• Journal of the Korea Society of Computer and Information
• /
• v.29 no.8
• /
• pp.23-32
• /
• 2024

NVS (Novel View Synthesis) is a field in computer vision that reconstructs new views of a scene from a set of input views. Real-time rendering and high performance are essential for NVS technology to be effectively utilized in various applications. Recently, 3D-GS (3D Gaussian Splatting) has gained popularity due to its faster training and inference times compared to those of NeRF (Neural Radiance Fields)-based methodologies. However, since 3D-GS reconstructs a 3D (Three-Dimensional) scene by splitting and cloning (Density Control) Gaussian points, the number of Gaussian points continuously increases, causing the model to become heavier as training progresses. To address this issue, we propose two methodologies: 1) Gaussian blending, an improved density control methodology that removes unnecessary Gaussian points, and 2) a performance enhancement methodology using a depth estimation model to minimize the loss in representation caused by the blending of Gaussian points. Experiments on the Tanks and Temples Dataset show that the proposed methodologies reduce the number of Gaussian points by up to 4% while maintaining performance.

• Journal of Sensor Science and Technology
• /
• v.7 no.2
• /
• pp.124-130
• /
• 1998

High-speed 3D vision systems are essential for autonomous robot or vehicle control applications. In our study, a stereo vision process has been developed. It consists of three steps : extraction of edges in right and left images, matching corresponding edges and calculation of the 3D map. This process is implemented in a VME 150/40 Imaging Technology vision system. It is a modular system composed by a display, an acquisition, a four Mbytes image frame memory, and three computational cards. Programmable accelerator computational modules are running at 40 MHz and are based on TMS320C31 DSP with a $64{\times}32$ bit instruction cache and two $1024{\times}32$ bit internal RAMs. Each is equipped with 512 Kbytes static RAM, 4 Mbytes image memory, 1 Mbytes flash EEPROM and a serial port. Data transfers and communications between modules are provided by three 8 bit global video bus, and three local configurable pipeline 8 bit video bus. The VME bus is dedicated to system management. Tasks between DSPs are distributed as follows: two DSPs are used to edges detection, one for the right image and the other for the left one. The last processor computes the matching process and the 3D calculation. With $512{\times}512$ pixels images, this sensor generates dense 3D maps at a rate of about 1 Hz depending of the scene complexity. Results can surely be improved by using a special suited multiprocessors cards.

• International Journal of Advanced Culture Technology
• /
• v.11 no.1
• /
• pp.389-394
• /
• 2023

Based on intelligent visual information and 5G, this paper studies the intelligent visual communication of landscape bamboo buildings, and provides a new method of intelligent perception and interactive computing for the real world, which can represent, model, Perception and cognition; through the integration of virtual and real, the situational understanding of the human-machine-material fusion environment and the interaction with nature. The 5G network can well meet the combination of high-bandwidth uplink transmission and low-latency downlink control. At the same time, 5G-based AR intelligent inspection, remote operation and maintenance guidance, and machine vision inspection. Taking the bamboo building as an example, through field inspections to analyze tourism Bamboo buildings before and after development, and the intelligentization of bamboo buildings based on 5G and visual modeling.

• Journal of Korea Multimedia Society
• /
• v.24 no.2
• /
• pp.267-273
• /
• 2021

Recently, the small-size mobile robots with remote-control are rapidly growth which market of mobile is increased in the world. Especially, the smart-phones are widely used for interface device in the small size of a mobile robot. The research goal is control system design which is applied to miniaturization of a mobile robot using smart-phone and it can be confirmed performance for designed system. Meanwhile, the fabrication of mini-mobile robot can also be remote-control operation through the WIFI performance of a smart-phone. The smart-phone is used to remote-control for robot operation which control data transmit to robot via the WIFI network. To drive the robot, we can observe by the smart-phone screen and it can easily adjust the robot drive condition and direction by smart-phone button. Consequentially, there was no malfunction and images were printed out well. However, in drive, because of blind spot, robot was bumped into obstacle. Therefore, the additional test is necessary to sensor for blind spot which sensor can be equipment to mobile robot. In addition, the experiment with robot object recognition is needed.

• 제어로봇시스템학회:학술대회논문집
• /
• 1988.10b
• /
• pp.810-815
• /
• 1988

This paper presents a mobile robot that travels employing visual information. The mobile robot is equipped solely with a TV camera as a sensor, and views from the TV camera are transferred to a separately installed micro computer through an image acquisition device. An acquired image of a view is processed there and the information necessary for travel is yielded. Instructions based on the information are then sent from the micro computer to the mobile robot, which causes the mobile robot next action. Among several application programs that have already been developed for the mobile robot other than the entire control program, this paper focuses its attention on the travelling control of the mobile robot in a model environment with obstructions as well as an overview of the whole system. The behaviour the present mobile robot takes when it travels among obstructions was investigated by an experiment, and satisfactory results were obtained.

