Research On Discrete Point Detection And Smoothing Filter Algorithm Based On Pointcloud Data

Society is gradually entering the era of big data.With the emergence of new earth science concepts such as "digital earth","smart city",digital maps and geographic information products.And with the deepening of research,closer and closer to the real 3D spatial information technology has gradually become the hotpot int the field of geographic information science in China.Among them,the three-dimensional laser scanning technology is continuously being studied due to its extremely high data acquisition capability.In the point cloud data obtained by scanning,a certain amount of noise may be mixed into due to the conditions of the observed object,the instrument itself,or external factors.But because the point cloud data is massive,it is a big problem to improve the efficiency of point cloud filtering while maintaining the model characteristics of the point cloud data.Aiming at the shortcomings of the existing point cloud discrete point filtering algorithms,this paper proposes an outlier detection algorithm that comprehensively uses point cloud morphological information and statistical information.This algorithm can uniformly distributed discrete points and clustered discrete points in the point cloud data at the same time.This paper uses discrete point filtering algorithms that be commonly used such as radius filtering algorithm,statistical filtering algorithm,and LOF(Local Anomaly Factor)filtering algorithm for comparison and conducting qualitative and quantitative experiments.The results prove that the effect and efficiency can meet the needs of industrial production.In the point cloud smoothing algorithm,in this paper,a curvature factor is added to the bilateral filtering factor characterized by maintaining edges,which improves the influence of bilateral filtering on the accuracy of normal noise points on the filtering,and reduces the final point cloud data model.This paper uses Laplace filtering algorithm,bilateral filtering algorithm and fuzzy C-means filtering algorithm to compare with the improved bilateral filtering algorithm by this paper.The results prove that the improved bilateral filtering factor can better overcome the problem of overfitting and keep characteristics,and the flatter area in the point cloud model can better filter the burr points without increasing the time cost and meeting the needs of industrial production.