• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.5
• /
• pp.2253-2272
• /
• 2018

We propose a deep learning method for multi-focus image fusion. Unlike most existing pixel-level fusion methods, either in spatial domain or in transform domain, our method directly learns an end-to-end fully convolutional two-stream network. The framework maps a pair of different focus images to a clean version, with a chain of convolutional layers, fusion layer and deconvolutional layers. Our deep fusion model has advantages of efficiency and robustness, yet demonstrates state-of-art fusion quality. We explore different parameter settings to achieve trade-offs between performance and speed. Moreover, the experiment results on our training dataset show that our network can achieve good performance with subjective visual perception and objective assessment metrics.

• Journal of Sensor Science and Technology
• /
• v.26 no.1
• /
• pp.1-6
• /
• 2017

The advanced driver assistant system (ADAS) requires the collection of a large amount of information including road conditions, environment, vehicle status, condition of the driver, and other useful data. In this regard, large-scale sensor networks can be an appropriate solution since they have been designed for this purpose. Recent advances in sensor network technology have enabled the management and monitoring of large-scale tasks such as the monitoring of road surface temperature on a highway. In this paper, we consider the estimation and fusion problems of the large-scale sensor networks used in the ADAS. Hierarchical fusion architecture is proposed for an arbitrary topology of the large-scale sensor network. A robust cluster estimator is proposed to achieve robustness of the network against outliers or failure of sensors. Lastly, a robust hierarchical data fusion scheme is proposed for the communication channel between the clusters and fusion center, considering the non-Gaussian channel noise, which is typical in communication systems.

• Proceedings of the Korean Statistical Society Conference
• /
• 2003.05a
• /
• pp.237-242
• /
• 2003

Various kinds of estimation methods have been developed for imputation of categorical missing data. They include modal category method, logistic regression, and association rule. In this study, we propose two imputation methods (neural network fusion and voting fusion) that combine the results of individual imputation methods. A Monte-Carlo simulation is used to compare the performance of these methods. Five factors used to simulate the missing data are (1) true model for the data, (2) data size, (3) noise size (4) percentage of missing data, and (5) missing pattern. Overall, neural network fusion performed the best while voting fusion is better than the individual imputation methods, although it was inferior to the neural network fusion. Result of an additional real data analysis confirms the simulation result.

• The KIPS Transactions:PartC
• /
• v.16C no.1
• /
• pp.83-90
• /
• 2009

Wireless Sensor Network (WSN) is a wireless network that gathers information from remote area with autonomously configured routing path. We propose a fusion based routing for a 'convergecast' in which all sensors periodically forward collected data to a base station. Previous researches dealt with only full-fusion or no-fusion case. Our Fusion rate based Spanning Tree (FST) can provide effective routing topology in terms of total cost according to all ranges of fusion rate f ($0{\leq}f{\leq}1$). FST is optimum for convergecast in case of no-fusion (f = 0) and full-fusion (f = 1) and outperforms the Shortest Path spanning Tree (SPT) or Minimum Spanning Tree (MST) for any range of f (0 < f < 1). Simulation of 100-node WSN shows that the total length of FST is shorter than MST and SPT nearby 31% and 8% respectively in terms of topology lengths for all range of f. As a result, we confirmed that FST is a very useful WSN topology.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.15 no.2
• /
• pp.147-154
• /
• 2012

In this paper, we revise the JDL data fusion model to have an ability of distributed data fusion(DDF). Data fusion is a function that produces valuable information using data from multiple sources. After the network centric warfare concept was introduced, the data fusion was required to be expanded to DDF. We identify the data transfer and control between nodes is the core function of DDF. The previous data fusion models can not be used for DDF because they don't include that function. Therefore, we revise the previous JDL data fusion model by adding the core function of DDF and propose this new model as a model for DDF. We show that our model is adequate and useful for DDF by using several examples.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.17 no.1
• /
• pp.77-85
• /
• 2008

The objective of the study is to construct a sensor fusion system for tool-condition monitoring (TCM) that will lead to a more efficient and economical drill usage. Drill-wear monitoring has an important attribute in the automatic machining processes as it can help preventing the damage of tools and workpieces, and optimizing the drill usage. In this study, we present the architectures of a multi-layer feed-forward neural network with Levenberg-Marquardt training algorithm based on sensor fusion for the monitoring of drill-wear condition. The input features to the neural networks were extracted from AE, vibration and current signals using the wavelet packet transform (WPT) analysis. Training and testing were performed at a moderate range of cutting conditions in the dry drilling of steel plates. The results show good performance in drill- wear monitoring by the proposed method of sensor fusion and neural network analysis.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.2461-2466
• /
• 2003

This paper presents a simple sensor fusion algorithm using neural network for navigation of mobile robots with obstacle avoidance and trap recovery. The multiple sensors input sensor data to the input layer of neural network activating the input nodes. The multiple sensors used include optical encoders, ultrasonic sensors, infrared sensors, a magnetic compass sensor, and GPS sensors. The proposed sensor fusion algorithm is combined with the VFH(Vector Field Histogram) algorithm for obstacle avoidance and AGPM(Adaptive Goal Perturbation Method) which sets adaptive virtual goals to escape trap situations. The experiment results show that the proposed low-level fusion algorithm is effective for real-time navigation of mobile robot.

• Journal of the military operations research society of Korea
• /
• v.34 no.3
• /
• pp.79-92
• /
• 2008

The purpose of this paper is to design an algorithm for data fusion of sensors data in the helicopter ASE system, using Bayesian Network, which was selected among several knowledge base data fusion methods after consideration and applied to this study. The result of the algorithm analysis shows that Bayesian Network is effective method for solving this problem.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.493-495
• /
• 2023

Deep learning methods and attention mechanisms have been incorporated to improve facial emotion recognition, which has recently attracted much attention. The fusion approaches have improved accuracy by combining various types of information. This research proposes a fusion network with self-attention and local attention mechanisms. It uses a multi-layer perceptron network. The network extracts distinguishing characteristics from facial images using pre-trained models on RAF-DB dataset. We outperform the other fusion methods on RAD-DB dataset with impressive results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1761-1777
• /
• 2021

An image with infrared features and visible details is obtained by processing infrared and visible images. In this paper, a fusion method based on Laplacian pyramid and generative adversarial network is proposed to obtain high quality fusion images, termed as Laplacian-GAN. Firstly, the base and detail layers are obtained by decomposing the source images. Secondly, we utilize the Laplacian pyramid-based method to fuse these base layers to obtain more information of the base layer. Thirdly, the detail part is fused by a generative adversarial network. In addition, generative adversarial network avoids the manual design complicated fusion rules. Finally, the fused base layer and fused detail layer are reconstructed to obtain the fused image. Experimental results demonstrate that the proposed method can obtain state-of-the-art fusion performance in both visual quality and objective assessment. In terms of visual observation, the fusion image obtained by Laplacian-GAN algorithm in this paper is clearer in detail. At the same time, in the six metrics of MI, AG, EI, MS_SSIM, Q_abf and SCD, the algorithm presented in this paper has improved by 0.62%, 7.10%, 14.53%, 12.18%, 34.33% and 12.23%, respectively, compared with the best of the other three algorithms.