HOME > Research > Scene situation transition recognition group

Scene situation transition recognition group

In order to achieve "Friendly Informative Cyber Space", it is necessary that we make computer that can detect a person or an object(or a group) which is located inside of the subject environment, recognize his or its(or their) whereabouts and movement, and think subjectively about a static or dynamic image like a person. For these complex problems solving, we do research on Time-varying Image Processing(Video Analysis, Monitoring), Human Tracking, Behavior Analysis, Situation Recognition, and so on.

Keywards:
Time-varying Image Processing(Video Analysis, Monitoring), Human Tracking, Behavior Analysis, Situation Recognition

Core Techniques:
Pattern Recognition, Multivariate Analysis(Principal Component Analysis, Factor Analysis), Machine Learning (AdaBoost), Voronoi Tesselation Method(Wavefront Method, Fast-forward Method), Region Segmentation Method(Graph-cut, Labeling), EM Algorithm, Tracking Moving Object(MeanShift Method, CamShift Method, Optical Flow Analysis)

Examples:
Sakurajima smoke state monitoring system and a fallout ash forecast
Source: Kosei Nagatani, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe,
"Research on Sakurajima smoke state monitoring system and a fallout ash forecast",
Meeting on Image Recognition and Understanding 2011, IS4-34 (2011.7)

Auto-monitoring system in the elderly-people institution by video processing
Source: Ryuji Sakamoto, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe, Hiroshige Sannou, Toshihide Fukumoto, Asutsugu Kurono,
"Research for auto-monitoring method in the elderly-people institution by video processing",
The 64th Joint Conference of Electrical and Electronics Engineers in Kyushu, 08-2P-11 (2011.9)

Arrangement of Monitoring Cameras Based on Non-diffracting Voronoi Diagram
Source: Yuki Kiyama, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe,
"Research on placement problem of Monitoring Camera system Based on (High-order) Non-diffracting Voronoi Diagram",
Meeting on Image Recognition and Understanding 2012, IS3-68 (2012.8)

Auto-detection of the potential dangerous situation during driving
Source: Kazuya Suzuki, Ryuji Sakamoto, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe,
"Research on automatic crossing pedestrian detection and tracking method for danger prediction during driving",
The 65th Joint Conference of Electrical and Electronics Engineers in Kyushu, 09-2P-08 (2012.9)

Automated analysis of the walking race form
Source: Kimi Nagatomo, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe,
"A study on the automated analysis of the walking race form",
The 65th Joint Conference of Electrical and Electronics Engineers in Kyushu, 10-1A-14 (2012.9)

Shot boundary detection of movie
Source: Kotaro Yokomura, Masayuki Kashima, Kiminori Sato, Mutsumi Watanabe,
"A study on the scene extraction of the movie",
The 65th Joint Conference of Electrical and Electronics Engineers in Kyushu, 09-2P-07 (2012.9)

Members