Abstract: This paper presents a 3D clothing fitting via the feature matching. Firstly, the method fits each vertex and its local neighborhood of the clothing and human models by constructing the biquadratic Bezier surface. It can estimate the Gaussian curvature and the mean curvature of each vertex. Then, the feature points of two mesh models are extracted according to curvatures. In order to reduce the wrong matching and avoid the multiple matching relations, it uses the curvature constraint conditions or few interactive operations to obtain the correct corresponding feature points. The coordinate transformation matrix in 3D space is calculated to realize the rough match by the minimal distance function. Finally, the clothing model is scaled and matched again to achieve the accurate matching effect. The experimental results show that the 3D fitting algorithm is convenient and effective.