• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.4
• /
• pp.632-638
• /
• 2002

Extensible 3D(X3D) is a software standard for defining interactive web and broadcast-based 3D content integrated with multimedia. The data size of Web3D representation based on polygon meshes is so large that transferring practical data fast is a hard problem. This paper proposes 3D object structure, a new framework for compact 3D representation with high quality surface shape. By utilizing a free form surface technique, qualified surface are transferred with limited amount of data size and rendered. 3D graphic structure can be regarded ad both polygon meshes and free form surfaces. Therefore, it can be easily integrated to existing Web3D data formats, for example VRML ＆ XML. 3D object structure also enables modeling free form surface shapes intuitively with polygon modeling like operations.

• The KIPS Transactions:PartC
• /
• v.12C no.3 s.99
• /
• pp.443-448
• /
• 2005

Efficient using of the web cache is becoming important factors that decide system management efficiency in the web base system. The cache performance depends heavily on the replacement algorithm which dynamically selects a suitable subset of objects for caching in a finite cache space. In this paper, the web caching algorithm is proposed for the efficient operation of the web base system. The algorithm is designed based on a divided scope that considered size reference characteristic and heterogeneity on web object. With the experiment environment, the algorithm is compared with conservative replacement algorithms, we have confirmed more than $15\%$ of an performance improvement.

• Korean Journal of Remote Sensing
• /
• v.9 no.2
• /
• pp.7-19
• /
• 1993

In stereo vision, when we use two parallel axis images, small portion of object is contained and B/H(Base-line to Height) ratio is limited due to the size of object and depth information is inaccurate. To overcome these difficulities we take a non-parallel axis image which is rotated $\theta$ about y-axis and match other parallel-axis image. Epipolar lines of non-parallel axis image are not same as those of parallel-axis image and we can't match these two images directly. In this paper, we transform the non-parallel axis image geometrically with camera parameters, whose epipolar lines are alingned parallel. NCC(Normalized Cross Correlation) is used as match measure, area-based matching technique is used find correspondence and 9$\times$9 window size is used, which is chosen experimentally. Focal length which is necessary to get depth information of given object is calculated with least-squares method by CCD camera characteristics and lenz property. Finally, we select 30 test points from given object whose elevation is varied to 150 mm, calculate heights and know that height RMS error is 7.9 mm.

• Journal of Korea Multimedia Society
• /
• v.22 no.2
• /
• pp.167-177
• /
• 2019

Even though so much progresses have been achieved in Multiple Object Tracking (MOT), most of reported MOT methods are not still satisfactory for commercial embedded products like Pan-Tilt-Zoom (PTZ) camera. In this paper, we propose a real-time multiple pedestrians tracking method for embedded environments. First, we design a new light weight convolutional neural network(CNN)-based pedestrian detector, which is constructed to detect even small size pedestrians, as well. For further saving of processing time, the designed detector is applied for every other frame, and Kalman filter is employed to predict pedestrians' positions in frames where the designed CNN-based detector is not applied. The pose orientation information is incorporated to enhance object association for tracking pedestrians without further computational cost. Through experiments on Nvidia's embedded computing board, Jetson TX2, it is verified that the designed pedestrian detector detects even small size pedestrians fast and well, compared to many state-of-the-art detectors, and that the proposed tracking method can track pedestrians in real-time and show accuracy performance comparably to performances of many state-of-the-art tracking methods, which do not target for operation in embedded systems.

• Journal of Internet of Things and Convergence
• /
• v.9 no.1
• /
• pp.97-104
• /
• 2023

This paper is related to object segmentation using ESRGAN(Enhanced Super Resolution GAN) and SSS(Semantic Soft Segmentation). The segmentation performance of the object segmentation method using Mask R-CNN and SSS proposed by the research team in this paper is generally good, but the segmentation performance is poor when the size of the objects is relatively small. This paper is to solve these problems. The proposed method aims to improve segmentation performance of small objects by performing super-resolution through ESRGAN and then performing SSS when the size of an object detected through Mask R-CNN is below a certain threshold. According to the proposed method, it was confirmed that the segmentation characteristics of small-sized objects can be improved more effectively than the previous method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.24 no.3
• /
• pp.298-303
• /
• 2014

