A novel algorithm based on fuzzy entropy iteration is proposed to simplify point cloud data. Better detail features of the streamlined point cloud model are retained while the algorithm′s operating efficiency is improved. To retain the boundary characteristics, X-Y boundary of all point cloud data is extracted quickly. Curvature of each data point is calculated. Data points except boundary points are grouped according to the curvatures. Then the number of data points and the average of the curvatures in each group are computed. Fuzzy sets of point cloud model are constructed according to the curvatures. Then the minimum fuzzy entropy is calculated to obtain the curvature threshold, so that the data points can be divided best. The data points are diluted when their curvatures are less than the threshold. The dilution ratio depends on the iteration number. When satisfying the requirements of number, the data points are processed when their curvatures are bigger than the threshold. In this iteration process, new minimum fuzzy entropy and new curvature threshold are obtained. If not satisfying the requirements of number, the data points whose curvatures are bigger than the threshold are retained. The experimental results show that the proposed algorithm approximates the point cloud model and has a satisfactory computing efficiency.