• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.8B
• /
• pp.609-614
• /
• 2008

Upstream data frame of TDM-PON includes various types of overheads, and there exist guard bands between consecutive frames from different ONUs. Although they are indispensible in synchronization and performance, they cause bandwidth waste at the same time. To solve this problem, a new TDM-PON that uses two types of wavelengths in upstream transmission is suggested. By even distribution of two wavelengths among ONUs and overhead overlap between frames that use different wavelengths, almost 100% bandwidth efficiency could be achieved. A serializer that multiplexes signals from two wavelengths is implemented for this purpose.

• The KIPS Transactions:PartB
• /
• v.13B no.5 s.108
• /
• pp.515-524
• /
• 2006

Background extraction is an important technique to find the moving objects in video surveillance system. This paper proposes a new on-line background extraction method for color video using vector order statistics. In the proposed method, using the fact that background occurs more frequently than objects, the vector median of color pixels in consecutive frames Is treated as background at the position. Also, the objects of current frame are consisted of the set of pixels whose distance from background pixel is larger than threshold. In the paper, the proposed method is compared with the on-line multiple background extraction based on Gaussian mixture model(GMM) in order to evaluate the performance. As the result, its performance is similar or superior to the method based on GMM.

• Journal of Sensor Science and Technology
• /
• v.21 no.3
• /
• pp.211-216
• /
• 2012

Fires are the most common disaster and early fire detection is of great importance to minimize the consequent damage. Simple sensors including smoke detectors are widely used for the purpose but they are able to sense fires only at close proximity. Recently, due to the rapid advances of relevant technologies, vision-based fire sensing has attracted growing attention. In this paper, a novel visual sensing technique to automatically detect fire is presented. The proposed technique consists of multiple steps of image processing: pixel-level, block-level, and frame level. At the first step, fire flame pixel candidates are selected based on their color values in YIQ space from the image of a camera which is installed as a vision sensor at a fire scene. At the second step, the dynamic parts of flames are extracted by comparing two consecutive images. These parts are then represented in regularly divided image blocks to reduce pixel-level detection error and simplify following processing. Finally, the temporal change of the detected blocks is analyzed to confirm the spread of fire. The proposed technique was tested using real fire images and it worked quite reliably.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.8
• /
• pp.742-747
• /
• 2015

The quality of steel plate products is determined by the number of defects and the process problems are estimated by shapes of defects. Therefore pinholes defects of cold rolled steel have to be controlled. In order to improve productivity and quality of products, within each production process, the product is inspected by an adequate inspection system individually in the lines of steelworks. Among a number of inspection systems, we focus on the pinholes detection system. In this paper, we propose an embedded system using FPGA which can detect pinholes defects. The proposed system is smaller and more flexible than a traditional system based on expensive frame grabbers and PC. In order to detect consecutive defects, FPGAs acquire two dimensional image and process the image in real time by using correlation of lines. The proposed pinholes detection algorithm decreases arithmetic operations of image processing and also we designed the hardware to shorten the data path between logics due to decreasing propagation delay. The experimental results show that the proposed embedded system detects the reliable number of pinholes in real time.

• Journal of Korea Multimedia Society
• /
• v.24 no.7
• /
• pp.890-902
• /
• 2021

Although the object detection accuracy with still images has been significantly improved with the advance of deep learning techniques, the object detection problem with video data remains as a challenging problem due to the real-time requirement and accuracy drop with occlusion. In this research, we propose a method in pig detection for video monitoring environment. First, we determine a motion, from a video data obtained from a tilted-down-view camera, based on the average size of each pig at each location with the training data, and extract key frames based on the motion information. For each key frame, we then apply YOLO, which is known to have a superior trade-off between accuracy and execution speed among many deep learning-based object detectors, in order to get pig's bounding boxes. Finally, we merge the bounding boxes between consecutive key frames in order to reduce false positive and negative cases. Based on the experiment results with a video data set obtained from a pig farm, we confirmed that the pigs could be detected with an accuracy of 97% at a processing speed of 37fps.

