• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.385-398
• /
• 2021

The complexity of deep learning models affects the real-time performance of gesture recognition, thereby limiting the application of gesture recognition algorithms in actual scenarios. Hence, a residual learning neural network based on a deep convolutional neural network is proposed. First, small convolution kernels are used to extract the local details of gesture images. Subsequently, a shallow residual structure is built to share weights, thereby avoiding gradient disappearance or gradient explosion as the network layer deepens; consequently, the difficulty of model optimisation is simplified. Additional convolutional neural networks are used to accelerate the refinement of deep abstract features based on the spatial importance of the gesture feature distribution. Finally, a fully connected cascade softmax classifier is used to complete the gesture recognition. Compared with the dense connection multiplexing feature information network, the proposed algorithm is optimised in feature multiplexing to avoid performance fluctuations caused by feature redundancy. Experimental results from the ISOGD gesture dataset and Gesture dataset prove that the proposed algorithm affords a fast convergence speed and high accuracy.

• Nuclear Engineering and Technology
• /
• v.55 no.6
• /
• pp.2276-2282
• /
• 2023

We deploy artificial neural networks to unfold neutron spectra from measured energy-integrated quantities. These neutron spectra represent an important parameter allowing to compute the absorbed dose and the kerma to serve radiation protection in addition to nuclear safety. The built architectures are inspired from convolutional neural networks. The first architecture is made up of residual transposed convolution's blocks while the second is a modified version of the U-net architecture. A large and balanced dataset is simulated following "realistic" physical constraints to train the architectures in an efficient way. Results show a high accuracy prediction of neutron spectra ranging from thermal up to fast spectrum. The dataset processing, the attention paid to performances' metrics and the hyper-optimization are behind the architectures' robustness.

• Economic and Environmental Geology
• /
• v.27 no.1
• /
• pp.41-47
• /
• 1994

A fast Hankel transform (FHT) algorithm (Anderson, 1982) is applied to numerical evaluation of many Green's tensor integrals encountered in three-dimensional electromagnetic modeling using integral equations. Efficient computation of Hankel transforms is obtained by a combination of related and lagged convolutions which are available in the FHT. We express Green's tensor integrals for a layered half-space, and rewrite those to a form of related functions so that the FHT can be applied in an efficient manner. By use of the FHT, a complete or full matrix of the related Hankel transform can be rapidly and accurately calculated for about the same computation time as would be required for a single direct convolution. Computing time for a five-layer half-space shows that the FHT is about 117 and 4 times faster than conventional direct and multiple lagged convolution methods, respectively.

• Journal of the Korean Chemical Society
• /
• v.55 no.2
• /
• pp.169-176
• /
• 2011

The electrochemical investigation of N,N-bis (2,5-di-tert-butylphenyl)-3,4,9,10 perylenebis (dicarboximide) laser dye have been carried out using cyclic voltammetry and convolution - deconvolution voltammetry combined with digital simulation technique at a platinum electrode in 0.1 mol/L tetrabutyl ammonium perchlorate (TBAP) in solvent 1,2 dichloroethane ($CH_2Cl-CH_2Cl$). The investigated dye was reduced via consumption of two sequential electrons to form radical anion and dianion (EE mechanism). In switching the potential to positive scan, the compound was oxidized by loss of two electrons, which were followed by a fast aggregation process ($EC_1EC_2$ mechanism). The electrode reaction pathway and the chemical and electrochemical parameters of the investigated compound were determined using cyclic voltammetry and convolutive voltammetry. The extracted electrochemical parameters were verified and confirmed via digital simulation method.

• International Journal of Computer Science & Network Security
• /
• v.23 no.3
• /
• pp.177-186
• /
• 2023

Hyperspectral imaging technology is one of the most efficient and fast-growing technologies in recent years. Hyperspectral image (HSI) comprises contiguous spectral bands for every pixel that is used to detect the object with significant accuracy and details. HSI contains high dimensionality of spectral information which is not easy to classify every pixel. To confront the problem, we propose a novel RGB channel Assimilation for classification methods. The color features are extracted by using chromaticity computation. Additionally, this work discusses the classification of hyperspectral image based on Domain Transform Interpolated Convolution Filter (DTICF) and 3D-CNN with Bi-directional-Long Short Term Memory (Bi-LSTM). There are three steps for the proposed techniques: First, HSI data is converted to RGB images with spatial features. Before using the DTICF, the RGB images of HSI and patch of the input image from raw HSI are integrated. Afterward, the pair features of spectral and spatial are excerpted using DTICF from integrated HSI. Those obtained spatial and spectral features are finally given into the designed 3D-CNN with Bi-LSTM framework. In the second step, the excerpted color features are classified by 2D-CNN. The probabilistic classification map of 3D-CNN-Bi-LSTM, and 2D-CNN are fused. In the last step, additionally, Markov Random Field (MRF) is utilized for improving the fused probabilistic classification map efficiently. Based on the experimental results, two different hyperspectral images prove that novel RGB channel assimilation of DTICF-3D-CNN-Bi-LSTM approach is more important and provides good classification results compared to other classification approaches.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.205-208
• /
• 2022