• Journal of Institute of Control, Robotics and Systems
• /
• v.11 no.4
• /
• pp.304-313
• /
• 2005

The robots that will be needed in the near future are human-friendly robots that are able to coexist with humans and support humans effectively. To realize this, robot need to recognize his position and direction for intelligent performance in an unknown environment. And the mobile robots may navigate by means of a number of monitoring systems such as the sonar-sensing system or the visual-sensing system. Notice that in the conventional fusion schemes, the measurement is dependent on the current data sets only. Therefore, more of sensors are required to measure a certain physical parameter or to improve the accuracy of the measurement. However, in this research, instead of adding more sensors to the system, the temporal sequence of the data sets are stored and utilized for the accurate measurement. As a general approach of sensor fusion, a UT -Based Sensor Fusion(UTSF) scheme using Unscented Transformation(UT) is proposed for either joint or disjoint data structure and applied to the landmark identification for mobile robot navigation. Theoretical basis is illustrated by examples and the effectiveness is proved through the simulations and experiments. The newly proposed, UT-Based UTSF scheme is applied to the navigation of a mobile robot in an unstructured environment as well as structured environment, and its performance is verified by the computer simulation and the experiment.

• Proceedings of the IEEK Conference
• /
• 2003.07c
• /
• pp.2577-2580
• /
• 2003

In this paper we present a teleoperation system of an internet-based mobile robot based on H.263 image compression algorithm. We control remotely the networked mobile robot with vision over the internet under unknown environments in the real time. The main feature of this system is that local operators need to download and install a client program and so they can command the robot in a remote location. Image information of remote location is compressed by H.263 video codec algorithm and then transmitted to a local operator. We constructed and tested a networked autonomous mobile robot in the real field. The experimental results show the usefulness of the robot under test.

• Journal of Multimedia Information System
• /
• v.7 no.3
• /
• pp.197-204
• /
• 2020

Superpixel methods are widely used in the preprocessing stage as a method to reduce computational complexity by simplifying images while maintaining the characteristics of the images in the computer vision applications. It is common to generate superpixels of similar size and shape based on the pixel values rather than considering the characteristics of the image. In this paper, we propose a method to control the sizes and shapes of generated superpixels, considering the contents of an image. The proposed method consists of two steps. The first step is to over-segment an image so that the boundary information of the image is well preserved. In the second step, generated superpixels are merged based on similarity to produce the target number of superpixels, where the shapes of superpixels are controlled by limiting the maximum size and the proposed roundness metric. Experimental results show that the proposed method preserves the boundaries of the objects in an image more accurately than the existing method.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.2435-2440
• /
• 2003

Manipulation of the nano/micro scale object has been a key technology in biology as the sizes of DNA, chromosome, nucleus, cell and embryo are within such order. For instance, for embryo cell manipulation, the cell injection is performed manually. The operator often spends over a year to carry out a cell manipulation project. Since the typical success rate of such operation is extremely low, automation of such biological cell manipulation has been asked. As the operator spends most of his time in finding the position of cell in the Petri dish and in injecting bio-material to the cell from the best orientation. In this paper, we propose a new strategy and a vision system, by which one can find, recognize and track nucleus, polar body, and zona pellucida of the embryo cell for automatic biomanipulation. The deformable template matching algorithm has been used in recognizing the nucleus and polar body of each cell. Result suggests that it outperforms the conventional methods.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.746-750
• /
• 2005

This paper describes image processing techniques that can detect, segment, and analyze the mango's physical properties such as size, shape, surface area, and color from images. First, images of mangoes taken by a digital camera are analyzed and segmented. The segmentation is done based on constructed hue model of the sample mangoes. Some morphological and filtering techniques are then applied to clean noises before fitting spline curve on the mango boundary. From the clean segmented image, the mango projected area can be computed. The shape of the mango is then analyzed using some structuring models. Color is also spatially analyzed and indexed in the database for future classification. To obtain the surface area, the mango is peeled. The scanned image of its peels is then segmented and filtered using similar approach. With calibration parameters, the surface area could then be computed. We employed the system to evaluate physical properties of a mango cultivar called "Nam Dokmai". There were sixty mango samples in three various sizes graded by an experienced farmer's eyes and hands. The results show the techniques could be a good alternative and more feasible method for grading mango comparing to human's manual grading.