• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.656-658
• /
• 2022

The purpose of infrared and visible image fusion is to integrate images of different modes with different details into a result image with rich information, which is convenient for high-level computer vision task. Considering many deep networks only work in a single scale, this paper proposes a novel image fusion based on three-scale dense network to preserve the content and key target features from the input images in the fused image. It comprises an encoder, a three-scale block, a fused strategy and a decoder, which can capture incredibly rich background details and prominent target details. The encoder is used to extract three-scale dense features from the source images for the initial image fusion. Then, a fusion strategy called l₁-norm to fuse features of different scales. Finally, the fused image is reconstructed by decoding network. Compared with the existing methods, the proposed method can achieve state-of-the-art fusion performance in subjective observation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.9
• /
• pp.2361-2376
• /
• 2023

The electrocardiogram (ECG) signal is commonly used to screen and diagnose cardiovascular diseases. In recent years, deep neural networks have been regarded as an effective way for automatic ECG disease diagnosis. The convolutional neural network is widely used for ECG signal extraction because it can obtain different levels of information. However, most previous studies adopt single scale convolution filters to extract ECG signal features, ignoring the complementarity between ECG signal features of different scales. In the paper, we propose a dual-scale network with convolution filters of different sizes for 12-lead ECG classification. Our model can extract and fuse ECG signal features of different scales. In addition, different spatial and time periods of the feature map obtained from the 12-lead ECG may have different contributions to ECG classification. Therefore, we add a spatial-temporal attention to each scale sub-network to emphasize the representative local spatial and temporal features. Our approach is evaluated on PTB-XL dataset and achieves 0.9307, 0.8152, and 89.11 on macro-averaged ROC-AUC score, a maximum F1 score, and mean accuracy, respectively. The experiment results have proven that our approach outperforms the baselines.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.11
• /
• pp.5465-5470
• /
• 2012

FPGA(Field Programmable Gate Array) devices are widely used due to their merits in circuit development time, and development cost. Among various FPGA technologies, SRAM-based FPGAs have large cell capacity so that they are attractive for complex circuit design and their reconfigurability. However, they are weak in space environment where radiation energy particles cause Single Event Upset(SEU). In this paper, we designed a controller supervising SRAM-based FPGA to protect configuration memory inside. The controller is implemented on an Anti-Fusing FPGA. Radiation test was performed on the implemented computer board and the result show that our controller provides better SEU-resilience than TMR-only system.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.1
• /
• pp.72-77
• /
• 2008

Identification of humans from multiple view points is an important task for surveillance and security purposes. For optimal performance the system should use the maximum information available from sensors. Multimodal biometric systems are capable of utilizing more than one physiological or behavioral characteristic for enrollment, verification, or identification. Since gait alone is not yet established as a very distinctive feature, this paper presents an approach to fuse face and gait for identification. In this paper we will use the single camera case i.e both the face and gait recognition is done using the same set of images captured by a single camera. The aim of this paper is to improve the performance of the system by utilizing the maximum amount of information available in the images. Fusion in considered at decision level. The proposed algorithm is tested on the NLPR database.

• Journal of Broadcast Engineering
• /
• v.19 no.2
• /
• pp.166-173
• /
• 2014

Recently, human action recognition have been developed for various broadcasting and video process. Since a video can consist of various scenes, keypoint approaches have been more attracted than template based methods for real application. Keypoint approahces tried to find regions having motion in video, and made 3-dimensional patches. Then, descriptors using histograms were computed from the patches, and a classifier based on machine learning method was applied to detect actions in video. However, a single classifier was difficult to handle various human actions. In order to improve this problem, approaches using multi classifiers were used to detect and to recognize objects. Thus, we propose a new human action recognition using decision-level fusion with support vector machine and sparse representation. The proposed method extracted descriptors based on keypoint approach from a video, and acquired results from each classifier for human action recognition. Then, we applied weights which were acquired by training stage to fuse each results from two classifiers. The experiment results in this paper show better result than a previous fusion method.

• Smart Structures and Systems
• /
• v.16 no.2
• /
• pp.295-328
• /
• 2015

In recent years the monitoring of structural behavior through acquisition of vibrational data has become common practice. In addition, recent advances in sensor development have made the collection of diverse dynamic information feasible. Other than the commonly collected acceleration information, Global Position System (GPS) receivers and non-contact, optical techniques have also allowed for the synchronous collection of highly accurate displacement data. The fusion of this heterogeneous information is crucial for the successful monitoring and control of structural systems especially when aiming at real-time estimation. This task is not a straightforward one as measurements are inevitably corrupted with some percentage of noise, often leading to imprecise estimation. Quite commonly, the presence of noise in acceleration signals results in drifting estimates of displacement states, as a result of numerical integration. In this study, a new approach based on a time domain identification method, namely the Unscented Kalman Filter (UKF), is proposed for correcting the "drift effect" in displacement or rotation estimates in an online manner, i.e., on the fly as data is attained. The method relies on the introduction of artificial white noise (WN) observations into the filter equations, which is shown to achieve an online correction of the drift issue, thus yielding highly accurate motion data. The proposed approach is demonstrated for two cases; firstly, the illustrative example of a single degree of freedom linear oscillator is examined, where availability of acceleration measurements is exclusively assumed. Secondly, a field inspired implementation is presented for the torsional identification of a tall tower structure, where acceleration measurements are obtained at a high sampling rate and non-collocated GPS displacement measurements are assumed available at a lower sampling rate. A multi-rate Kalman Filter is incorporated into the analysis in order to successfully fuse data sampled at different rates.

