• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.70-73
• /
• 2021

최근 영상 및 비디오 분야에 심층 신경망(DNN, Deep Neural Network)을 사용한 연구가 다양하게 진행됨에 따라 High Dynamic Range (HDR) 이미징 기술에서도 기존의 방법들 보다 우수한 성능을 보이는 심층 신경망 모델들이 등장하였다. 하지만, 심층 신경망을 사용한 방법은 큰 연산량과 많은 GPU 메모리를 사용한다는 문제점이 존재하며, 이는 심층 신경망 기반 기술들의 현실 적용 가능성에 제한이 되고 있다. 이에 본 논문에서는 제한된 연산량과 GPU 메모리 조건에서도 사용 가능한 다중 노출 HDR 경량화 심층 신경망을 제안한다. Kalantari Dataset에 대해 기존 HDR 모델들과의 성능 평가를 진행해 본 결과, PSNR-µ와 PSNR-l 수치에서 GPU 메모리 사용량 대비 우수한 성능을 보임을 확인하였다.

• ETRI Journal
• /
• v.45 no.4
• /
• pp.666-677
• /
• 2023

In the last few years, convolutional neural networks (CNNs) have demonstrated good performance while solving various computer vision problems. However, since CNNs exhibit high computational complexity, signal processing is performed on the server side. To reduce the computational complexity of CNNs for edge computing, a lightweight algorithm, such as a MobileNet, is proposed. Although MobileNet is lighter than other CNN models, it commonly achieves lower classification accuracy. Hence, to find a balance between complexity and accuracy, additional hyperparameters for adjusting the size of the model have recently been proposed. However, significantly increasing the number of parameters makes models dense and unsuitable for devices with limited computational resources. In this study, we propose a novel MobileNet architecture, in which the number of parameters is adaptively increased according to the importance of feature maps. We show that our proposed network achieves better classification accuracy with fewer parameters than the conventional MobileNet.

• The Journal of Korea Robotics Society
• /
• v.19 no.2
• /
• pp.188-195
• /
• 2024

This paper proposes a photorealistic real-time dense 3D mapping system that utilizes a neural network-based image enhancement method and mesh-based map representation. Due to the characteristics of the underwater environment, where problems such as hazing and low contrast occur, it is hard to apply conventional simultaneous localization and mapping (SLAM) methods. At the same time, the behavior of Autonomous Underwater Vehicle (AUV) is computationally constrained. In this paper, we utilize a neural network-based image enhancement method to improve pose estimation and mapping quality and apply a sliding window-based mesh expansion method to enable lightweight, fast, and photorealistic mapping. To validate our results, we utilize real-world and indoor synthetic datasets. We performed qualitative validation with the real-world dataset and quantitative validation by modeling images from the indoor synthetic dataset as underwater scenes.

• Annual Conference of KIPS
• /
• 2007.05a
• /
• pp.1237-1239
• /
• 2007

This paper proposes a virtual storage system for smart home/office embedded devices by utilizing AoE and HyperSCSI protocols. It relies on current Ethernet infrastructures and aims to provide a low-cost solution for the storage limitation problem on embedded devices. We choose AoE (ATA over Ethernet) and HyperSCSI in our implementation because they are relatively lightweight compared to other network storage protocols such as NFS(Network File System), CIFS(Common Internet File System) and iSCSI(internet Small Computer System Interface). The design and architecture of our proposed virtual system as well as the prototype demonstration are presented in this paper.

• The KIPS Transactions:PartC
• /
• v.13C no.1 s.104
• /
• pp.1-10
• /
• 2006

Ubiquitous computing, a new type of network environment has been generating much interest recently and has been actively studied. In ubiquitous computing, the sensor network which consists of low electric power ad-hoc network-based sensors and sensor nodes, is particularly the most important factor The sensor network serves as the mediator between ubiquitous computing and the actual environment. Related studies are focused on network . management through lightweight hardware using RFID. However, to apply these to actual environment, more practical scenarios as well as more secured studies equipped with secures and efficiency features are needed. Therefore, this study aims to build a wireless network based on PTD for multi users, which provides the largest utility in individual networks, and propose an appropriate management method. The proposed method is designed to enhance security and efficiency related to various services required in wireless networks, based on the reliable peripheral devices for users or pm. using pm, which has been applied to electronic commerce transactions in existing papers, this study also proposed an appropriate management method that is suitable for a dynamic environment and setting a temporary group to provide various services.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.15 no.5
• /
• pp.981-986
• /
• 2020

The Internet of Things refers to a space-of-things connection network configured to allow things with built-in sensors and communication functions to interact with people and other things, regardless of the restriction of place or time.IoT is a network developed for the purpose of services for human convenience, but the scope of its use is expanding across industries such as power transmission, energy management, and factory automation. However, the communication protocol of IoT, MQTT, is a lightweight message transmission protocol based on the push technology and has a security vulnerability, and this suggests that there are risks such as personal information infringement or industrial information leakage. To solve this problem, we designed a synchronous MQTT security channel that creates a secure channel by using the characteristic that different chaotic dynamical systems are synchronized with arbitrary values in the lightweight message transmission MQTT protocol. The communication channel we designed is a method of transmitting information to the noise channel by using characteristics such as random number similarity of chaotic signals, sensitivity to initial value, and reproducibility of signals. The encryption method synchronized with the proposed key value is a method optimized for the lightweight message transmission protocol, and if applied to the MQTT of IoT, it is believed to be effective in creating a secure channel.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.8
• /
• pp.1588-1596
• /
• 2015

