• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.07a
• /
• pp.321-322
• /
• 2012

본 논문은 선명도를 향상시킨 고해상도 깊이맵을 생생 방법을 제안한다. 현재 저해상도 깊이맵으로부터 생성되는 고해상도 깊이맵은 원 깊이맵과 유사도를 높이는 것에 초점이 맞춰져 있다. 본 논문은 기존 보간법들을 바탕으로 깊이맵에 고주파 성분을 사용하여 깊이맵의 선명도를 증가시킨다. 제안 방법은 저해상도 깊이맵으로부터 고주파 데이터를 생성 후, 깊이맵에 고주파 성분을 적용한 다음에 보간을 통하여 깊이맵을 고해상도 깊이맵으로 변환한다.

• Smart Media Journal
• /
• v.1 no.3
• /
• pp.22-28
• /
• 2012

As the smart media becomes more popular, the demand for high-quality 3D images and depth maps is increasing. However, performance of the current technologies to acquire depth maps is not sufficient. The depth maps from stereo matching methods have low accuracy in homogeneous regions. The depth maps from depth cameras are noisy and have low-resolution due to technical limitations. In this paper, we introduce the state-of-the-art algorithms for depth map enhancement and up-sampling from conventional methods using only depth maps to the latest algorithms referring to both depth maps and their corresponding color images. We also present depth map enhancement algorithms for hybrid camera systems in detail.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.11a
• /
• pp.112-113
• /
• 2012

최근 카메라와 디스플레이의 발전에 따라 고해상도 영상이 요구되고 있다. 하지만 깊이를 획득하는 깊이센서 장치는 색상 영상에 미치지 못하는 저해상도 깊이맵을 주로 제공한다. 이에 따라 저해상도의 깊이맵을 고해상도 깊이맵으로 상향변환이 필요하다. 하지만 대부분의 보간법들은 edge에서 blur가 발생하는 경우가 있다. 따라서 본 논문에서는 distance transform(DT)를 이용하여 edge의 선명도를 향상시킨 고해상도 깊이맵 생성 방법을 제안한다.

• Journal of Broadcast Engineering
• /
• v.13 no.5
• /
• pp.596-601
• /
• 2008

In this paper, we propose a new scheme to generate region-of-interest (ROI) enhanced depth maps in the hybrid camera system, which is composed of a low-resolution depth camera and a high-resolution stereoscopic camera. The proposed method creates an ROI depth map for the left image by carrying out a three-dimensional (3-D) warping operation onto the depth information obtained from the depth camera. Then, we generate a background depth map for the left image by applying a stereo matching algorithm onto the left and right images captured by the stereoscopic camera. Finally, we merge the ROI map with the background one to create the final depth map. The proposed method provides higher quality depth information on ROI than the previous methods.

• Journal of Broadcast Engineering
• /
• v.9 no.3
• /
• pp.185-195
• /
• 2004

This paper presents a multi-depth map fusion method for the 3D scene reconstruction. It fuses depth maps obtained from the stereo matching technique and the depth camera. Traditional stereo matching techniques that estimate disparities between two images often produce inaccurate depth map because of occlusion and homogeneous area. Depth map obtained from the depth camera is globally accurate but noisy and provide a limited depth range. In order to get better depth estimates than these two conventional techniques, we propose a depth map fusion method that fuses the multi-depth maps from stereo matching and the depth camera. We first obtain two depth maps generated from the stereo matching of 3-view images. Moreover, a depth map is obtained from the depth camera for the center-view image. After preprocessing each depth map, we select a depth value for each pixel among them. Simulation results showed a few improvements in some background legions by proposed fusion technique.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.11a
• /
• pp.137-138
• /
• 2011

최근 깊이센서가 컴퓨터비전 등의 영상처리 분야에서 다양하게 활용되고 있다. 그러나 깊이센서에서 생성된 깊이맵의 해상도가 낮기 때문에 고해상도로 상향변환이 필요하다. 현재까지 저해상도의 깊이맵을 고해상도의 깊이맵으로 변환하는 방법들이 많이 제안되었다. 하지만 이러한 방법들은 객체의 에지 개선에만 국한되어 있다. 따라서 본 논문에서는 객체의 에지 뿐만아니라, 객체의 내부를 개선하는 방법을 제안한다. 제안방법은 기존에서 활용되어 온 보간법들에 고주파 성분을 적용하여 개선된 고해상도 깊이맵을 얻는다.

• Journal of Broadcast Engineering
• /
• v.23 no.1
• /
• pp.3-10
• /
• 2018

Depth map has been utilized in a varity of fields. Recently research on generating depth map by artificial neural network (ANN) has gained much interest. This paper validates the feasibility of implementing the ready-made depth map generation by convolutional neural network (CNN). First, for a given image, a depth map is generated by the weighted average of a saliency map as well as a motion history image. Then CNN network is trained by test images and depth maps. The objective and subjective experiments are performed on the CNN and showed that the CNN can replace the ready-made depth generation method.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.620-624
• /
• 2009

In this paper, we present a new camera system combining a high-quality 3-D scanner and hybrid camera system to generate a multiview video-plus-depth. In order to get the 3-D video using the hybrid camera system and 3-D scanner, we first obtain depth information for background region from the 3-D scanner. Then, we get the depth map for foreground area from the hybrid camera system. Initial depths of each view image are estimated by performing 3-D warping with the depth information. Thereafter, multiview depth estimation using the initial depths is carried out to get each view initial disparity map. We correct the initial disparity map using a belief propagation algorithm so that we can generate the high-quality multiview disparity map. Finally, we refine depths of the foreground boundary using extracted edge information. Experimental results show that the proposed depth maps generation method produces a 3-D video with more accurate multiview depths and supports more natural 3-D views than the previous works.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.11a
• /
• pp.34-35
• /
• 2017

본 논문에서는 영상으로부터 생성된 깊이맵을 합성곱 신경망(CNN)으로 재생성하는 방법을 제안한다. 합성곱 신경망은 영상인식, 영상분류에 좋은 성능을 보여주는데, 이 기술을 깊이맵 생성에 활용하여 기 제작된 깊이맵 생성 기법을 간단한 합성곱 신경망으로 구현하고자 한다. 성능 실험에서는 10개의 비디오 세트에 제안 방법을 적용한 결과, 만족스러운 결과를 얻었다.

• Journal of Broadcast Engineering
• /
• v.17 no.6
• /
• pp.933-944
• /
• 2012

In this paper, we propose a new method to convert a low-resolution depth map into its high-resolution one called distance transform-based bilateral upsampling. Since the proposed method controls the spatial domain weighting function based on distance transform values of the depth map, it increases the input depth map resolution while preserving edge sharpness. The proposed method is composed of three main steps: distance transform, spatial weighting control, and image interpolation. Experimental results show that our method outperforms the conventional bilateral upsampling in terms of the quality of output depth maps.