• Journal of Korea Society of Industrial Information Systems
• /
• v.22 no.3
• /
• pp.27-38
• /
• 2017

This Paper Presents a Method for Stitching a Large Number of Images Recorded by a Large Number of Individual Users Through a Cellular Phone Camera at a Venue. In Contrast to 360 Camera Solutions that Use Existing Fixed Rigs, these Conditions must Address New Challenges Such as Time Synchronization, Repeated Transformation Matrix Calculations, and Camera Sensor Mismatch Correction. In this Paper, we Solve this Problem by Updating the Transformation Matrix Using Time Synchronization Method Using Audio, Sensor Mismatch Removal by Color Transfer Method, and Global Operation Stabilization Algorithm. Experimental Results Show that the Proposed Algorithm Shows better Performance in Terms of Computation Speed and Subjective Image Quality than that of Screen Stitching.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.41 no.5
• /
• pp.33-40
• /
• 2004

In this paper, we constructs face database by using wavelet comparison, and compare face recognition rate by using principle component analysis (Principal Component Analysis : PCA) algorithm. General face recognition method constructs database, and do face recognition by using normalized size. Proposed method changes image of normalized size (92${\times}$112) to 1 step, 2 step, 3 steps to wavelet compression and construct database. Input image did compression by wavelet and a face recognition experiment by PCA algorithm. As well as method that is proposed through an experiment reduces existing face image's information, the processing speed improved. Also, original image of proposed method showed recognition rate about 99.05%, 1 step 99.05%, 2 step 98.93%, 3 steps 98.54%, and showed that is possible to do face recognition constructing face database of large quantity.

• Journal of Broadcast Engineering
• /
• v.2 no.1
• /
• pp.45-55
• /
• 1997

In this paper, we propose a fast algorithm with adaptive thresholding for the wavelet transform (WT) based blocking artifact reduction. In the fast algorithm, all processings that are equivalent to the processing in WT domain of the first and second scale are performed in spatial domain. In the adaptive thresholding, the threshold values used to classify the block boundary are selected adaptively according to each input image by using the statistical properties of the WT of the coded signal at block boundary and at block center, which can be obtained in spatial domain. Experimental results showed that the proposed fast algorithm is about 10 times faster than the WT-based algorithm. It also was found that the postprocessing with proposed adaptive thresholding yields some PSNR improvement and better subjective quality over that with nonadaptive thresholding which has best performance at high compression ratios of a certain .image, even at low compression ratios.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.1-8
• /
• 2024

On-device AI technology, which can operate AI models at the edge devices to support real-time processing and privacy enhancement, is attracting attention. As intelligent IoT is applied to various industries, services utilizing the on-device AI technology are increasing significantly. However, general deep learning models require a lot of computational resources for inference and learning. Therefore, various lightweighting methods such as quantization and pruning have been suggested to operate deep learning models in embedded edge devices. Among the lightweighting methods, we analyze how to lightweight and apply deep learning models to edge computing devices, focusing on pruning technology in this paper. In particular, we utilize dynamic and static pruning techniques to evaluate the inference speed, accuracy, and memory usage of a lightweight AI vision model. The content analyzed in this paper can be used for intelligent video control systems or video security systems in autonomous vehicles, where real-time processing are highly required. In addition, it is expected that the content can be used more effectively in various IoT services and industries.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.228-231
• /
• 2022

Most CCTVs operated by public institutions for crime prevention and parking enforcement are located on roads. The angle of these CCTV's view is often changed for various reasons, such as bolt loosening by vibration or shocking by vehicles and workers, etc. In order to effectively provide AI services based on the collected images, the service target area(ROI, Region Of Interest) must be provided without interruption within the image. This is also related to the viewpoint of effective operation of computing power for image analysis. This study explains how to maximize the application of artificial intelligence technology by setting the ROI based on the marker on the road, setting the image analysis to be possible only within the area, and studying the process of finding the ROI.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.19-22
• /
• 2001