In this paper, we proposed a robust object extraction and tracking algorithm in the IR image sequence acquired in the sea environment. In order to extract size-invariant object, we detect horizontal and vertical edges by using DWT and combine it to generate saliency map. To extract object region, binarization technique is applied to saliency map. The correspondences between objects in consecutive frames are defined by the calculating minimum weighted Euclidean distance as a matching measure. Finally, object trajectories are determined by considering false correspondences such as entering object, vanishing objects and false object and so on. The proposed algorithm can find trajectories robustly, which has shown by experimental results.

• Journal of Korea Multimedia Society
• /
• v.21 no.12
• /
• pp.1425-1430
• /
• 2018

In this paper, we propose a CNN based deep learning algorithm for semantic segmentation of images. In order to improve the accuracy of semantic segmentation, we combined pixel level object classification and image level object classification. The image level object classification is used to accurately detect the characteristics of an image, and the pixel level object classification is used to indicate which object area is included in each pixel. The proposed network structure consists of three parts in total. A part for extracting the features of the image, a part for outputting the final result in the resolution size of the original image, and a part for performing the image level object classification. Loss functions exist for image level and pixel level classification, respectively. Image-level object classification uses KL-Divergence and pixel level object classification uses cross-entropy. In addition, it combines the layer of the resolution of the network extracting the features and the network of the resolution to secure the position information of the lost feature and the information of the boundary of the object due to the pooling operation.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.21 no.5
• /
• pp.1109-1122
• /
• 1996

Generally, moving images contain the various components in motions, which reange from a static object and background to a fast moving object. To extract the accurate motion parameters, we must consider the various motions. That requires a wide search egion in motion estimation. The wide search, however, causes a high computational complexity. If we have a few knowledge about the motion direction and magnitude before motion estimation, we can determine the search location and search window size using the already-known information about the motion. In this paper, we present a local adaptive motion estimation approach that predicts a block motion based on spatio-temporal neighborhood blocks and adaptively defines the search location and search window size. This paper presents a technique for reducing computational complexity, while having high accuracy in motion estimation. The proposed algorithm is introduced the forward and backward projection techniques. The search windeo size for a block is adaptively determined by previous motion vectors and prediction errors. Simulations show significant improvements in the qualities of the motion compensated images and in the reduction of the computational complexity.

• Journal of Broadcast Engineering
• /
• v.19 no.6
• /
• pp.916-924
• /
• 2014

It is possible to improve the accuracy of the motion estimation for a video by applying a variable block size. However, it has limits in the zoom motion estimation. In this paper, we propose a method for estimating the zoom motion with variable block size. The proposed method separates the background within the object picture by depth information obtained from a depth camera, and only the object regions are applied to zoom scale, but the background is not applied. In addition, the object regions select efficiently variable block size mode in consideration of the generated motion vectors and the accuracy of motion estimation at the same time. Simulation results show the accuracy of the motion estimation and the number of motion vectors for the proposed method. It is verified that the proposed method can reduce the number of motion while maintaining the similar accuracy of motion estimation than the conventional motion estimation methods.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.17 no.1
• /
• pp.123-128
• /
• 2018

In this paper, we propose a multi-vehicle object detection algorithm using feature point matching that tracks efficient vehicle objects. The proposed algorithm extracts the feature points of the vehicle using the FAST algorithm for efficient vehicle object tracking. And True if the feature points are included in the image segmented into the 5X5 region. If the feature point is not included, it is processed as False and the corresponding area is blacked to remove unnecessary object information excluding the vehicle object. Then, the post processed area is set as the maximum search window size of the vehicle. And A minimum search window using the outermost feature points of the vehicle is set. By using the set search window, we compensate the disadvantages of the search window size of mean-shift algorithm and track vehicle object. In order to evaluate the performance of the proposed method, SIFT and SURF algorithms are compared and tested. The result is about four times faster than the SIFT algorithm. And it has the advantage of detecting more efficiently than the process of SUFR algorithm.