• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.754-756
• /
• 2005

본 논문은 연속 초음파 영상에서 움직임 보상 시간 필터를 적용하여 영상의 품질을 향상시키는 방법을 제안한다. 비선형 광학 흐름 추정 방법을 이용하여 화소 단위의 움직임을 추정하고, 이를 바탕으로 시간적 재귀 필터링을 적용한다 화소 단위 움직임 벡터의 양이 작을 경우 필터링을 크게 하고, 움직임 벡터의 양이 클 경우 필터링을 작게 적용한다. 그 결과 프로브에 의한 전역적 움직임과 측정 대상물에 의한 국부적 움직임으로 발생되는 블러 현상을 극소화하고 잡음을 감소시켜 영상의 품질을 향상시켰다.

• Journal of Korea Multimedia Society
• /
• v.22 no.10
• /
• pp.1149-1159
• /
• 2019

We present a novel deep learning architecture for obtaining a latent image from a single blurry image, which contains dynamic motion blurs through object/camera movements. The proposed architecture consists of two sub-modules: blur image restoration and optical flow estimation. The tasks are highly related in that object/camera movements make cause blurry artifacts, whereas they are estimated through optical flow. The ablation study demonstrates that training multi-task architecture simultaneously improves both tasks compared to handling them separately. Objective and subjective evaluations show that our method outperforms the state-of-the-arts deep learning based techniques.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.464-465
• /
• 2012

장면에는 초점을 두는 중요한 영역이 있다. 초점에 의한 영상 효과는 사실감 뿐 아니라 작가들의 매시지를 효과적으로 전달하는데 현저히 도움이 된다. 본 논문에서는 영화 영상 기법 중 달리줌을 컴퓨터 그래픽스에 적용/렌더링하여 초점 효과를 극적으로 향상시키는 방법에 대하여 제안한다. 달리줌과 더불어, thin-lens 카메라 모델 기반 디포커스 블러를 추가하여, 보다 극적인 효과를 실시간에 얻을 수 있다. 이러한 효과는 역동적인 원근감을 제공하여 물체를 강조하는 다양한 특수효과에 쓰일 수 있다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.04a
• /
• pp.169-170
• /
• 2009

영상 해상도 향상 기술은 영상 처리의 많은 분야에서 사용되는 전처리 기술로서, 최근들어 감시 카메라 시스템에서의 영상 해상도 향상을 위한 연구가 진행되고 있다. 보간 과정에서의 블러링으로 인한 화질 저하를 해결하기 위해서, 본 논문에서는 하위 레벨 보간을 이용한 에러 추정과 영상 해상도 향상방법을 제안한다. 제안하는 방법에서는 하위 레벨 보간을 통해서 보간 과정에서 발생하는 손실 정보를 추정하고, 추정한 손실 정보를 보간 결과에 적용하여 영상 복원의 결과를 향상시킨다. 동일한 영상을 이용한 실험을 통해서 기존의 방법들보다 0.38~1.75dB의 객관적 화질의 개선을 확인하였고 주관적 화질 비교에서도 향상되었음을 확인하였다. 제안하는 방법은 감시 카메라 시스템을 비롯한 영상 확대를 위한 응용 환경에서 활용될 수 있다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.343-344
• /
• 2021

최근 영상의 활용도의 증가에 따라, 비정형 영상 데이터에 대한 양이 기하급수적으로 증가하였다. 디지털 영상을 획득할 시에 처리/압축/저장/전송/재생산 등의 과정을 거치면서 왜곡을 수반하게 되며 영상의 품질을 저하시키는 요인이 된다. 영상의 품질은 활용 결과에도 큰 영향을 미치기 때문에 품질이 저하된 영상은 분류를 하는 것이 중요하다. 하지만 사람이 수신된 모든 영상에 대해서 직접 분류를 하는 것은 많은 시간과 비용이 소요된다는 문제점이 존재한다. 따라서 본 논문에서는 사람이 인지하는 주관적인 영상 품질 평가와 유사하게 품질에 대한 평가를 위한 왜곡영상의 특징정보를 검출 및 분석하는 방안에 대해서 제안한다. 본 방법은 사람이 영상을 인지할 때 가장 많이 사용되는 요소인 색상에 대한 선명도, 블러와 노이즈에 대한 특징정보를 이용한다. 검출된 특징정보를 공간 도메인으로 변환함으로써 왜곡 영상별 특성을 분석하였다. 실험을 위해서 IQA 데이터베이스인 LIVE를 이용하였으며, 원본영상 및 5가지 유형의 왜곡영상으로 구성되어 있다. 실험결과 품질이 좋은 영상과 왜곡영상에 대한 특성을 검출할 수 있었다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.637-639
• /
• 2021

As autonomous driving research is actively progressing, lane detection is an essential technology in ADAS (Advanced Driver Assistance System) to locate a vehicle and maintain a route. Lane detection is detected using an image processing algorithm such as Hough transform and RANSAC (Random Sample Consensus). This paper implements a linear shape detection algorithm using OpenCV on Raspberry Pi 3 B+. Thresholds were set through OpenCV Gaussian blur structure and Canny edge detection, and lane recognition was successful through linear detection algorithm.

