conference in February 2006. PRMU is a national conference on pattern recognition and multimedia understanding. Satoshi Ito describes the method developed by HG_ Precise Team.

Conference Code: 2006-02-PRMU-TL
Period: 2006.2.23 (Thu) - 2.24 (Fri)

Time: 10:45-11:15
Speaker: Satoshi Ito
Title: High Precision Gait Recognition Using Large Scale PC Cluster
Bibliography: Satoshi Ito, Yuuki Horita, Kenji Kaneda, Takuya Nanri, Yasuyuki Shimohata, Kenjiro Taura, Mihoko Otake, Tomomasa Sato, Nobuyuki Otsu, “High Precision Gait Recognition Using Large Scale PC Cluster”, IEICE Technical Report TL2005-82, PRMU2005-217(2006-2), pp.91-96, 2006. (In Japanese)

Abstract:
A good video surveillance system needs a high-precise gait recognition method. As such a method, a gait recognition method using cubic higher-order local auto-correlation (CHLAC) features has been proposed. However, in this method a recognition rate is decreased in the case of recognizing various sizes of objects. To overcome this problem, it can be considered to increase the number of CHLAC features, but this requires huge computation time. Our method made it possible to use a larger number of CHLAC features than a traditional CHLAC method by using parallel computing techniques. We applied this method to a data set, which contains objects whose scales change dynamically, and achieved a higher recognition rate than that of the traditional CHLAC method.