• Agricultural and Biosystems Engineering
• /
• v.1 no.2
• /
• pp.95-99
• /
• 2000

Nondestructive methods such as ultrasonic and magnetic resonance imaging systems have many advantages but still much expensive. And they do not give exact color information and may miss some details. If it is allowed to destruct some biological objects to get interior and exterior informations, constructing 3D image form a series of slices sectional images gives more useful information with relatively low cost. In this paper, a PC based automatic 3D model generator was developed. The system was composed of three modules. The first module was the object handling and image acquisition module, which fed and sliced the object sequentially and maintains the paraffine cool to be in solid state and captures the sectional image consecutively. The second one was the system control and interface module, which controls actuators for feeding, slicing, and image capturing. And the last was the image processing and visualization module, which processed a series of acquired sectional images and generated 3D volumetric model. Handling module was composed of the gripper, which grasped and fed the object and the cutting device, which cuts the object by moving cutting edge forward and backward. sliced sectional images were acquired and saved in a form of bitmap file. 2D sectional image files were segmented from the background paraffine and utilized to generate the 3D model. Once 3-D model was constructed on the computer, user could manipulated it with various transformation methods such as translation, rotation, scaling including arbitrary sectional view.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2003.06a
• /
• pp.1055-1058
• /
• 2003

The robustness and reliability of vision algorithms is the key issue in robotic research and industrial applications. In this paper robust real time visual tracking in complex scene is considered. A common approach to increase robustness of a tracking system is the use of different model (CAD model etc.) known a priori. Also fusion or multiple features facilitates robust detection and tracking of objects in scenes of realistic complexity. Voting-based fusion of cues is adapted. In voting. a very simple or no model is used for fusion. The approach for this algorithm is tested in a 3D Cartesian robot which tracks a toy vehicle moving along 3D rail, and the Kalman filter is used to estimate the motion parameters. namely the system state vector of moving object with unknown dynamics. Experimental results show that fusion of cues and motion estimation in a tracking system has a robust performance.

• The Journal of Korea Robotics Society
• /
• v.18 no.3
• /
• pp.346-351
• /
• 2023

With the release of numerous open driving datasets, the demand for domain adaptation in perception tasks has increased, particularly when transferring knowledge from rich datasets to novel domains. However, it is difficult to solve the change 1) in the sensor domain caused by heterogeneous LiDAR sensors and 2) in the environmental domain caused by different environmental factors. We overcome domain differences in the semi-supervised setting with 3-stage model parameter training. First, we pre-train the model with the source dataset with object scaling based on statistics of the object size. Then we fine-tine the partially frozen model weights with copy-and-paste augmentation. The 3D points in the box labels are copied from one scene and pasted to the other scenes. Finally, we use the knowledge distillation method to update the student network with a moving average from the teacher network along with a self-training method with pseudo labels. Test-Time Augmentation with varying z values is employed to predict the final results. Our method achieved 3rd place in ECCV 2022 workshop on the 3D Perception for Autonomous Driving challenge.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.9
• /
• pp.1171-1180
• /
• 1995

In this paper, we present a 3-D object recognition system in which the 3-D Hough transform domain is employed to represent the 3-D objects. In object modeling step, the features for recognition are extracted from the CAD models of objects to be recognized. Since the approach is based on the CAD models, the accuracy and flexibility are greatly improved. In matching stage, the sensed image is compared with the stored model, which is assumed to yield a distortion (location and orientation) in the 3-D Hough transform domain. The high dimensional (6-D) parameter space, which defines the distortion, is decomposed into the low dimensional space for an efficient recognition. At first we decompose the distortion parameter into the rotation parameter and the translation parameter, and the rotation parameter is further decomposed into the viewing direction and the rotational angle. Since we use the 3-D Hough transform domain of the input images directly, the sensitivity to the noise and the high computational complexity could be significantly alleviated. The results show that the proposed 3-D object recognition system provides a satisfactory performance on the real range images.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.2
• /
• pp.93-100
• /
• 2014

3D object recognition using only 2D images is a difficult work because each images are generated different to according to the view direction of cameras. Because SIFT algorithm defines the local features of the projected images, recognition result is particularly limited in case of input images with strong perspective transformation. In this paper, we propose the object recognition method that improves SIFT algorithm by using several sequential images captured from rotating 3D object around a rotation axis. We use the geometric relationship between adjacent images and merge several images into a generated feature space during recognizing object. To clarify effectiveness of the proposed algorithm, we keep constantly the camera position and illumination conditions. This method can recognize the appearance of 3D objects that previous approach can not recognize with usually SIFT algorithm.

