• Journal of the Korea Computer Graphics Society
• /
• v.20 no.2
• /
• pp.1-11
• /
• 2014

The quality of 2D-to-3D conversion depends on the accuracy of the assigned depth to scene objects. Manual depth painting for given objects is labor intensive as each frame is painted. Specifically, a human is one of the most challenging objects for a high-quality conversion, as a human body is an articulated figure and has many degrees of freedom (DOF). In addition, various styles of clothes, accessories, and hair create a very complex silhouette around the 2D human object. We propose an efficient method to estimate visually pleasing depths of a human at every frame in a monocular video. First, a 3D template model is matched to a person in a monocular video with a small number of specified user correspondences. Our pose estimation with sequential joint angular constraints reproduces a various range of human motions (i.e., spine bending) by allowing the utilization of a fully skinned 3D model with a large number of joints and DOFs. The initial depth of the 2D object in the video is assigned from the matched results, and then propagated toward areas where the depth is missing to produce a complete depth map. For the effective handling of the complex silhouettes and appearances, we introduce a partial depth propagation method based on color segmentation to ensure the detail of the results. We compared the result and depth maps painted by experienced artists. The comparison shows that our method produces viable depth maps of humans in monocular videos efficiently.

• ETRI Journal
• /
• v.37 no.4
• /
• pp.766-771
• /
• 2015

Robust motion estimation for human-computer interactions played an important role in a novel method of interaction with electronic devices. Existing pose estimation using a monocular camera employs either ego-motion or exo-motion, both of which are not sufficiently accurate for estimating fine motion due to the motion ambiguity of rotation and translation. This paper presents a hybrid vision-based pose estimation method for fine-motion estimation that is specifically capable of extracting human body motion accurately. The method uses an ego-camera attached to a point of interest and exo-cameras located in the immediate surroundings of the point of interest. The exo-cameras can easily track the exact position of the point of interest by triangulation. Once the position is given, the ego-camera can accurately obtain the point of interest's orientation. In this way, any ambiguity between rotation and translation is eliminated and the exact motion of a target point (that is, ego-camera) can then be obtained. The proposed method is expected to provide a practical solution for robustly estimating fine motion in a non-contact manner, such as in interactive games that are designed for special purposes (for example, remote rehabilitation care systems).

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.389-395
• /
• 2022

Because of improvement of deep learning techniques, deep learning using computer vision such as classification, detection and segmentation has also been used widely at many fields. Expecially, automatic driving is one of the major fields that applies computer vision systems. Also there are a lot of works and researches to combine multiple tasks in a single network. In this study, we propose the network that predicts the individual depth of pedestrians and vehicles. Proposed model is constructed based on YOLOv3 for object detection and Monodepth for depth estimation, and it process object detection and depth estimation consequently using encoder and decoder based on hard parameter sharing. We also used attention module to improve the accuracy of both object detection and depth estimation. Depth is predicted with monocular image, and is trained using self-supervised training method.

• Smart Media Journal
• /
• v.12 no.11
• /
• pp.48-56
• /
• 2023

The technology of Three-dimensional human posture estimation is used in sports, motion recognition, and special effects of video media. Among various methods for this, multi-view 3D human pose estimation is essential for precise estimation even in complex real-world environments. But Existing models for multi-view 3D human posture estimation have the disadvantage of high order of time complexity as they use 3D feature maps. This paper proposes a method to extend an existing monocular viewpoint multi-frame model based on Transformer with lower time complexity to 3D human posture estimation for multi-viewpoints. To expand to multi-viewpoints our proposed method first generates an 8-dimensional joint coordinate that connects 2-dimensional joint coordinates for 17 joints at 4-vieiwpoints acquired using the 2-dimensional human posture detector, CPN(Cascaded Pyramid Network). This paper then converts them into 17×32 data with patch embedding, and enters the data into a transformer model, finally. Consequently, the MLP(Multi-Layer Perceptron) block that outputs the 3D-human posture simultaneously updates the 3D human posture estimation for 4-viewpoints at every iteration. Compared to Zheng[5]'s method the number of model parameters of the proposed method was 48.9%, MPJPE(Mean Per Joint Position Error) was reduced by 20.6 mm (43.8%) and the average learning time per epoch was more than 20 times faster.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.6
• /
• pp.24-34
• /
• 2007

