This paper addresses the nonlinear shape registration by using the constrained implicit representation for the shape. First, we adopt the level set of the implicit function to represent the shape and model the shape registration, where a global-to-local strategy is applied. Then, in order to improve the regularity, the scale and the bending constraints are introduced to the global and local parameters respectively. Further, a first order variational model is deduced and solved by the gradient descent method. Finally, the comparison of the proposed algorithm with some state-of-the-art approaches on a variety of data sets validates our better performance.