To address the large amount of redundant data in the original point cloud, a point cloud simplification algorithm based on fast point feature histograms (FPFH) feature extraction was proposed, effectively taking into account the retention of feature information and overall integrity. Firstly, the algorithm sought and retained the boundary points of the original model. Then, the FPFH value of non-boundary points were calculated, thus producing the feature value of the point cloud. After sorting the feature values, the non-boundary points were divided into the feature region and the non-feature region, retaining the points in the feature region. Finally, the non-feature region was divided into k sub-intervals, and the improved farthest point sampling algorithm was employed to sample each sub-interval. The proposed algorithm was compared with the farthest point sampling algorithm, non-uniform grid method, k-means algorithm, and adaptive curvature entropy algorithm, and the simplified point cloud was evaluated by the standardized information entropy evaluation method. Experimental results show that the proposed algorithm outperforms other simplification algorithms. In addition, the visualization results indicate that the proposed algorithm can not only ensure the integrity of the simplified model but also retain most of the feature information of the point cloud.