A method to recognize the moving human activity about posture and velocity
Zhang Lijie, Lv Shaozhong, Guo Xu · 2011
This work describes a method to recognize the moving human activity about posture and velocity, which can be useful for various further advanced human activity analysis. Considering the width information of the body can cover the outline of the body, numbers of the non-zero pixels in every row of the image including the detected moving human are used as the body characteristics and according to the width characteristics a neural network categorizer used to recognize body postures is designed in this paper. In order to decrease the calculation error of the velocity caused by the perspective projection, area of the moving target is proposed to modify the velocity calculated in the two-dimensional coordinate system. The effectiveness of the proposed method in this paper is demonstrated with experiments in an indoor environment.