The field of human motion analysis has researched with designing the rehabilitation robot system. In the present, there are many robot systems dealing with the human motion but they are heavy and hard to use simply at home. Moreover they especially have a particular purpose focused on the single human motion then the usability of these systems is low and fragmentary about a variety of motions. To classify the various motions, we propose the approach to categorization analysis for human motion by using information of the Kinect and IMUs. It is focused on the human motion which is classified into the walking, standing up and down, and falling down motion. The two COG points of the body from the sensing results are defined as the new indexes such as distance and incline, which are the indicator to classify the human motion. To verify the verification, the theoretical human model is designed and the experiment for the various human motions is carried out by using Kinect and IMU.