• Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
• /
• 2006.04a
• /
• pp.463-467
• /
• 2006

Several segmentation algorithms have been proposed for DTM generation or building modeling from airborne LiDAR data. Three components are important for accurate segmentation: (i) the adjacent relationship of n-nearest points or mesh, etc. (ii) the effective decision parameters of height, slope, curvature, and plane condition, (iii) grouping methods. In this paper, we created the topology of point cloud data using the contour tree and implemented the region-growing Terrain and non-terrain points were classified correctly in the segmented data, which can be used also for feature classification.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.2
• /
• pp.435-449
• /
• 2023

The state-of-the-art video-based point cloud compression(V-PCC) has a high efficiency of compressing 3D point cloud by projecting points onto 2D images. These images are then padded and compressed by High-Efficiency Video Coding(HEVC). Pixels in padded 2D images are classified into three groups including origin pixels, padded pixels and unoccupied pixels. Origin pixels are generated from projection of 3D point cloud. Padded pixels and unoccupied pixels are generated by copying values from origin pixels during image padding. For padded pixels, they are reconstructed to 3D space during geometry reconstruction as well as origin pixels. For unoccupied pixels, they are not reconstructed. The rate distortion optimization(RDO) used in HEVC is mainly aimed at keeping the balance between video distortion and video bitrates. However, traditional RDO is unreliable for padded pixels and unoccupied pixels, which leads to significant waste of bits in geometry reconstruction. In this paper, we propose a new RDO scheme which takes 3D-Distortion into account instead of traditional video distortion for padded pixels and unoccupied pixels. Firstly, these pixels are classified based on the occupancy map. Secondly, different strategies are applied to these pixels to calculate their 3D-Distortions. Finally, the obtained 3D-Distortions replace the sum square error(SSE) during the full RDO process in intra prediction and inter prediction. The proposed method is applied to geometry frames. Experimental results show that the proposed algorithm achieves an average of 31.41% and 6.14% bitrate saving for D1 metric in Random Access setting and All Intra setting on geometry videos compared with V-PCC anchor.

• Journal of IKEEE
• /
• v.27 no.4
• /
• pp.486-493
• /
• 2023

In this paper, we propose an automatic hand gesture area extraction and recognition technique using FMCW radar-based point cloud and LSTM. The proposed technique has the following originality compared to existing methods. First, unlike methods that use 2D images as input vectors such as existing range-dopplers, point cloud input vectors in the form of time series are intuitive input data that can recognize movement over time that occurs in front of the radar in the form of a coordinate system. Second, because the size of the input vector is small, the deep learning model used for recognition can also be designed lightly. The implementation process of the proposed technique is as follows. Using the distance, speed, and angle information measured by the FMCW radar, a point cloud containing x, y, z coordinate format and Doppler velocity information is utilized. For the gesture area, the hand gesture area is automatically extracted by identifying the start and end points of the gesture using the Doppler point obtained through speed information. The point cloud in the form of a time series corresponding to the viewpoint of the extracted gesture area is ultimately used for learning and recognition of the LSTM deep learning model used in this paper. To evaluate the objective reliability of the proposed technique, an experiment calculating MAE with other deep learning models and an experiment calculating recognition rate with existing techniques were performed and compared. As a result of the experiment, the MAE value of the time series point cloud input vector + LSTM deep learning model was calculated to be 0.262 and the recognition rate was 97.5%. The lower the MAE and the higher the recognition rate, the better the results, proving the efficiency of the technique proposed in this paper.

• Journal of Korean Society of Forest Science
• /
• v.112 no.2
• /
• pp.195-208
• /
• 2023

This study aimed to use three-dimensional point cloud data (PCD) obtained from Terrestrial Laser Scanning (TLS) and Mobile Laser Scanning (MLS) to evaluate a deep learning-based species classification model for two tree species: Pinus koraiensis and Larix kaempferi. Sixteen models were constructed based on the three conditions: LiDAR platform (TLS and MLS), down-sampling intensity (1024, 2048, 4096, 8192), and deep learning model (PointNet, PointNet++). According to the classification accuracy evaluation, the highest kappa coefficients were 93.7% for TLS and 96.9% for MLS when applied to PCD data from the PointNet++ model, with down-sampling intensities of 8192 and 2048, respectively. Furthermore, PointNet++ was consistently more accurate than PointNet in all scenarios sharing the same platform and down-sampling intensity. Misclassification occurred among individuals of different species with structurally similar characteristics, among individual trees that exhibited eccentric growth due to their location on slopes or around trails, and among some individual trees in which the crown was vertically divided during tree segmentation.

