• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제5권9호
• /
• pp.1613-1631
• /
• 2011

Several metrics have been reported in the literature to assess stereo image quality, mostly based on visual attention or human visual sensitivity based distortion prediction with the help of disparity information, which do not consider the combined aspects of human visual processing. In this paper, visual attention and depth assisted stereo image quality assessment model (VAD-SIQAM) is devised that consists of three main components, i.e., stereo attention predictor (SAP), depth variation (DV), and stereo distortion predictor (SDP). Visual attention is modeled based on entropy and inverse contrast to detect regions or objects of interest/attention. Depth variation is fused into the attention probability to account for the amount of changed depth in distorted stereo images. Finally, the stereo distortion predictor is designed by integrating distortion probability, which is based on low-level human visual system (HVS), responses into actual attention probabilities. The results show that regions of attention are detected among the visually significant distortions in the stereo image pair. Drawbacks of human visual sensitivity based picture quality metrics are alleviated by integrating visual attention and depth information. We also show that positive correlation with ground-truth attention and depth maps are increased by up to 0.949 and 0.936 in terms of the Pearson and the Spearman correlation coefficients, respectively.

• 인터넷정보학회논문지
• /
• 제21권3호
• /
• pp.113-121
• /
• 2020

Human Pose Estimation (HPE) which localizes the human body joints becomes a high potential for high-level applications in the field of computer vision. The main challenges of HPE in real-time are occlusion, illumination change and diversity of pose appearance. The single RGB image is fed into HPE framework in order to reduce the computation cost by using depth-independent device such as a common camera, webcam, or phone cam. However, HPE based on the single RGB is not able to solve the above challenges due to inherent characteristics of color or texture. On the other hand, depth information which is fed into HPE framework and detects the human body parts in 3D coordinates can be usefully used to solve the above challenges. However, the depth information-based HPE requires the depth-dependent device which has space constraint and is cost consuming. Especially, the result of depth information-based HPE is less reliable due to the requirement of pose initialization and less stabilization of frame tracking. Therefore, this paper proposes a new method of HPE which is robust in estimating self-occlusion. There are many human parts which can be occluded by other body parts. However, this paper focuses only on head self-occlusion. The new method is a combination of the RGB image-based HPE framework and the depth information-based HPE framework. We evaluated the performance of the proposed method by COCO Object Keypoint Similarity library. By taking an advantage of RGB image-based HPE method and depth information-based HPE method, our HPE method based on RGB-D achieved the mAP of 0.903 and mAR of 0.938. It proved that our method outperforms the RGB-based HPE and the depth-based HPE.

• 대한인간공학회지
• /
• 제18권1호
• /
• pp.41-53
• /
• 1999

The effect of menu width and depth on the efficiency of information search and menu preference was investigated to identify an optimal menu structure for EIG which reflects the characteristics of human information processing. Information search time increased stepwisely as the menu width exceeded 6 items and linearly as the level of menu depth increased. The linear relationship between the error rate and the number of depth levels seems to be caused by the increase in the items to be remembered. When a menu structure was constructed by combining different menu depths and widths, it was observed that making the menu width wider rather than the depth deeper allows better information search. The menu structure rated as the most preferable and the easiest to user was that of pyramidal form. Such a result seems to come from its structural similarity to general categories which people get used to and implies that one should consider user preference as well as efficiency of search when he/she designs an EIG menu.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 신호처리소사이어티 추계학술대회 논문집
• /
• pp.171-174
• /
• 2003

In this paper, we propose the method of stereo images composition using adaptive dense disparity estimation. For the correct composition of stereo image and 3D virtual object, we need correct marker position and depth information. The existing algorithms use position information of markers in stereo images for calculating depth of calibration object. But this depth information may be wrong in case of inaccurate marker tracking. Moreover in occlusion region, we can't know depth of 3D object, so we can't composite stereo images and 3D virtual object. In these reasons, the proposed algorithm uses adaptive dense disparity estimation for calculation of depth. The adaptive dense disparity estimation is the algorithm that use pixel-based disparity estimation and the search range is limited around calibration object.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 ITC-CSCC -3
• /
• pp.1666-1669
• /
• 2002

In the process of creating architectural scenes from photographs using Model-based Stereo 〔1〕, the geometric model is used as prior information to solve correspondence problems and recover the depth or disparity of real scenes. This paper presents an Image Subregioning algorithm that divides left and right images into several rectangular sub-images. The division is done according to the estimated depth of real scenes using a Heuristic Approach. The depth difference between the reality and the model can be partitioned into each depth level. This reduces disparity search range in the Similarity Function. For architectural scenes with complex depth, experiments using the above approach show that accurate disparity maps and better results when rendering scenes can be achieved by the proposed algorithm.

