• The KIPS Transactions:PartB
• /
• v.9B no.5
• /
• pp.563-570
• /
• 2002

Robust extraction of 3D facial features and global motion information from 2D image sequence for the MPEG-4 SNHC face model encoding is described. The facial regions are detected from image sequence using multi-modal fusion technique that combines range, color and motion information. 23 facial features among the MPEG-4 FDP (Face Definition Parameters) are extracted automatically inside the facial region using color transform (GSCD, BWCD) and morphological processing. The extracted facial features are used to recover the 3D shape and global motion of the object using paraperspective camera model and SVD (Singular Value Decomposition) factorization method. A 3D synthetic object is designed and tested to show the performance of proposed algorithm. The recovered 3D motion information is transformed into global motion parameters of FAP (Face Animation Parameters) of the MPEG-4 to synchronize a generic face model with a real face.

• Journal of Korea Multimedia Society
• /
• v.10 no.9
• /
• pp.1117-1124
• /
• 2007

In this paper, an algorithm that can define the threshold value automatically proposed in order to detect a brain ventricle in MRI brain images. After the wavelet transform, edge sharpness, which means the average magnitude of detail signals on the contour of the object, was computed by using the magnitude of horizontal and vertical detail signals. The contours of a brain ventricle were detected by increasing the threshold value repeatedly and computing edge sharpness. When the edge sharpness became maximal, the optimal threshold was determined, and the detection of a brain ventricle was accomplished finally. In this paper, we compared the proposed algorithm with the geodesic active contour model numerically and verified the efficiency of the proposed algorithm by applying real MRI brain images.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.1
• /
• pp.195-200
• /
• 2021

With the development of transportation and mobiles demand for tourism travel is increasing and related industries are also developing significantly. The combination of augmented reality and tourism contents one of the areas of digital media technology, is also actively being studied, and artificial intelligence is already combined with the tourism industry in various directions, enriching tourists' travel experiences. In this paper, we propose a system that scans miniature models produced by reducing tourist areas, finds the relevant tourist sites based on models learned using deep learning in advance, and provides relevant information and 3D models as AR services. Because model learning and object detection are carried out using YOLOv3 neural networks, one of various deep learning neural networks, object detection can be performed at a fast rate to provide real-time service.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.12
• /
• pp.1794-1799
• /
• 2022

Recently, the increasing number of overloaded vehicles on the road poses a risk to traffic safety, such as falling objects, road damage, and chain collisions due to the abnormal weight distribution, and can cause great damage once an accident occurs. therefore we propose to build an object detection-based AI model to identify overloaded vehicles that cause such social problems. In addition, we present a simple yet effective method to construct an object detection model for the large-scale vehicle images. In particular, we utilize the large-scale of vehicle image sets provided by open AI-Hub, which include the overloaded vehicles. We inspected the specific features of sizes of vehicles and types of image sources, and pre-processed these images to train a deep learning-based object detection model. Also, we propose an integrated system for tracking the detected vehicles. Finally, we demonstrated that the detection performance of the overloaded vehicle was improved by about 23% compared to the one using raw data.

• Journal of Broadcast Engineering
• /
• v.27 no.3
• /
• pp.391-401
• /
• 2022

As deep learning-based object detection and recognition research have been developed recently, the scope of application to industry and real life is expanding. But deep learning-based systems in the construction system are still much less studied. Calculating materials in the construction system is still manual, so it is a reality that transactions of wrong volumn calculation are generated due to a lot of time required and difficulty in accurate accumulation. A fast and accurate automatic drawing recognition system is required to solve this problem. Therefore, we propose an AI-based automatic drawing recognition accumulation system that detects and recognizes steel materials in construction drawings. To accurately detect steel materials in construction drawings, we propose data augmentation techniques and spatial attention modules for improving small object detection performance based on YOLOv4. The detected steel material area is recognized by text, and the number of steel materials is integrated based on the predicted characters. Experimental results show that the proposed method increases the accuracy and precision by 1.8% and 16%, respectively, compared with the conventional YOLOv4. As for the proposed method, Precision performance was 0.938. The recall was 1. Average Precision AP_0.5 was 99.4% and AP_0.5:0.95 was 67%. Accuracy for character recognition obtained 99.9.% by configuring and learning a suitable dataset that contains fonts used in construction drawings compared to the 75.6% using the existing dataset. The average time required per image was 0.013 seconds in the detection, 0.65 seconds in character recognition, and 0.16 seconds in the accumulation, resulting in 0.84 seconds.