• Journal of the Korean Society for Precision Engineering
• /
• v.32 no.7
• /
• pp.633-641
• /
• 2015

In this study, we proposed a three-dimensional (3D) scanning system based on laser-vision technique and rotary mechanism for automatic 3D model reconstruction. The proposed scanning system consists of a laser projector, a camera, and a turntable. For laser-camera calibration a new and simple method was proposed. 3D point cloud data of the surface of scanned object was fully collected by integrating extracted laser profiles, which were extracted from laser stripe images, corresponding to rotary angles of the rotary mechanism. The obscured laser profile problem was also solved by adding an addition camera at another viewpoint. From collected 3D point cloud data, the 3D model of the scanned object was reconstructed based on facet-representation. The reconstructed 3D models showed effectiveness and the applicability of the proposed 3D scanning system to 3D model-based applications.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.8 no.6
• /
• pp.26-34
• /
• 1999

This study shows an alternative method for the determination of object's position, based on a computer vision method. This approach develops the vision system model to define the reciprocal relationship between the 3-D real space and 2-D image plane. The developed model involves the bilinear six-view parameters, which is estimated using the relationship between the camera space location and real coordinates of known position. Based on estimated parameters in independent cameras, the position of unknown object is accomplished using a sequential estimation scheme that permits data of unknown points in each of the 2-D image plane of cameras. This vision control methods the robust and reliable, which overcomes the difficulties of the conventional research such as precise calibration of the vision sensor, exact kinematic modeling of the robot, and correct knowledge of the relative positions and orientation of the robot and CCD camera. Finally, the developed vision control method is tested experimentally by performing determination of object position in the space using computer vision system. These results show the presented method is precise and compatible.

• International conference on construction engineering and project management
• /
• 2009.05a
• /
• pp.1468-1473
• /
• 2009

Efficient communication of construction information has been critical for successful project performance. Building Information Modeling (BIM) has appeared as a tool for efficient communication. Through 3D CAD objects, it is possible to check interception and collisions of each object in advance. In addition, 4D simulation based on 3D objects integrated with time information makes it realize to go over scheduling and to perceive potential errors in scheduling. However, current scheduling simulation is still at a stage of animation due to manual integration of 3D objects and scheduling data. Accordingly, this study aims to develop an integrated model of 3D CAD objects that automatically creates scheduling information.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.606-611
• /
• 2009

This paper describes a method for achieving a novel design within a class of 3D objects that would create a preferred impression on users. Physical parameters of the 3D objects that might strongly contribute to their visual impressions are sought through computational investigation of the impression ratings obtained for learning samples. "Car body" was selected as the class of 3D objects to be investigated. A morphable 3D model of car bodies that describes the variations in appearance using a smaller number of parameters was obtained. Based on each car body's rating for the impression of speediness obtained by paired comparison, the visual impression was transformed by manipulating the parameters defined in the morphable 3D model. The validity of the proposed method was confirmed by psychological experiments. A new scheme is also proposed to properly re-sample a novel object of a peculiar shape so that such an object could also be represented by the morphable 3D model.

• Journal of Broadcast Engineering
• /
• v.28 no.3
• /
• pp.329-332
• /
• 2023

3D Object Detection using deep neural network has been developed a lot for obstacle detection in autonomous vehicles because it can recognize not only the class of target object but also the distance from the object. But in the case of 3D Object Detection models, the detection performance for distant objects is lower than that for nearby objects, which is a critical issue for autonomous vehicles. In this paper, we introduce a technique that increases the performance of 3D object detection models, particularly in recognizing distant objects, by generating virtual 3D vehicle data and adding it to the dataset used for model training. We used a spherical point tracing method that leverages the characteristics of 3D LiDAR sensor data to create virtual vehicles that closely resemble real ones, and we demonstrated the validity of the virtual data by using it to improve recognition performance for objects at all distances in model training.