Multi-exposure high dynamic range (HDR) image reconstruction, the task of reconstructing an HDR image from multiple low dynamic range (LDR) images in a dynamic scene, often produces ghosting artifacts caused by camera motion and moving objects and also cannot deal with washed-out regions due to over or under-exposures. While there has been many deep-learning-based methods with motion estimation to alleviate these problems, they still have limitations for severely moving scenes. They also require large parameter counts, especially in the case of state-of-the-art methods that employ attention modules. To address these issues, we propose a frequency domain approach based on the idea that the transform domain coefficients inherently involve the global information from whole image pixels to cope with large motions. Specifically we adopt Residual Fast Fourier Transform (RFFT) blocks, which allows for global interactions of pixels. Moreover, we also employ Depthwise Overparametrized convolution (DO-conv) blocks, a convolution in which each input channel is convolved with its own 2D kernel, for faster convergence and performance gains. We call this LFFNet (Lightweight Frequency Fusion Network), and experiments on the benchmarks show reduced ghosting artifacts and improved performance up to 0.6dB tonemapped PSNR compared to recent state-of-the-art methods. Our architecture also requires fewer parameters and converges faster in training.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.6
• /
• pp.437-445
• /
• 2004

In this paper. we describe a novel method for transforming audio contents according to user's preference on sound field. Sound field effect technologies. which transform or simulate acoustic environments as user's preference, are very important for enlarging the reality of acoustic scene. However huge amount of computational power is required to process sound field effect in real time. so it is hard to implement this functionality at the portable audio devices such as MP3 player. In this paper, we propose an efficient method for providing sound field effect to audio contents independent of terminal's computational power through processing this functionality at the server using user's sound field preference, which is transfered from terminal side. To describe sound field preference, user can use perceptual acoustic parameters as well as the URI address of room impulse response signal. In addition, a novel fast convolution method is presented to implement a sound field effect engine as a result of convoluting with a room impulse response signal at the realtime application. and verified to be applicable to real-time applications through experiments. To verify the evidence of benefit of proposed method we performed two subjective listening tests about sound field descrimitive ability and preference on sound field processed sounds. The results showed that the proposed sound field preference can be applicable to the public.

• Geophysics and Geophysical Exploration
• /
• v.16 no.3
• /
• pp.139-144
• /
• 2013

A computer program has been developed to estimate time-domain electromagnetic (EM) responses for a onedimensional model with multiple source and receiver dipoles that are finite in length. The time-domain solution can be obtained by applying an inverse fast Fourier transform (FFT) to frequency-domain fields for efficiency. Frequency-domain responses are first obtained for 10 logarithmically equidistant frequencies per decade, and then cubic spline interpolated to get the FFT input. In the case of phases, the phase curve must be made to be continuous prior to the spline interpolation. The spline interpolated data are convolved with a source current waveform prior to FFT. In this paper, only a step-off waveform is considered. This time-domain code is verified with an analytic solution and EM responses for a marine hydrocarbon reservoir model. Through these comparisons, we can confirm that the accuracy of the developed program is fairly high.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.18 no.2
• /
• pp.104-113
• /
• 2019

The road management institute needs lots of cost to repair road surface damage. These damages are inevitable due to natural factors and aging, but maintenance technologies for efficient repair of the broken road are needed. Various technologies have been developed and applied to cope with such a demand. Recently, maintenance technology for road surface damage repair is being developed using image information collected in the form of a black box installed in a vehicle. There are various methods to extract the damaged region, however, we will discuss the image recognition technology of the deep neural network structure that is actively studied recently. In this paper, we introduce a new neural network which can estimate the road damage and its location in the image by region-based convolution neural network algorithm. In order to develop the algorithm, about 600 images were collected through actual driving. Then, learning was carried out and compared with the existing model, we developed a neural network with 10.67% accuracy.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.11
• /
• pp.1713-1719
• /
• 2022

As the noise canceler with a small processing delay increases the sampling frequency, a better-quality output can be obtained. For a single buffer, processing delay occurs because it is impossible to write new data while the processor is processing the data. When synthesizing with anti-noise and output signal, this processing delay creates additional buffering overhead to match the phase. In this paper, we propose an accelerator structure that minimizes processing delay and increases processing speed by alternately performing read and write operations using the Symmetric Even-Odd-buffer. In addition, we compare the structural differences between the two methods of noise cancellation (Fast Fourier Transform noise cancellation and adaptive Least Mean Square algorithm). As a result, using an Symmetric Even-Odd-buffer the processing delay was reduced by 29.2% compared to a single buffer. The proposed Symmetric Even-Odd-buffer structure has the advantage that it can be applied to various canceling algorithms.