• The Journal of Korea Robotics Society
• /
• v.3 no.3
• /
• pp.212-218
• /
• 2008

Collision avoidance is a fundamental and important task of an autonomous mobile robot for safe navigation in real environments with high uncertainty. Obstacles are classified into static and dynamic obstacles. It is difficult to avoid dynamic obstacles because the positions of dynamic obstacles are likely to change at any time. This paper proposes a scheme for vision-based avoidance of dynamic obstacles. This approach extracts object candidates that can be considered moving objects based on the labeling algorithm using depth information. Then it detects moving objects among object candidates using motion vectors. In case the motion vectors are not extracted, it can still detect the moving objects stably through their color information. A robot avoids the dynamic obstacle using the dynamic window approach (DWA) with the object path estimated from the information of the detected obstacles. The DWA is a well known technique for reactive collision avoidance. This paper also proposes an algorithm which autonomously registers the obstacle color. Therefore, a robot can navigate more safely and efficiently with the proposed scheme.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.4
• /
• pp.471-480
• /
• 1999

This paper presents design and performance improvement of a new digital tomosynthesis (DTS) system for object-detector synchronous rotation. Firstly, a new DTS system, called OSDR (Object-Detector Synchronous Rotation) is suggested and designed to acquire X-ray digital images. Secondly, the shape distortion of DTS images generated by an image intensifier is modeled. And a new synthesis algorithm, which overcomes the limitations of the existing synthesis algorithm, is suggested to improve the sharpness of the synthesized image. Also an artifact analysis of the DTS system is performed. Thirdly, some performance indices, which evaluate quantitatively performance improvement, are defined. And the experimental verification of the performance improvement is accomplished by the ODSR system newly designed. The advantages of the ODSR system are expressed quantitatively, compared with an existing system.

• Proceedings of the IEEK Conference
• /
• 2001.06d
• /
• pp.219-222
• /
• 2001

Moving objects in video data are main elements for video analysis and retrieval. In this paper, we propose a new algorithm for tracking and segmenting moving objects in color image sequences that include complex camera motion such as zoom, pan and rotating. The Proposed algorithm is based on the Mean-shift color segmentation and stochastic region matching method. For segmenting moving objects, each sequence is divided into a set of similar color regions using Mean-shift color segmentation algorithm. Each segmented region is matched to the corresponding region in the subsequent frame. The motion vector of each matched region is then estimated and these motion vectors are summed to estimate global motion. Once motion vectors are estimated for all frame of video sequences, independently moving regions can be segmented by comparing their trajectories with that of global motion. Finally, segmented regions are merged into the independently moving object by comparing the similarities of trajectories, positions and emerging period. The experimental results show that the proposed algorithm is capable of segmenting independently moving objects in the video sequences including complex camera motion.

• Proceedings of the IEEK Conference
• /
• 2001.06d
• /
• pp.89-92
• /
• 2001

In this paper, we propose a new approach for tracking a moving object in moving image sequences using active contour models and optical flow. In our approach object segmentation is achieved by active contours, and object tracking is done by motion estimation based on optical flow. To get more dynamic characteristics, Lagrangian dynamics combined to the active contour models. For the optical flow computation, a method, which is based on Spatiotempo-ral Energy Models, is employed to perform robust tracking under poor environments. A prototype real tracking system has been developed and applied to a contents-based video retrieval systems.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.739-742
• /
• 2001

동영상은 각 프레임 사이에 시간적으로나 공간적으로 많은 양의 정보가 중복되어 있다. 이러한 중복 정보를 줄이는 표현 방법들 중에 하나로, 동영상을 커다란 하나의 영상으로 정합하여 중복 정보를 줄이는 모자이크 기법이 있다. 두 개 이상의 영상을 정합하기 위해서는 영상간의 카메라 파라미터가 필요한데, 본 논문에서는 Mellin 변환을 사용하여 카메라 파라미터를 구하였다. 이때 3차원 공간모델은 직교 투사법을 사용하였으며, 영상의 움직임 모델로는 4개의 파라미터(평행이동, 확대/축소, 회전)를 사용한 어파인 움직임 모델을 사용하였다. 이렇게 구현된 파노라마 영상은 동영상에서 움직이는 물체를 검출하거나 추적하고, 동영상을 편집하는데 응용될 수 있다. 또한 본 연구의 최종 목적인 3D 영상의 배경을 구현하는 데 좀 더 사실적인 영상을 제공할 수 있다.