• The Journal of the Korea Contents Association
• /
• v.4 no.4
• /
• pp.189-196
• /
• 2004

Most of the recent researches for ATM switches have been based on multistage interconnection network known as regularity and self-routing property. These networks can switch packets simultaneously and in parallel. However, they are blocking networks in the sense that packet is capable of collision with each other Mainly Banyan network have been used for structure. There are several ways to reduce the blocking or to increase the throughput of banyan-type switches: increasing the internal link speeds, placing buffers in each switching node, using multiple path, distributing the load evenly in front of the banyan network and so on. Therefore, this paper proposes the use of recirculating shuffle-exchange network to reduce the blocking and to improve hardware complexity. This structures are recirculating shuffle-exchange network as simplified in hardware complexity and Rank network with tree structure which send only a packet with highest priority to the next network, and recirculate the others to the previous network. after it decides priority number on the Packets transferred to the same destination, The transferred Packets into banyan network use the function of self routing through decomposition and composition algorithm and all they arrive at final destinations. To analyze throughput, waiting time and packet loss ratio according to the size of buffer, the probabilities are modeled by a binomial distribution of packet arrival. If it is 50 percentage of load, the size of buffer is more than 15. It means the acceptable packet loss ratio. Therefore, this paper simplify the hardware complexity as use of recirculating shuffle-exchange network instead of bitonic sorter.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.4
• /
• pp.77-84
• /
• 2012

Face classification has wide applications in intelligent video surveillance, content retrieval, robot vision, and human-machine interface. Pose and expression changes, and arbitrary illumination are typical problems for face recognition. When the face is captured at a distance, the image quality is often degraded by blurring and noise corruption. This paper investigates the efficacy of multi-classifier decision level fusion for face classification based on the photon-counting linear discriminant analysis with two different cost functions: Euclidean distance and negative normalized correlation. Decision level fusion comprises three stages: cost normalization, cost validation, and fusion rules. First, the costs are normalized into the uniform range and then, candidate costs are selected during validation. Three fusion rules are employed: minimum, average, and majority-voting rules. In the experiments, unfocusing and motion blurs are rendered to simulate the effects of the long distance environments. It will be shown that the decision-level fusion scheme provides better results than the single classifier.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.49 no.11
• /
• pp.47-55
• /
• 2012

In this paper, we propose a method to reconstruct accurate 3D shape object by using multi-view images which are disturbed by motion blur. In multi-view deblurring, more precise PSF estimation can be done by using the geometric relationship between multi-view images. The proposed method first estimates initial 2D PSFs from individual input images. Then 3D PSF candidates are projected on the input images one by one to find the best one which are mostly consistent with the initial 2D PSFs. 3D PSF consists with direction and density and it represents the 3D trajectory of object motion. 야to restore 3D shape by using multi-view images computes the similarity map and estimates the position of 3D point. The estimated 3D PSF is again projected to input images and they replaces the intial 2D PSFs which are finally used in image deblurring. Experimental result shows that the quality of image deblurring and 3D reconstruction improves significantly compared with the result when the input images are independently deblurred.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.85-92
• /
• 2023

In this paper, we propose a method for automatically generating story-based abbreviated summaries from long-form dramas and movies. From the shooting stage, the basic premise was to compose a frame with illusion of depth considering the golden division as well as focus on the object of interest to focus the viewer's attention in terms of content delivery. To consider how to extract the appropriate frames for this purpose, we utilized elemental techniques that have been utilized in previous work on scene and shot detection, as well as work on identifying focus-related blur. After converting the videos shared on YouTube to frame-by-frame, we divided them into a entire frame and three partial regions for feature extraction, and calculated the results of applying Laplacian operator and FFT to each region to choose the FFT with relative consistency and robustness. By comparing the calculated values for the entire frame with the calculated values for the three regions, the target frames were selected based on the condition that relatively sharp regions could be identified. Based on the selected results, the final frames were extracted by combining the results of an offline change point detection method to ensure the continuity of the frames within the shot, and an edit decision list was constructed to produce an abbreviated summary of 62.77% of the footage with F1-Score of 75.9%