• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.49 no.12
• /
• pp.651-658
• /
• 2000

This paper presents a new method to extract traffic information such as number of passing vehicles and average speed by a pair of stereo road images. The whole process consists of the extraction of vehicles and the tracking of the extracted vehicles. For the extraction of vehicles, the outline of each vehicle is obtained by using binary region growing technique applied to disparity map based on multi-resolution stereo matching. The Kalman filter tracking algorithm is applied to the extracted vehicle outlines to determine the flow of vehicles. Experimental results show that the proposed method significantly improved recognition rate of vehicles over the conventional methods-frame difference and background elimination methods.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.891-894
• /
• 2001

본 논문에서는 NTGST (noise·tolerant generalized symmetry transform)와 snake를 이용하여 복잡배경으로부터 여러 물체의 윤곽선을 동시에 검출하는 방법을 제안하였다. 먼저 NTCST의 대칭도 맵(symmetry map)을 이용하여 복잡한 배경에 혼재하는 여러 물체들의 위치를 찾은 다음, 이들 각 물체에 snake의 초기 윤곽들을 자동 설정해 줌으로써 기존 snake 알고리즘의 초기 윤곽 설정의 어려움과 다중 물체 윤곽선 검출의 어려움을 동시에 해결하였다. 이때 NTGST의 대칭도 맵으로부터 설정된 snake의 초기 윤곽은 실제 물체의 윤곽선 가까이에 위치할 뿐만 아니라 물체의 형태를 잘 반영하므로 요철이 있는 물체의 윤곽선도 기존의 방법보다 적은 반복횟수로 정확하게 검출 할 수 있다. 다양한 합성 영상과 실영상에 적용한 결과 복잡배경으로부터도 다중 물체의 윤곽선을 효과적으로 추출함을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.33-34
• /
• 2016

본 논문에서는 HVS(human visual system)의 특성을 고려한 새로운 스케일러블 코딩방법을 제안한다. 제안된 방법은 먼저 영상 내에서 관심영역(saliency map)을 찾고 관심영역을 제외한 부분에 에지 보존 필터를 적용한다. 그 영상은 정해진 양자 파라미터 값으로 인코딩 되어 제안된 코딩 시스템의 베이스 층(base layer)이 된다. 기존 스케일러블 코딩 표준에서의 베이스 층과 다르게 본 논문의 베이스 층은 관심 있는 중요영역(foreground)을 보존하고 또한 배경(background)의 에지 성분도 보존한다. 기본 층이 전송되면 개선층(enhancement layer)은 원 영상과 복원된 베이스 층 영상간의 차분 영상에서 관심영역 순으로 보내진다. 실험은 HEVC 를 바탕으로 수행되었고 스케일러블 코딩 표준인 SHVC 와 관심영역에서 비교를 했을 때 제안된 알고리즘이 더 높은 PSNR 을 가지는 것을 확인하였다. 또한 전체적으로 지각적인 품질(perceptual quality) 또한 향상되었음을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.06a
• /
• pp.215-218
• /
• 2017

본 논문에서는 바둑 대국 동영상에서 배경 영상과의 차이를 이용하여 바둑판 내에서의 움직임을 분석하고, 분석 결과를 이용하여 바둑돌의 착수 위치 및 바둑돌의 종류를 인식하는 자동 바둑 기보 저장 알고리즘을 제안한다. 카메라의 내부 특성이 변하지 않고 렌즈 왜곡이 존재하지 않는다고 가정하였을 때, 바둑판 위에 움직임이 없는 배경 영상과 현재의 영상 간의 차이의 변화량을 블록 단위로 누적한 블록 단위 움직임 맵(Block Motion Map)을 기반으로 움직임의 존재 여부를 판단하고, 착수 후 물체의 움직임이 없어진 영상을 배경 영상으로 갱신하며, 해당 영상과 이전 배경 영상의 패치(Patch)를 이용하여 착수 위치 및 바둑돌의 종류를 인식한다.

