Search

Objectives: Point cloud has no topological structure, current deep learning semantic segmentation algorithm is difficult to capture geometric features implied in irregular points. In addition, the point cloud is in three-dimensional space with a large amount of data size. If we blindly expand the captive field size during extract neighborhood information, it will increase the number of model parameters, which will make model training difficult. Methods: We propose a point cloud semantic segmentation model based on the dilated convolution and combining elementary geometric features such as angle as the model input. First, during feature extraction, basic geometric features such as the relative coordinates, distance and angle between the centroid and the neighboring points are used as the model input to mine the geometric information. Second, in the process of building local neighborhoods, we expand the image dilated convolution operator to point cloud processing, the point cloud dilated operator can expand the receptive field size with no increasing the number of parameters of the model. Finally, the dilated convolution operator, multi-geometric features encoding modules and U-Net architecture are combined to form a complete point cloud semantic segmentation model. Results: The results show that compared with the traditional neighborhood structure, the overall accuracy (OA) of dilated neighborhood structure is increased by 1.4%. Compared with the model that only uses coordinates as input, multi-geometric features encoding module is increased by 10.7%. The final model based on the two proposed algorithms get mean intersection over union and OA are 91.2% and 68.2%, respectively. Conclusions: The dilated neighborhood structure can effectively extract point cloud information in a larger range without increasing the number of model parameters. multigeometric features encoding module can maximize the capture of shape information in the neighborhood. [ABSTRACT FROM AUTHOR]

For mobile LiDAR point cloud data, a new hybrid index structure combining global KD-tree and local Octree is proposed to improve the efficiency of data organization and management, which is named as KD-OcTree index. Firstly, global KD-tree reconstructs the spatial neighborhood relations by defining the segmenting dimension and segmenting planes, for the purpose of ensuring the balance of the whole index. Then, local Octree is constructed in the leafs of KD-tree, which can avoid some shortcomings such as the unbalance of point cloud distribution, deeper Octree, large amount of non-point space, and so on. Lastly, we take three real scenes' point clouds as test data to process. The experimental results and comparative analysis show that the KD-OcTree index can not only improve the speed of constructing index and neighborhood searching, but also improve the effect of data-processing and influence the reliability of classification. [ABSTRACT FROM AUTHOR]

In industrial metrology, a multi-type instrument is only combined to completing all measurement task, and measurement data type and accuracy of each type instrument is different, so valid adjustment model is need to setting up and solving spatial position and attitude parameter of multi-type instrument. Based on observation equation with condition equations, the article established generalized USMN adjustment model and combined adjust for different type instruments, such as laser tracker, total station and theodolite. Furthermore, because of different data type, efficient weight matric model is build up to get accuracy weight, and this model can get precision position without abundant redundant observation. Based on angular resection and coordinate transformation, spatial network preliminary computation model can solve approximate parameter without theodolite each aiming angle and scalebar measurement. At last, measurement datum is solved with generalized USMN adjustment model, the solution result shows that the model can solve different instrument and raise the point accuracy. [ABSTRACT FROM AUTHOR]

The Position and Orientation System (POS) is a critical element in mobile mapping systems. Because of the complex system integration, the center of the POS is hard to observe. Thus, a cooper­ative target could be used for tracking the motion stale of a POS. if the relative position between tar­get and A POS is determined. Based on the positioning equation for mobile mapping systems, we ana­lyze the calibration techniques for cooperative target lever-arm parameters. An error model was de­signed for predicting the accuracy of mounting parameters of cooperative targets, with consideration of the impact from errors such as laser ranging error, angular measurement error, position error, orien­tation error, and scale factor error. Outdoor experiments were carried out to verify the feasibility of this calibration method, verifying that it could be used for testing the dynamic position accuracy of POS. [ABSTRACT FROM AUTHOR]