In numerous content-based video applications, it is important to extract from a video sequence a representation for humans in motion. This task is difficult, because humans are not rigid objects and they are capable of performing a wide variety of actions. However, often, human movements can be categorized into repetitive and rhythmic patterns of motion. Identifying the motion pattern of a human significantly alleviates the task of construction of its representation. We propose here a model-based recognition of the generic posture of human walking in dynamic scenes. We model the human body as an articulated object connected by joints and rigid parts, and model the human walking as a periodic motion. The recognition task is to fit the model walker sequence to the walker in the live video (data walker sequence). We achieve this by determining the period of the data walker sequence and finding its phase with respect to the model walker sequence. We present promising results of how our system performs with a live video sequence.