• Proceedings of the KOSOMBE Conference
• /
• v.1997 no.05
• /
• pp.95-100
• /
• 1997

In this paper, we propose a coding method to improve compression efficiency for MR image. This can be achieved by combining coding and segmentation scheme which removes noisy background region, which is meaningless for diagnosis, in MR image. The wavelet coder encodes only diagnostically significant foreground regions refering to segmentation map. Our proposed algorithm provides about 15% of bitrate reduction when compared with the same coder which is not combined with segmentation scheme. And the proposed scheme shows better reconstructed image Qualify than JPEG at the same compression ratio.

• Journal of Industrial Technology
• /
• v.28 no.B
• /
• pp.143-148
• /
• 2008

A real-time face detection is to find human faces robustly under the cluttered background free from the effect of occlusion by other objects or various lightening conditions. We propose a face detection system for real-time applications using wavelet decomposition method based on Gabor features. Firstly, skin candidate regions are extracted from the given image by skin color filtering and projection method. Then Gabor-feature based template matching is performed to choose face cadidate from the skin candidate regions. The chosen face candidate region is transformed into 2-level wavelet decomposition images, from which feature vectors are extracted for classification. Based on the extracted feature vectors, the face candidate region is finally classified into either face or nonface class by the Levenberg-Marguardt back-propagation neural network.

• 한국HCI학회:학술대회논문집
• /
• 2007.02a
• /
• pp.630-635
• /
• 2007

In this paper, we propose a system for multiple people tracking using fragment based histogram matching. Appearance model is based on IHLS color histogram which can be calculated efficiently using integral histogram representation. Since histograms will loss all spatial information, we define a fragment based region representation which retain spatial information, robust against occlusion and scale issue by using disparity information. Multiple people labeling is maintained by creating online appearance representation for each people detected in scene and calculating fragment vote map. Initialization is performed automatically from background segmentation step.

• Proceedings of the IEEK Conference
• /
• 2002.07b
• /
• pp.904-907
• /
• 2002

In this paper, we describe an algorithm to automatically generate an intermediate scene using the bidirectional disparity morphing from the parallel stereopair. To compute the disparity between two reference images, we use the 2-step fast block matching algorithm that restricts the searching range and accelerates the speed of the computation of the disparity. We also define three occluding patterns so as to smooth the computed disparities, especially for occluded regions. They are derived from the peculiar properties of the disparity map. The smoothed disparity maps present that the false disparities are well corrected and the boundary between foreground and background becomes sharper. We discuss the advantages of this algorithm compared to the commonly used schemes and we show some experimental results with real data.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.3
• /
• pp.131-143
• /
• 1998

This paper presents a high performance disparity extraction algorithm that generate a dense and accurate disparity map using low-resolution disparity histogram. Disparity distribution of background and object areas can besegmented from low-resolution disparity histogram. These information can be used to reduce the search area and search range of the high-resolution image resulting reliable disparity information in high speed. The computationally efficient matching pixel count(MPC) similarity measure technique is useed extensively toremove the redundancies inherent in the area-based matching method, and also results robust matching at the boundary region. Resulting maches are further improved using iterative support algorithm and post processing. We have obtained good results on randomdot stereogram and real images obtained in our carmera system.

• Proceedings of the KSRS Conference
• /
• 2008.10a
• /
• pp.332-335
• /
• 2008

Newly developing SAR (Synthetic Aperture Radar) sensors commonly include high resolution X-band those data are expected to contribute various applications. Recent few studies are presenting potential of X-band SAR data in forest related application. This study tried to investigate the relationship between forest stand parameters and multi-band SAR normalized backscattering. Multi-band SAR data was radiometric corrected to compare signal from different forest stand condition. Then correlation coefficients were estimated between attribute of forest stand map and normalized backscattering coefficients. Although overall correlation coefficients are not high, only X-band shows strong relationship with DBH class than other bands. The signal of C- and L-band is composed of a large number of discrete tree components such as leaves, stems, even background soil. In forest, strength of radar backscattering is affected by complex parameters. Further study might be considered more various forest stand parameters such as canopy density, stand height, volume, and biomass.