Application systems that easy to access a information have been developed by IT growth and a human life variation. In this paper, we propose a application system to register a 3D footwear model using a monocular camera. In General, a human motion analysis research to body movement. However, this system research a new method to use a foot movement. This paper present a system process and show experiment results. For projection to 2D foot plane from 3D shoe model data, we construct processes that a foot tracking, a projection expression and pose estimation process. This system divide from a 2D image analysis and a 3D pose estimation. First, for a foot tracking, we propose a method that find fixing point by a foot characteristic, and propose a geometric expression to relate 2D coordinate and 3D coordinate to use a monocular camera without a camera calibration. We make a application system, and measure distance error. Then, we confirmed a registration very well.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.6
• /
• pp.143-148
• /
• 2022

It is very important for metaverse, mobile robot, and user location services to analyze the images continuously taken using a mobile smartphone or robot's monocular camera to estimate the camera's location. So far, PnP-related techniques have been applied to calculate the position. In this paper, the camera's moving direction is obtained using the essential matrix in the epipolar geometry applied to successive images, and the camera's continuous moving position is calculated through geometrical equations. A new estimation method was proposed, and its accuracy was verified through simulation. This method is completely different from the existing method and has a feature that it can be applied even if there is only one or more matching feature points in two or more images.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.34S no.11
• /
• pp.126-137
• /
• 1997

This paper describes the algorithm which can estimate road following direction and deetect obstacle using a monocular vision system. This algorithm can estimate the course of vehicle using the vanishing point properties and detect obstacle by statistical method. The proposed algorithm is composed of four steps, which are lane prediction, lane extraction, road following parameter estimation and obstacle detection. It is designed for high processing speed and high accuracy. The former is achieved by a small area named sub-windown in lane existence area, the later is realized by using connected edge points of lane. We would like to present that the new mehod can detect obstacle using the simple statistical method. The paracticalities of the processing speed, the accuracy of the algorithm and proposing obstacle detection method, have been justified through the experiment applied VTR image of the real road to the algorithm.

• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.10
• /
• pp.983-988
• /
• 2011

This paper presents a robust method for ground plane detection in vision-based applications based on a monocular sequence of images with a non-stationary camera. The proposed method, which is based on the reliable estimation of the homography between two frames taken from the sequence, aims at designing a practical system to detect road surface from traffic scenes. The homography is computed using a feature matching approach, which often gives rise to inaccurate matches or undesirable matches from out of the ground plane. Hence, the proposed homography estimation minimizes the effects from erroneous feature matching by the evaluation of the difference between the predicted and the observed matrices. The method is successfully demonstrated for the detection of road surface performed on experiments to fill an information void area taken place from geometric transformation applied to captured images by an in-vehicle camera system.

• Journal of Advanced Marine Engineering and Technology
• /
• v.36 no.4
• /
• pp.512-519
• /
• 2012

This paper proposes a 3-D pose (positions and orientations) estimation system based on the recognition of circular ring patterns. To deal with monocular vision-based pose estimation problem, we specially design a circular ring pattern that has a simplicity merit in view of object recognition. A pose estimation procedure is described in detail, which utilizes the geometric transformation of a circular ring pattern in 2-D perspective projection space. The proposed method is evaluated through the analysis of accuracy and precision with respect to 3-D pose estimation of a quadrotor-type vehicle in 3-D space.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.45-54
• /
• 2022

Image depth estimation is a technology that is the basis of various image analysis. As analysis methods using deep learning models emerge, studies using deep learning in image depth estimation are being actively conducted. Currently, most deep learning-based depth estimation models are being trained with clean and ideal images. However, due to the lack of data on adverse conditions such as haze or fog, the depth estimation may not work well in such an environment. It is hard to sufficiently secure an image in these environments, and in particular, obtaining non-homogeneous haze data is a very difficult problem. In order to solve this problem, in this study, we propose a method of synthesizing non-homogeneous haze images and a learning method for a monocular depth estimation deep learning model using this method. Considering that haze mainly occurs outdoors, datasets mainly containing outdoor images are constructed. Experiment results show that the model with the proposed method is good at estimating depth in both synthesized and real haze data.