Outliers Detection Method Based on Dynamic Standard Deviation Threshold Using Neighborhood Density Constraints for Three Dimensional Point Cloud

In 3 D point clouds reverse engineering, the outliers detection plays a key role on the subsequent processing. However, when the point cloud has big change density distribution, the detection of outliers becomes very difficult. In order to get a feasible detection result, improve the detection ability and adaptivity, an outliers detection method was proposed, which based on dynamic standard deviation threshold using k-neighborhood density constraints. This method fully considered the density difference of the obtained point cloud, and introduced the density characteristics into calculation of the determining threshold. Firstly, the target point cloud by pass-through filtering was extracted, and the invalid points were removed. Then the detection principle was analyzed, the k-neighborhood density estimation method was presented. Finally the dynamic standard deviation threshold constrained by the k-neighborhood density was calculated, the different constraints for outer regions and inlier regions were obtained, and a better detection result for point cloud with big change density distribution was got. Experimental results show that the method can apply to the highly variable density distribution point cloud, get a feasible detection result, improve detection effect and performance, and is positive to practical applications.