Abstract
In this paper, we propose a new method of feature homogenization to prevent changes in the features according to human subtle direction change, in the detection of the human behavior of CHLAC. The conventional human behavior detection using CHLAC, it shows a high performance, but the situation that moving greatly changes the direction of the vector, abnormal operation normal operation and different operation if it is difficult to determine. We propose a method in addition to improved processing mask pattern based on the correlation and vector averaging process based on an amplification to suppress the change in the amount of features according to the direction of movement of the person. This provides a robust human motion specific person to change direction.