• 디지털융복합연구
• /
• 제11권1호
• /
• pp.243-248
• /
• 2013

본 논문에서는 2D/3D 동영상 변환을 위해 깊이가 할당될 전경을 초점 정보와 색상분석 기반의 그룹화를 이용하여 추출하고, 전경의 깊이를 초점 정보와 움직임 정보를 이용하여 생성하는 방법을 제안하였다. 2D영상에서 전경을 추출하기 위해 영상의 초점 정보의 움직임을 추정하여 전경 후보 영상을 생성하고, 전경 후보 영상에 존재하는 객체 내부의 홀 영역을 색상 분석을 이용한 채움 과정을 수행하여 전경 영역을 추출하였다. 생성된 전경 영역에 깊이를 할당하기 위해 해당 프레임에 존재하는 초점 값을 분석하여 초기 깊이 정보를 생성하고 움직임 정보를 가중하여 깊이 정보를 할당하였다. 생성된 깊이 정보의 품질을 평가하기 위해 기존에 제안된 알고리즘의 결과 영상과 비교하였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2008년도 하계종합학술대회
• /
• pp.815-816
• /
• 2008

In this paper, we evaluate depth image preprocessing for 3D data service based on DIBR over T-DMB. We evaluate two preprocessing methods of depth images. These are gaussian smoothing and adaptive smoothing. The results show that adaptive smoothing is more suitable for images with sharp transition of depth.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제17권2호
• /
• pp.252-259
• /
• 2017

In this paper, we propose a novel method for extracting an accurate depth of a trap that causes RTS(Random Telegraph Signal) noise. The error rates of the trap depth rely on the mean time constants and its ratio. Here, we determined how many data of the capture and emission time constant are necessary in order to reduce the trap depth error caused by an inaccurate mean time constant. We measured the capture and emission time constants up to 100,000 times in order to ensure that the samples had statistical meaning. As a result, we demonstrated that at least 1,000 samples are necessary to satisfy less than 10% error for trap depth. This result could be used to improve the accuracy of RTS noise analysis.

• 한국정보통신학회논문지
• /
• 제19권1호
• /
• pp.127-135
• /
• 2015

본 연구에서는 DIBR(Depth-Image-Based Rendering) 기법을 사용한 다시점 중간 영상 생성을 하는데 있어서 최종 중간 영상의 화질을 개선하기 위한 방법으로 깊이 정보의 확장과 메쉬 구성 (즉, 표면 재구성) 그리고 이 방법들을 교차 적용하여 어떤 것이 가장 좋은 결과를 낼 수 있는 지에 대해서 실험하였다. 마이크로소프트사에서 제공하는 브레이크 댄서와 발레 영상을 사용하여 실험 결과를 살펴보았고 다양한 틈새 영역 채움 알고리즘을 적용시켜 분석해 보았다. 실험 결과는 기존의 점 구름만을 활용하는 것보다 깊이 정보를 확장하는 방법과 메쉬 구성 방법을 모두 시켰을 때 가장 좋은 결과가 나타났다. 그리고 틈새 영역 채움 알고리즘을 적용하기 전에는 깊이 확장만으로도 충분히 화질이 향상됨을 확인할 수 있었다.

• 디지털산업정보학회논문지
• /
• 제12권1호
• /
• pp.27-33
• /
• 2016

In this study we propose a method for 3D landform reconstruction and object modeling method by generating contour lines on the map using a depth sensor which abstracts characteristics of geological layers from the depth map. Unlike the common visual camera, the depth-sensor is not affected by the intensity of illumination, and therefore a more robust contour and object can be extracted. The algorithm suggested in this paper first abstracts the characteristics of each geological layer from the depth map image and rearranges it into the proper order, then creates contour lines using the Bezier curve. Using the created contour lines, 3D images are reconstructed through rendering by mapping RGB images of the visual camera. Experimental results show that the proposed method using depth sensor can reconstruct contour map and 3D modeling in real-time. The generation of the contours with depth data is more efficient and economical in terms of the quality and accuracy.