摘要： Coordinate measuring machines (CMMs) are widely used in evaluating the profile error of the freeform surfaces. Traditional trigger probes are inefficient at collecting large scale measurement sets. A CMM equipped with a sweep scanning probe dramatically improves the efficiency of inspection. However, the large scale data obtained by scanning is challenging for most existing profile error evaluation methods, which are based on iterative algorithms. This paper proposes an efficient and accurate method to evaluate the profile error of freeform surfaces. To simultaneously simplify the calculation and retain the accuracy of evaluation, a new method to extract key points from scanning data set is presented. First, the key points are defined as those measured points with regional outstanding deviations from the design surface. Second, wavelet decomposition is utilized to decompose the curves formed by the deviations of the measured points in the scanning data set, and the key points are extracted according to the decomposition results. In addition, the key points set, as a representation of the scanning data, are evaluated using Sequential Quadratic Programming (SQP) algorithm. Finally, a simulation example and an actual machined part are used to test the proposed evaluation method. The results prove that the proposed method is both accurate and efficient.