• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.244-248
• /
• 2020

본 논문에서는 초고속 카메라 기반 투구 궤적 분석 시스템 자동화를 위한 투구 유무 판단 방법을 제안한다. 기존의 카메라 기반 투구 궤적 분석 시스템에서는 궤적 분석을 위한 투구 영상을 수동으로 확보해야 하는 한계가 있다. 이를 해결하기 위해 투수가 던진 공이 포수를 향해 등가속도 운동을 한다는 특성을 이용하여 자동으로 영상 시퀀스 내 투구 유무를 판단한다. 먼저 프레임 별로 야구공 위치를 추정하고, 추정된 공 위치들을 사용해 공 궤적을 모델링한다. 이후 모델링 된 각 궤적 별로 투구 시작 위치와 종료 위치를 획득하고, 시작 위치와 종료 위치에 대한 편차 값을 기준으로 투구 유무를 결정한다. 제안하는 방법의 정확도를 측정하기 위해 실제 야구 경기를 촬영한 영상을 이용하여 실험하였고, 영상 내 모든 투구를 정확하게 판단했음을 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.456-458
• /
• 2023

원본 이미지를 변형하여 학습용 데이터를 확장하는 기법에 대해서는 이전부터 꾸준히 논의된 바가 있다. 턴 테이블과 크로마 키를 이용하여 객체의 영상을 촬영하고 프레임을 추출하여 이미지 분류, 영상 내 객체 탐지 등에 사용이 가능한 데이터 셋의 확장 구축 방안에 대해 다루며, 성능 분석 결과 평균 90% 이상의 객체 검출률을 보였으며 객체 탐지 모델의 경우에서 보다 높은 정확도를 보임을 확인할 수 있었다. 영상내 단일 객체를 인지하기 위한 상황을 위해 본 논문이 제시하는 데이터셋 구축 방안은 충분한 효과를 보일 수 있을 것으로 기대된다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.18-19
• /
• 2024

High Dynamic Range 는 디지털 카메라에 내장된 영상 보정 장치인 Image Signal Processor 의 주요 기능 중 하나로서, 영상의 밝고 어두운 정도의 범위를 넓혀, 피사체가 더 또렷하게 보이도록 한다. 초당 수십 프레임을 촬영하는 경우, 실시간 보정처리를 위해 ISP 에 사용되는 기능 및 알고리즘은 신속성과 효율성이 요구된다. 본 연구는 ISP 에 적합한 HDR 알고리즘을 선정을 목표로 하여, Histogram Equalization 과 Contrast Limited Adaptive Histogram Equalization 을 소개한다. 이어 해당 알고리즘들을 컴퓨터 프로그래밍으로 구현, CMOS 이미지 센서를 통해 추출한 raw image 를 보정하여 각 알고리즘의 성능을 검토하였다.

• Land and Housing Review
• /
• v.14 no.2
• /
• pp.125-135
• /
• 2023

Drones are used in various fields, including land survey, transportation, forestry/agriculture, marine, environment, disaster prevention, water resources, cultural assets, and construction, as their industrial importance and market size have increased. In this study, image data for deep learning was collected using a mavic3 drone capturing images at a shooting altitude was 20 m with ×7 magnification. Swin Transformer and UperNet were employed as the backbone and architecture of the deep learning model. About 800 sheets of labeled data were augmented to increase the amount of data. The learning process encompassed three rounds. The Cross-Entropy loss function was used in the first and second learning; the Tversky loss function was used in the third learning. In the future, when the crack detection model is advanced through convergence with the Internet of Things (IoT) through additional research, it will be possible to detect patching or potholes. In addition, it is expected that real-time detection tasks of drones can quickly secure the detection of pavement maintenance sections.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.44 no.5
• /
• pp.12-25
• /
• 2016

In this study, in order to facilitate analysis in a continuous sequence, including the concept of the landscape experience time, countless frames of a continuous landscape were extracted. The amount of visual elements in each frame was data-converted numerically to take advantage of the quantitative data necessary for landscape planning and design was calculated in the rhythm of the sequence. In Order to shoot video with the flow of the line of sight of experience in landscape districts and landscape control points along the landscape corridor which is a continuous path, each of the corresponding computer motion techniques. This study developed a CRVP Koo computer program to effectively calculate the continuous visual number of specific landscape components by extracting uncounted frames at regular intervals, and after verifying, attempting to apply this to the target site. Through the applied result, it was possible to extract the digitized quantitative rhythm for each component of each landscape, the margin of error is very small when compared with the results of manual in photoshop, it was able to overcome the drawbacks of the manual. Using the rhythm of the derived sequence, and those close to the experience of the landscape, it was possible to achieve quantitative analysis derived from a variety of perspectives as well as was possible to be used as quantitative basis data and analysis technique for landscape planning and design.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.7
• /
• pp.29-36
• /
• 2008

