This paper presents a complete methodology for human gait analysis using a marker-free system. The acquisition system devised is composed of three synchronized and calibrated CCD cameras. The aim of this work is to recognize the leg of a walking human in gray level image sequences, to reconstruct it in the three dimensional space and then to analyze the movements of the body during gait activity. An articulated 3D model of the human body, based on tapered superquadric curves, is first introduced. A motion-based segmentation, using morphological operators is then applied to the image sequences in order to extract the boundaries of the leg in motion. A reconstruction process, based on the use of a least median of squares (LMS) regression is next performed, in order to determine the location of the human body in 3D space. Finally, a spatial coherence is imposed on the reconstructed curves in order to better fit the anatomy of the leg and to take the articulated model into account. Each stage of the proposed methodology was tested both on synthetic images and on real world images of walking humans. After having reconstructed the 3D model of the human, some describing gait parameters were extracted and analyzed in order to detect pathologies or abnormalities