This paper presents a novel method for human recognition using gait information. The moving object position at first is detected by using the difference of two continuous frames. And then, based on the detected moving object, all of gait information was exploited. Based on the head shoulder and back contour, the walking direction can be extracted. Depend on this parameter, the lateral or frontal characteristic was focused in the features detection process. One all of these are exploited, a feature vector is constructed to serve for applying in the recognition process. To evaluate the effect of this proposed and compare with another method, we also present some simulation results which are performed on indoor outdoor environment.