This paper proposes algorithm for the system that can search for an image of a frame unit, and we implement it. A system already inserts in images after generating the cord that mechanical decoding and identification are possible. We are independent of an external noise in a frame unit, and a system to propose at these papers can search for an image recorded by search condition to include recording date, recording time, a recording place or filming course etc. This system is composed by image insertion wealth to insert data to an image to data image code generation wealth, a frame generating data image code you apply a code generation rule to be fixed in order to express to a price to have continued like data entry wealth, GPS locator values and direction price receiving an image signal, image decoding signals and an image search signal to include search condition, and to have continuity from users each of an image. Also, image decoding we decipher about the noise that was already added from the outsides in a telerecording process, a copy process or storage processes inserted in images by real time, and searching image information by search condition. Consequently we implement decoder, and provide the early system that you use, and we easily insert data code among images. and we can search. and maximization can get precision regarding an image search and use satisfaction as we use algorithm to propose at these papers.

• Journal of Broadcast Engineering
• /
• v.26 no.6
• /
• pp.680-691
• /
• 2021

Due to the recent developments of attaining 3D contents by using devices such as 3D scanners, the diversity of the contents being used in AR(Augmented Reality)/VR(Virutal Reality) fields is significantly increasing. There are several ways to represent 3D data, and using point clouds is one of them. A point cloud is a cluster of points, having the advantage of being able to attain actual 3D data with high precision. However, in order to express 3D contents, much more data is required compared to that of 2D images. The size of data needed to represent dynamic 3D point cloud objects that consists of multiple frames is especially big, and that is why an efficient compression technology for this kind of data must be developed. In this paper, a motion estimation and compensation method for dynamic point cloud objects using 3D DCT is proposed. This will lead to switching the 3D video frames into I frames and P frames, which ensures higher compression ratio. Then, we confirm the compression efficiency of the proposed technology by comparing it with the anchor technology, an Intra-frame based compression method, and 2D-DCT based V-PCC.

• Journal of the Korea Computer Graphics Society
• /
• v.24 no.1
• /
• pp.9-16
• /
• 2018

In this paper, we introduce a visualization framework for cell image data obtained from optical diffraction tomography (ODT), including a method for representing cell morphology in 3D virtual environment and a color mapping protocol. Unlike commonly known volume data sets, such as CT images of human organ or industrial machinery, that have solid structural information, the cell image data have rather vague information with much morphological variations on the boundaries. Therefore, it is difficult to come up with consistent representation of cell structure for visualization results. To obtain desired visual representation of cellular structures, we propose an interactive visualization technique for the ODT data. In visualization of 3D shape of the cell, we adopt a volume rendering technique which is generally applied to volume data visualization and improve the quality of volume rendering result by using empty space jittering method. Furthermore, we provide a layer-based independent rendering method for multiple transfer functions to represent two or more cellular structures in unified render window. In the experiment, we examined effectiveness of proposed method by visualizing various type of the cell obtained from the microscope which can capture ODT image and fluorescence image together.

• Journal of Intelligence and Information Systems
• /
• v.15 no.2
• /
• pp.33-48
• /
• 2009

Capsule endoscopy is one of the most remarkable inventions in last ten years. Causing less pain for patients, diagnosis for entire digestive system has been considered as a most convenience method over a normal endoscope. However, it is known that the diagnosis process typically requires very long inspection time for clinical experts because of considerably many duplicate images of same areas in human digestive system due to uncontrollable movement of a capsule endoscope. In this paper, we propose a method for clinical diagnosticians to get highly valuable information from capsule-endoscopy video. Our software system consists of three global maps, such as movement map, characteristic map, and brightness map, in temporal domain for entire sequence of the input video. The movement map can be used for effectively removing duplicated adjacent images. The characteristic and brightness maps provide frame content analyses that can be quickly used for segmenting regions or locating some features(such as blood) in the stream. Our experiments show the results of four patients having different health conditions. The result maps clearly capture the movements and characteristics from the image frames. Our method may help the diagnosticians quickly search the locations of lesion, bleeding, or some other interesting areas.

• Journal of Broadcast Engineering
• /
• v.22 no.4
• /
• pp.448-461
• /
• 2017

This paper proposes a vehicle counting and speed estimation method for traffic situation analysis in road CCTV videos. The proposed method removes a distortion in the images using Inverse perspective Mapping, and obtains specific region for vehicle counting and speed estimation using lane detection algorithm. Then, we can obtain vehicle counting and speed estimation results from using optical flow at specific region. The proposed method achieves stable accuracy of 88.94% from several CCTV images by regional groups and it totally applied at 106,993 frames, about 3 hours video.