• Advances in nano research
• /
• v.12 no.2
• /
• pp.223-239
• /
• 2022

Object detection has always been to pursue objects with particular properties or representations and to predict details on objects including the positions, sizes and angle of rotation in the current picture. This was a very important subject of computer vision science. While vision-based object tracking strategies for the analysis of competitive videos have been developed, it is still difficult to accurately identify and position a speedy small ball. In this study, deep learning (DP) network was developed to face these obstacles in the study of tennis motion tracking from a complex perspective to understand the performance of athletes. This research has used CNN-VGG 16 to tracking the tennis ball from broadcasting videos while their images are distorted, thin and often invisible not only to identify the image of the ball from a single frame, but also to learn patterns from consecutive frames, then VGG 16 takes images with 640 to 360 sizes to locate the ball and obtain high accuracy in public videos. VGG 16 tests 99.6%, 96.63%, and 99.5%, respectively, of accuracy. In order to avoid overfitting, 9 additional videos and a subset of the previous dataset are partly labelled for the 10-fold cross-validation. The results show that CNN-VGG 16 outperforms the standard approach by a wide margin and provides excellent ball tracking performance.

• Applied Microscopy
• /
• v.50
• /
• pp.23.1-23.9
• /
• 2020

We propose an effective deep learning model to denoise scanning transmission electron microscopy (STEM) image series, named Noise2Atom, to map images from a source domain 𝓢 to a target domain 𝓒, where 𝓢 is for our noisy experimental dataset, and 𝓒 is for the desired clear atomic images. Noise2Atom uses two external networks to apply additional constraints from the domain knowledge. This model requires no signal prior, no noise model estimation, and no paired training images. The only assumption is that the inputs are acquired with identical experimental configurations. To evaluate the restoration performance of our model, as it is impossible to obtain ground truth for our experimental dataset, we propose consecutive structural similarity (CSS) for image quality assessment, based on the fact that the structures remain much the same as the previous frame(s) within small scan intervals. We demonstrate the superiority of our model by providing evaluation in terms of CSS and visual quality on different experimental datasets.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.2
• /
• pp.44.1-44.1
• /
• 2019

We generate new intermediate images between observed consecutive solar images using NVIDIA's SuperSloMo that is a novel video interpolation method. This technique creates intermediate frames between two successive frames to form a coherent video sequence for both spatially and temporally. By using SuperSloMo, we create 600 images (12-second interval) using the observed 121 SDO/AIA 304 Å images (1-minute interval) of a filament eruption event on December 3, 2012. We compare the generated images with the original 12-second images. For the generated 480 images the correlation coefficient (CC), the relative error (R1), and the normalized mean square error (R2) are 0.99, 0.40, and 0.86, respectively. We construct a video made of the generated images and find a smoother erupting movement. In addition, we generate nonexistent 2.4-second interval images using the original 12-second interval images, showing slow motions in the eruption. We will discuss possible applications of this method.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.6
• /
• pp.19-27
• /
• 2010

This paper proposes a method for enhancing detection performance and reducing computational cost when detecting a human face by applying B-spline active contour to the frame difference of consecutive images. Firstly, the method estimates amount of user's motion using kurtosis. If the kurtosis is smaller than a pre-defined threshold, it is considered that the amount of user's motion is insufficient and thus the contour fitting is not applied. Otherwise, the contour fitting is applied by exploiting the fact that the amount of motion is sufficient. Secondly, for the contour fitting, difference edges are detected by combining the distance transformation of the binarized frame difference and the edges of current frame. Lastly, the face is located by assigning the contour fitting process to the detected difference edges. Kurtosis-based motion amount estimation can reduce a computational cost and stabilize the results of the contour fitting. In addition, distance transformation-based difference edge detection can enhance the problems of contour lag and discontinuous difference edges. Experimental results confirm that the proposed method can reduce the face localization error caused by the contour lag and discontinuity of edges, and decrease the computational cost by omitting approximately 39% of the contour fitting.

• Journal of Korea Game Society
• /
• v.12 no.5
• /
• pp.79-88
• /
• 2012

In this paper, we present a real-time hand pose recognition method to provide an intuitive user interface through hand poses or movements without a keyboard and a mouse. For this, the areas of right and left hands are segmented from the depth camera image, and noise removal is performed. Then, the rotation angle and the centroid point of each hand area are calculated. Subsequently, a circle is expanded at regular intervals from a centroid point of the hand to detect joint points and end points of the finger by obtaining the midway points of the hand boundary crossing. Lastly, the matching between the hand information calculated previously and the hand model of previous frame is performed, and the hand model is recognized to update the hand model for the next frame. This method enables users to predict the hidden fingers through the hand model information of the previous frame using temporal coherence in consecutive frames. As a result of the experiment on various hand poses with the hidden fingers using both hands, the accuracy showed over 95% and the performance indicated over 32 fps. The proposed method can be used as a contactless input interface in presentation, advertisement, education, and game applications.