Detection Algorithm Of DEM Gross Error Considering Topographic Patterns

The airborne LiDAR (Light Detection and Ranging) point cloud data, which is utilized to generate Digital Elevation Model (DEM), has been one of the most effective ways to express terrain morphology. It is an important section to filter the original LiDAR point cloud data and acquire the terrain point cloud data in generating DEM. But the non-terrain points which has the characteristics of DEM gross error, remained because of the existing incomplete filtering methods, awfully affect the DEM accuracy and its use in actual project. Therefore, it has important practical significance to regard non-terrain points as gross error and eliminate it for high accuracy DEM. At present, there are two DEM gross error detection and eliminate algorithms-based on grid data and based on irregular data, especially the former. LiDAR point cloud data is stored in the form of irregular discrete points. Although the existing research of DEM gross error detection and elimination based on irregular data has made remarkable achievements, it is still the emphasis and difficulties in the research to guarantee high gross error elimination rate for various complicated terrain. In addition, it is also the emphasis to reduce the probability of terrain point being misjudged when elimination the DEM gross error. According to the above content, the following work has been carried out:(1) Researching about the organizing and managing pattern of LiDAR point cloud data in actual engineering project. According to the characteristics of high density and large quantity, rational and effective organization form of discrete point cloud is needed to improve the algorithm efficiency.(2) According to the summarized extraction methods of DEM topographic feature line and the terrain classification methods, there will be a terrain classification research about the filtered airborne LiDAR point cloud data and extract its DEM topographic feature line. Aimed at the problem of the ineffective detection algorithm owe to the natural terrain fluctuation, a new DEM coarse detection and elimination algorithm, taking topographic feature line into account, is proposed.(3) Research the theory of gross error detection and its localization method based on iteration method with variable weights. On the basis of analyzing DEM gross error detection and elimination algorithm based on all kinds of irregular grid data, a new DEM gross error detection and elimination algorithm fitted by moving surface based on iteration method with variable weights is proposed (the following called iteration method with variable weights).(4) Utilize the existing gross error elimination algorithm and iteration method with variable weights to detect and eliminate the gross error of flat ground, hill and mountain. Compare the gross error elimination rate and false of various algorithms.(5) Analyze the influence to DEM gross error detection and elimination algorithm by the number of neighborhood, the ratio of window and the attenuation coefficient of weights.It can be concluded:the result shows that the iteration method with variable weights has a high DEM gross error elimination rate. Meanwhile, it also greatly reduces the misjudgment rate. The research results of this paper have certain engineering practical value to build the large area DEM with high precision by using LiDAR data.For the further use of DEM based on LiDAR data instead of field survey work to improve the survey efficiency and reduce the cost has an important reference value.