• Fashion & Textile Research Journal
• /
• v.15 no.2
• /
• pp.294-301
• /
• 2013

Sustainable fashion & textile is more than eco fashion & textile with the concepts for the next generation's happiness, prosumer value, and community responsibility. This study considers methods to enhance fashion and textile industry sustainability in accordance to clothing types (material, product life and washing properties) and to investigate company strategies. Company strategies are of redesign with stock, volunteering & measuring trash amount, participation by evaluation stores, clerk environment education, hiring QC specialist and reinforcing partnerships. For the case of daily innerwear, throwing away and recycling is more efficient for the environment than laundering in the consumer use stage; subsequently, we recommend the use of polypropylene fiber (a cheap and an eco-friendly material) for this item that can be recycled and reformed after use. For the case of single layer clothing (such as sportswear, blouse or pants) we recommend the use of thermoplastic materials with welding or fuse assembling technology instead of a sewing method of seams as well as the recycle design that is simply melted and reformed into new clothing without an after use dissembling process. Secondhand use or resale is suitable for denim/jean items if the clothing has a storytelling or private history tag. Lastly, module-type jacket or coat shows the variety of styles with one clothing worn w/o collar or sleeve details and changed into vest/coat; in addition, it is possible to add or partly tear off some jacket/coat fibers of the felt material to reform it into a new design.

• Journal of Information Processing Systems
• /
• v.10 no.1
• /
• pp.1-22
• /
• 2014

Recent technological advances provide the opportunity to use large amounts of multimedia data from a multitude of sensors with different modalities (e.g., video, text) for the detection and characterization of criminal activity. Their integration can compensate for sensor and modality deficiencies by using data from other available sensors and modalities. However, building such an integrated system at the scale of neighborhood and cities is challenging due to the large amount of data to be considered and the need to ensure a short response time to potential criminal activity. In this paper, we present a system that enables multi-modal data collection at scale and automates the detection of events of interest for the surveillance and reconnaissance of criminal activity. The proposed system showcases novel analytical tools that fuse multimedia data streams to automatically detect and identify specific criminal events and activities. More specifically, the system detects and analyzes series of incidents (an incident is an occurrence or artifact relevant to a criminal activity extracted from a single media stream) in the spatiotemporal domain to extract events (actual instances of criminal events) while cross-referencing multimodal media streams and incidents in time and space to provide a comprehensive view to a human operator while avoiding information overload. We present several case studies that demonstrate how the proposed system can provide law enforcement personnel with forensic and real time tools to identify and track potential criminal activity.

• Journal of Industrial and Engineering Chemistry
• /
• v.67
• /
• pp.358-364
• /
• 2018

An isotropic pitch precursor for fabricating carbon fibres was prepared by co-carbonization of ethylene bottom oil(EBO) and polyvinyl chloride (PVC). Various pre-treatments of EBO and PVC, and a high heating rate of $3^{\circ}C/min$ with no holding time, were evaluated for their effects on the oxidative stabilization process and the mechanical stability of the resulting fibres. Our stabilization process enhanced the volatilization, oxidative reaction and decomposition properties of the precursor pitch, while the addition of PVC both decreased the onset time and accelerated the oxidative reaction. Aliphatic carbon groups played a critical role in stabilization. Microstructural characterization indicated that these were first oxidised to carbon-oxygen single bonds and then converted to carbon-oxygen double bonds. Due to the higher heating rate and lack of a holding step during processing,the resulting thermoplastic fibers did not completely convert to thermoset materials, allowing partially melted, adjacent fibres to fuse. Fiber surfaces were smooth and homogeneous. Of the various methods evaluated herein, carbon fibers derived from pressure-treated EBO and PVC exhibited the highest tensile strength. This work shows that enhancing the naphthenic component of a pitch precursor through the co-carbonization of pre-treated EBO with PVC improves the oxidative properties of the resulting carbon fibers.

• Korean Journal of Remote Sensing
• /
• v.33 no.2
• /
• pp.149-158
• /
• 2017

Fine temporal and spatial resolution of image data are necessary to monitor the phenology of vegetation. However, there is no single sensor provides fine temporal and spatial resolution. For solve this limitation, researches on spatiotemporal data fusion methods are being conducted. Among them, FSDAF (Flexible spatiotemporal data fusion) can fuse each band in high accuracy.In thisstudy, we applied MODIS NDVI and Landsat NDVI to enhance time resolution of NDVI based on FSDAF algorithm. Then we proposed the possibility of utilization in vegetation phenology monitoring. As a result of FSDAF method, the predicted NDVI from January to December well reflect the seasonal characteristics of broadleaf forest, evergreen forest and farmland. The RMSE values between predicted NDVI and actual NDVI (Landsat NDVI) of August and October were 0.049 and 0.085, and the correlation coefficients were 0.765 and 0.642 respectively. Spatiotemporal data fusion method is a pixel-based fusion technique that can be applied to variousspatial resolution images, and expected to be applied to various vegetation-related studies.