• ETRI Journal
• /
• v.29 no.5
• /
• pp.641-648
• /
• 2007

In this study, we suggest a new segmentation algorithm for processing airborne laser point cloud data which is more memory efficient and faster than previous approaches. The main principle is the reading of data points along a scan line and their direct classification into homogeneous groups as a single process. The results of our experiments demonstrate that the algorithm runs faster and is more memory efficient than previous approaches. Moreover, the segmentation accuracy is generally acceptable.

• The Journal of Korea Robotics Society
• /
• v.16 no.2
• /
• pp.130-136
• /
• 2021

In recent years, much study has been conducted in robotic grasping. The grasping algorithms based on deep learning have shown better grasping performance than the traditional ones. However, deep learning-based algorithms require a lot of data and time for training. In this study, a grasping algorithm using an artificial neural network-based graspability estimator is proposed. This graspability estimator can be trained with a small number of data by using a neural network based on the residual blocks and point clouds containing the shapes of objects, not RGB images containing various features. The trained graspability estimator can measures graspability of objects and choose the best one to grasp. It was experimentally shown that the proposed algorithm has a success rate of 90% and a cycle time of 12 sec for one grasp, which indicates that it is an efficient grasping algorithm.

• Korean Journal of Remote Sensing
• /
• v.37 no.6_1
• /
• pp.1527-1543
• /
• 2021

In this paper, we explore the applicability and utility of a technology that generating clustered solid building models based on point cloud automatically by applying it to various data. In order to improve the quality of the model of insufficient quality due to the limitations of the automatic building modeling technology, we develop the building shape modification and texture correction technology and confirmed the resultsthrough experiments. In order to explore the applicability of automatic building model generation technology, we experimented using point cloud and LiDAR (Light Detection and Ranging) data generated based on UAV, and applied building shape modification and texture correction technology to the automatically generated building model. Then, experiments were performed to improve the quality of the model. Through this, the applicability of the point cloud data-based automatic clustered solid building model generation technology and the effectiveness of the model quality improvement technology were confirmed. Compared to the existing building modeling technology, our technology greatly reduces costs such as manpower and time and is expected to have strengths in the management of modeling results.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.15 no.6
• /
• pp.269-279
• /
• 2020

This paper presents the design and implementation of embedded systems and autonomous path generation for autonomous speed sprayer. Autonomous Orchard Systems can be divided into embedded controller and path generation module. Embedded controller receives analog sensor data, on/off switch data and control linear actuator, break, clutch and steering module. In path generation part, we get 3D cloud point using Velodyne VLP16 LIDAR sensor and process the point cloud to generate maps, do localization, generate driving path. Then, it finally generates velocity and rotation angle in real time, and sends the data to embedded controller. Embedded controller controls steering wheel based on the received data. The developed autonomous speed sprayer is verified in test-bed with apple tree-shaped artworks.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.16 no.1
• /
• pp.31-49
• /
• 2012

In this paper, we propose a very efficient method which reconstructs the high resolution surface from a set of unorganized points. Our method is based on the level set method using adaptive octree. We start with the surface reconstruction model proposed in [20]. In [20], they introduced a very fast and efficient method which is different from the previous methods using the level set method. Most existing methods[21, 22] employed the time evolving process from an initial surface to point cloud. But in [20], they considered the surface reconstruction process as an elliptic problem in the narrow band including point cloud. So they could obtain very speedy method because they didn't have to limit the time evolution step by the finite speed of propagation. However, they implemented that model just on the uniform grid. So they still have the weakness that it needs so much memories because of being fulfilled only on the uniform grid. Their algorithm basically solves a large linear system of which size is the same as the number of the grid in a narrow band. Besides, it is not easy to make the width of band narrow enough since the decision of band width depends on the distribution of point data. After all, as far as it is implemented on the uniform grid, it is almost impossible to generate the surface on the high resolution because the memory requirement increases geometrically. We resolve it by adapting octree data structure[12, 11] to our problem and by introducing a new redistancing algorithm which is different from the existing one[19].

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2008.06a
• /
• pp.969-970
• /
• 2008