In the Internet of Things (IoT), resource-constrained devices such as sensors are capable of communicating and exchanging data over the Internet. The IETF standard group has specified an application protocol CoAP, which uses UDP as a transport protocol, allows such a lightweight device to transmit data. Also, the IETF recommended the DTLS binding for securing CoAP. However, additional features should be added to the DTLS protocol to resolve several problems such as packet loss, reordering, fragmentation and replay attack. Consequently, performance of DTLS is worse than TLS. It is highly required for lightweight devices powered by small battery to design and implement a security protocol in an energy efficient manner. This paper thus discusses about DTLS performance in the perspective of energy consumption. To analyze the performance, we implemented IEEE 802.15.4 based test network consisting of constrained sensor devices in the Cooja simulator. We measured energy consumptions required for each of DTLS client and server in the test network. This paper compares the energy consumption and amount of transmitted data of each flight of DTLS handshake, and the processing and receiving time. We present the analyzed results with regard to code size, cipher primitive and fragmentation as well.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.11
• /
• pp.115-124
• /
• 2013

In this paper, we suggest a secure data dissemination by Lightweight Individual Encryption Multicast scheme over wireless sensor networks using the individual encryption method with Forward Error Correction instead of the group key encryption method. In wireless sensor networks, a sink node disseminates multicast data to the number of sensor nodes to update the up to date software such as network re-programming and here the group key encryption method is the general approach to provide a secure transmission. This group key encryption approach involves re-key management to provide a strong secure content distribution, however it is complicated to provide group key management services in wireless sensor networks due to limited resources of computing, storage, and communication. Although it is possible to control an individual node, the cost problem about individual encryption comes up and the individual encryption method is difficult to apply in multicast data transmission on wireless sensor networks. Therefore we only use 0.16% of individually encrypted packets to securely transmit data with the unicast to every node and the rest 99.84% non-encrypted encoded packets is transmitted with the multicast for network performance.

• Korean Journal of Remote Sensing
• /
• v.38 no.2
• /
• pp.167-177
• /
• 2022

In order to obtain satellite image products using the image transmitted to the ground station after capturing the satellite images, many image pre/post-processing steps are involved. During the pre/post-processing, when converting from level 1R images to level 1G images, geometric correction is essential. An interpolation method necessary for geometric correction is inevitably used, and the quality of the level 1G images is determined according to the accuracy of the interpolation method. Also, it is crucial to speed up the interpolation algorithm by the level processor. In this paper, we proposed a lightweight CNN-based interpolation method required for geometric correction when converting from level 1R to level 1G. The proposed method doubles the resolution of satellite images and constructs a deep learning network with a lightweight deep convolutional neural network for fast processing speed. In addition, a feature map fusion method capable of improving the image quality of multispectral (MS) bands using panchromatic (PAN) band information was proposed. The images obtained through the proposed interpolation method improved by about 0.4 dB for the PAN image and about 4.9 dB for the MS image in the quantitative peak signal-to-noise ratio (PSNR) index compared to the existing deep learning-based interpolation methods. In addition, it was confirmed that the time required to acquire an image that is twice the resolution of the 36,500×36,500 input image based on the PAN image size is improved by about 1.6 times compared to the existing deep learning-based interpolation method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.7
• /
• pp.1706-1725
• /
• 2024

The unique U-shaped structure of U-Net network makes it achieve good performance in image segmentation. This network is a lightweight network with a small number of parameters for small image segmentation datasets. However, when the medical image to be segmented contains a lot of detailed information, the segmentation results cannot fully meet the actual requirements. In order to achieve higher accuracy of medical image segmentation, a novel improved U-Net network architecture called multi-scale encoder-decoder U-Net+ (MEDU-Net+) is proposed in this paper. We design the GoogLeNet for achieving more information at the encoder of the proposed MEDU-Net+, and present the multi-scale feature extraction for fusing semantic information of different scales in the encoder and decoder. Meanwhile, we also introduce the layer-by-layer skip connection to connect the information of each layer, so that there is no need to encode the last layer and return the information. The proposed MEDU-Net+ divides the unknown depth network into each part of deconvolution layer to replace the direct connection of the encoder and decoder in U-Net. In addition, a new combined loss function is proposed to extract more edge information by combining the advantages of the generalized dice and the focal loss functions. Finally, we validate our proposed MEDU-Net+ MEDU-Net+ and other classic medical image segmentation networks on three medical image datasets. The experimental results show that our proposed MEDU-Net+ has prominent superior performance compared with other medical image segmentation networks.