현재 수 많은 공중파 채널을 비롯하여 각종 케이블 TV 와 위성 TV 등의 채널이 대량으로 디지털화 되어 가고 있지만 파국의 디지털 TV 의 규격은 MPEG2 라는 표준 규격으로 통일 되어 우리에게 전해지고 있다. 이런 수많은 동영상 정보들은 디지털이라는 특성상 기록과 보관이 용이하므로 사용자는 수 많은 정보를 저장, 관리 할 수 있게 되었다. 따라서 방대한 데이터의 바다 속에서 원하는 정보를 검색하는데 있어서 기존의 방식 즉 일정 속도를 가진 검색 엔진을 사용하는 것 보다는 MPEG2 규격의 표준들을 보다 효율적으로 이용하여 사용자들이 동영상 정보 중에 이동하기 원하는 부분으로 이동하기 위한 비선형 검색엔진을 소개한다. 이를 이용한 PVR(Penonal Video Recorder)은 현재 hardware 부분은 완료되었고 일부 software 엔진을 적용하여 개발 중이다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.375-377
• /
• 2016

리포터의 짧은 현장 중계를 위해 중계차와 MNG 등 기존 중계시스템을 동원하는 것은 많은 시간과 비용을 요구한다. 최근 스마트폰의 연산 능력, 카메라의 화질, 모바일 데이터 네트워크의 속도, 그리고 스마트폰과 네트워크의 안정성이 비약적으로 발전함에 따라 스마트폰을 이용한 간단한 현장 중계 방송이 가능하게 되었다. 본 논문에서는 이를 위한 스마트폰 애플리케이션, 스튜디오에 위치한 관리서버와 송수신시스템, 스마트폰 영상과 스튜디오 영상의 실시간 부/복호화 모듈, RTP를 이용한 전송 모듈 등의 개발에 대해 기술하고, 개발 중인 시스템을 KBS의 '보이는 라디오' 실시간 중계방송에 활용한 사례에 대해 소개한다.

• Tunnel and Underground Space
• /
• v.11 no.2
• /
• pp.120-121
• /
• 2001

The purpose of this study is to investigate concrete flaw using seismic first arrival and various inversion method. Seismic wave propagation was calculated using finite element method in theoretical modelling and tomogram was made using various inversion methods in theoretical and experimental modelling. Five steps of seismic first arrival were selected from FEM results and these data were used to calculate seismic velocity section. According to the results, exact seismic first arrival picking method was proposed and experimental modelling was conducted.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.9
• /
• pp.1985-1997
• /
• 1997

This paper proposed a regularized iterative image restoration by using method of conjugate gradient. Compared with conventional iterative methods, method of conjugate gradient has a merit to converte toward a solution as a super-linear convergence speed. But because of those properties, there are several artifacts like ringing effects and the partial magnification of the noise in the course of restoring the images that are degraded by a defocusing blur and additive noise. So, we proposed the regularized method of conjugate gradient applying constraints. By applying the projectiong constraint and regularization parameter into that method, it is possible to suppress the magnification of the additive noise. As a experimental results, we showed the superior convergence ratio of the proposed mehtod compared with conventional iterative regularized methods.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.10a
• /
• pp.138-141
• /
• 2004

본 논문은 EBGM(Elastic Bunch Graph Matching)기법을 이용한 얼굴인식에 대해 다룬다. 대용량 영상 정보에 대해 차원 축소를 이용한 얼굴인식 기법인 주성분기법이나 선형판별기법에서는 얼굴 영상 전체의 정보를 이용하는 반면 본 논문에서는 얼굴의 눈, 코, 입 등과 같은 얼굴 특징점에 대해 주파수와 방향각이 다른 여러 개의 가버 커널과 영상 이미지의 컨볼루션(Convolution)의 계수의 집합(Jets)을 이용한 특징 데이터를 이용한다. 하나의 얼굴 영상에 대해서는 모든 영상이 같은 크기의 특징 데이터로 표현되는 Face Graph가 생성되며, 얼굴인식 과정에서는 추출된 제트의 집합에 대해서 상호 유사도(Similarity)의 크기를 비교하여 얼굴인식을 수행한다. 본 논문에서는 기존의 EBGM방법의 Face Graph 생성 과정을 보다 간략화 한 방법을 이용하여 얼굴인식 과정에서 계산량을 줄여 속도를 개선하였다.