• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2191-2208
• /
• 2022

Motion blur in PET (Positron emission tomography) images induced by respiratory motion will reduce the quality of imaging. Although exiting methods have positive performance for respiratory motion correction in medical practice, there are still many aspects that can be improved. In this paper, an improved 3D unsupervised framework, Res-Voxel based on U-Net network was proposed for the motion correction. The Res-Voxel with multiple residual structure may improve the ability of predicting deformation field, and use a smaller convolution kernel to reduce the parameters of the model and decrease the amount of computation required. The proposed is tested on the simulated PET imaging data and the clinical data. Experimental results demonstrate that the proposed achieved Dice indices 93.81%, 81.75% and 75.10% on the simulated geometric phantom data, voxel phantom data and the clinical data respectively. It is demonstrated that the proposed method can improve the registration and correction performance of PET image.

• Journal of Korea Multimedia Society
• /
• v.25 no.5
• /
• pp.676-684
• /
• 2022

As realistic media are widespread in various image processing areas, image or video compression is one of the key technologies to enable real-time applications with limited network bandwidth. Generally, image or video compression cause the unnecessary compression artifacts, such as blocking artifacts and ringing effects. In this study, we propose a Deep Residual Channel-attention Network, so called DRCAN, which consists of an input layer, a feature extractor and an output layer. Experimental results showed that the proposed DRCAN can reduced the total memory size and the inference time by as low as 47% and 59%, respectively. In addition, DRCAN can achieve a better peak signal-to-noise ratio and structural similarity index measure for compressed images compared to the previous methods.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.80-81
• /
• 2021

딥 러닝 모델 사용에 있어서, 일반적인 사용자가 이용할 수 있는 하드웨어 리소스는 제한적이기 때문에 기존 모델을 경량화 할 수 있는 프루닝 방법을 통해 제한적인 리소스를 효과적으로 활용할 수 있도록 한다. 그 방법으로, 여러 딥 러닝 모델들 중 비교적 파라미터 수가 많은 것으로 알려진 GAN 아키텍처에 네트워크 프루닝을 적용함으로써 비교적 무거운 모델을 적은 파라미터를 통해 학습할 수 있는 방법을 제시한다. 또한, 본 논문을 통해 기존의 SRGAN 논문에서 가장 효과적인 결과로 제시했던 16 개의 residual block 의 개수를 실제로 줄여 봄으로써 기존 논문에서 제시했던 결과와의 차이에 대해 서술한다.

• 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.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.338-339
• /
• 2021

In this paper, we proposed a multiclass image scene classification method based on transform learning. The method using the Residual Network (ResNet) model which pre-trained on the large image dataset ImageNet for image classification. Compared with the image classification method of the CNN model, it can greatly improve the classification accuracy and efficiency

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.656-659
• /
• 2020

초해상화 딥러닝 기법은 학습 시 수렴하기까지 최소 수백 번의 에폭을 필요로 하며 오랜 시간이 걸린다. 최근, 영상 인식용 딥러닝 모델에서는 학습 수렴 속도를 향상시키기 위해 픽셀, 채널간 불필요한 상호연관 정보를 제거하는 Deconvolution 기술이 제안되었다. 본 논문에서는 최초로 Deconvolution 기술을 초해상화 딥러닝 방법에 적용하여 학습 수렴 속도 증가를 시도했다. 영상 인식 딥러닝 기법과 다르게 초해상화 딥러닝 기법은 이미지 특성 추출 부분과 이미지 복원 부분의 정보를 보존하는 것이 중요하기 때문에, EDSR을 Baseline 모델로 사용하여 양쪽 끝의 레이어는 기존의 Convolution 연산을 그대로 유지하고, 중간 레이어의 ResBlock 내의 Convolution 연산만 Deconvolution 연산으로 바꿔서 구성하였다. 초해상화 벤치마크 데이터셋을 사용한 실험 결과, 수렴속도가 빨라지지 않는 결과를 도출했다. 본 논문에서는 Deconvolution 기술이 Baseline 모델의 성능을 개선하지 못하는 이유를 초해상화 분야에서 기본적으로 적용되는 Residual Learning 기법 때문으로 분석했다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1322-1324
• /
• 2022

최근 딥러닝의 발전에 따라 단일 이미지 초해상도 분야에 좋은 성과를 보여주고 있다. 그러나 보다 더 높은 성능을 획득하기 위해 네트워크의 깊이 및 파라미터의 수가 크게 증가하였고, 모바일 및 엣지 디바이스에 원활하게 적용되기 위하여 딥러닝 모델 경량화의 필요성이 대두되고 있다. 이에 본 논문에서는 초해상도 모델 중 하나인 EDSR(Enhanced Deep Residual Network)에 대조 학습 기반 지식 전이를 적용한 경량화 기법을 제안한다. 실험 결과 제안한 지식 전이 기법이 기존의 다른 지식 증류 기법보다 향상된 성능을 보임을 확인하였다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.519-520
• /
• 2023

본 논문에서는 최근 반려동물 등록제가 확대되고 있는 바, 기존의 마이크로 칩 삽입 방법을 회피하고 반려견 이미지를 통하여 개체를 인식하는 방법을 연구하였다. 반려견의 전체 이미지를 학습시켜 해당 개체를 식별하는 지능형 시스템을 ResNet 알고리즘을 이용하여 구현하고, 수집된 반려견의 개체 사진을 학습시켜 필요한 개체를 식별할 수 있도록 하였다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.6
• /
• pp.55-61
• /
• 2014

This paper related with the ECMA (Enchanced CMA) algorithm performance which is possible to simultaneously compensation of the amplitude and phase by appling the minimum disturbance techniques in the CMA adatpve equalizer. The ECMA can improving the gradient noise amplification problem, stability and roburstness performance by the minimum disturbance technique that is the minimization of the equalizer tap weight variation in the point of squared euclidiean norm and the decision directed mode, and then the now cost function were proposed in order to simultaneouly compensation of amplitude and phase of the received signal with the minimum increment of computational operations. The performance of ECMA algorithm was compared to present MCMA by the computer simulation. For proving the performance, the recovered signal constellation that is the output of equalizer output signal and the residual isi and Maximum Distortion charateristic and MSE learning curve that are presents the convergence performance in the equalizer and the overall frequency transfer function of channel and equalizer were used. As a result of computer simulation, the ECMA has more better compensation capability of amplitude and phase in the recovered constellation, and the convergence time of adaptive equalization has improved compared to the MCMA.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.3
• /
• pp.63-68
• /
• 2014

This paper related with the I-CMA (Improved-CMA) algorithm that is possible to compensates of phase in CMA adatpve equalizer which is used for the elemination of intersymbol interference in the multipath fading and band limit characteristics of channel. The new cost function is proposed for the eliminate the amplitude and phase simulataneous by modifying the cost fuction for get the error signal in present CMA algorithm. It has a merit to the algorithm simplicities and eliminats the PLL device for phase compensation after equalization. For proving this, the recovered signal constellation that is the output of equalizer output signal and the residual isi and Maximum Distortion charateristic learning curve that are presents the convergence performance in the equalizer and the overall frequency transfer function of channel and equalizer were used. As a result of computer simulation, the I-CMA has more good compensation capability of amplitude and phas in the recovered constellation. But the convergence time is slow due to the simultaneously phase compensation.