• Journal of Broadcast Engineering
• /
• v.28 no.3
• /
• pp.333-336
• /
• 2023

Recently, the transformer model used in natural language processing is also applied to the image super resolution field, showing good performance. However, these transformer based models have a disadvantage that they are difficult to use in small mobile devices because they are complex and have many learning parameters and require high hardware resources. Therefore, in this paper, we propose a knowledge distillation technique that can effectively reduce the size of a transformer based super resolution model. As a result of the experiment, it was confirmed that by applying the proposed technique to the student model with reduced number of transformer blocks, performance similar to or higher than that of the teacher model could be obtained.

• Journal of the Korea Computer Graphics Society
• /
• v.30 no.2
• /
• pp.11-19
• /
• 2024

Demand for small biomimetic robots that can carry out reconnaissance missions without being exposed to the enemy in underground spaces and narrow passages is increasing in order to increase the fighting power and survivability of soldiers in wartime situations. A small compound eye image sensor for environmental recognition has advantages such as small size, low aberration, wide angle of view, depth estimation, and HDR that can be used in various ways in the field of vision. However, due to the small lens size, the resolution is low, and the problem of resolution in the fused image obtained from the actual compound eye image occurs. This paper proposes a compound eye image quality enhancement algorithm based on Image Enhancement and ESRGAN to overcome the problem of low resolution. If the proposed algorithm is applied to compound eye image fusion images, image resolution and image quality can be improved, so it is expected that performance improvement results can be obtained in various studies using compound eye cameras.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.31 no.5
• /
• pp.466-471
• /
• 2011

Optical coherence tomography(OCT) is an emerging medical diagnostic tool that draws great attention in medical and biological fields. It has a 10-100 times higher spatial resolution than that of the clinical ultrasound but lower imaging depth such as 1-2 mm. In order to image internal organs, OCT needs an endoscopic probe. In this paper, the principle of Fourier-domain optical coherence tomography with high-speed imaging capability was introduced. An OCT endoscope based on MEMS technology was developed. It was attached to the Fourier-domain OCT system to acquire three-dimensional tomographic images of gastrointestinal tract of New Zealand white rabbit. The endoscope had a two-axis scanning mirror that was driven by electrostatic force. The mirror stirred an incident light to sweep two-dimensional plane by scanning. The outer diameter of the endoscope was 6 mm and the mirror diameter was 1.2 mm. A three-dimensional image rendered by 200 two-dimensional tomographs with $200{\times}500$ pixels was displayed within 3.5 seconds. The spatial resolution of the OCT system was 8 ${\mu}m$ in air.

• Journal of KIISE:Computer Systems and Theory
• /
• v.26 no.11
• /
• pp.1285-1295
• /
• 1999

본 논문은 방대한 크기의 볼륨 데이타를 효율적으로 렌더링하기 위한 셀 기반 웨이브릿 압축 방법을 제시한다. 이 방법은 볼륨을 작은 크기의 셀로 나누고, 셀 단위로 웨이브릿 변환을 한 다음 복원 순서에 따른 런-길이(run-length) 인코딩을 수행하여 높은 압축율과 빠른 복원을 제공한다. 또한 최근 복원 정보를 캐쉬 자료 구조에 효율적으로 저장하여 복원 시간을 단축시키고, 에러 임계치의 정규화로 비정규화된 웨이브릿 압축보다 빠른 속도로 정규화된 압축과 같은 고화질의 이미지를 생성하였다. 본 연구의 성능을 평가하기 위하여 {{}} 해상도의 볼륨 데이타를 압축하여 쉬어-？ 분해(shear-warp factorization) 알고리즘에 적용한 결과, 손상이 거의 없는 상태로 약 27:1의 압축율이 얻어졌고, 약 3초의 렌더링 시간이 걸렸다.Abstract This paper presents an efficient cell-based wavelet compression method of large volume data. Volume data is divided into individual cell of {{}} voxels, and then wavelet transform is applied to each cell. The transformed cell is run-length encoded according to the reconstruction order resulting in a fairly good compression ratio and fast reconstruction. A cache structure is used to speed up the process of reconstruction and a threshold normalization scheme is presented to produce a higher quality rendered image. We have combined our compression method with shear-warp factorization, which is an accelerated volume rendering algorithm. Experimental results show the space requirement to be about 27:1 and the rendering time to be about 3 seconds for {{}} data sets while preserving the quality of an image as like as using original data.

• Journal of Advanced Navigation Technology
• /
• v.27 no.5
• /
• pp.495-501
• /
• 2023

P-3C/K aircraft for maritime patrols that Republic of Korea Navy is using, is equipped with a variety of sensors and communication devices. Collected data from the aircraft is managed as tactical information by flight operators and stored. When the flight mission is completed, this information is transferred to tactical support center on the ground and played back or used for follow-up work through a analysis tool. During a flight mission, there are tens of thousands of detection objects within an hour in KADIZ (Korea air defense identification zone). In contrast, in TSD (tactical situation display), which displays a map when using the analysis tool, all detected objects are expressed as symbols. The increase in display symbols has a significant impact on the TSD image updating and consequently interferes with the smooth operation of operators. In this paper, we propose applying multiple threads and multiple layers to improve the performance of existing TSD. And the performance improvement is proven through the execution results.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.318-318
• /
• 2016

위성은 준 실시간으로 국토 전체의 관측과 미계측/비접근 지역의 관측도 가능하여 가뭄, 홍수 등 수재해와 관련된 분석 자료로 활용되고 있으며, 위성 기반의 수재해 모니터링 적용성에 대한 연구 또한 수행되고 있다. 위성에서 관측된 자료는 NASA, JAXA 등의 위성 관리 센터에서 알고리즘을 적용하여 인터넷으로 제공하고, 최근 K-water에서는 수자원분야의 위성활용을 위해 위성 자료 수집 시스템을 갖추어 Aqua/Terra MODIS, GPM, GCOM-W1 등의 위성 자료를 수집하고 있다. 위성 자료는 5분~16일 등의 다양한 주기로 제공되고 있으며, 자료 타입, 측정 시간 등의 간단한 정보만 파일명으로 표시되어 위성의 위치(경위도) 및 해당 지점의 위성 자료를 얻기 위해서는 위성 자료를 확인해야만 하는 번거로움이 따른다. 본 연구에서는 순차적으로 관측된 위성 자료의 시 공간적 속성정보를 추출하고 해당 정보를 영상과 함께 맵핑하여, 시간의 흐름에 따른 위성 궤도의 시각화 방안을 제시하였다. 위성 궤도의 시각화 방안으로 사용된 위성 자료는 Terra MODIS의 'MOD02SSH', GPM GMI 센서의 'GPROF' 자료 타입을 사용하였다. 'MOD02SSH'는 5분 동안 5km의 공간해상도로 측정한 자료가 1개의 파일이며, 'GPROF'는 5분 동안 4km의 공간해상도로 측정한다. 공전 주기의 검증을 위해 케플러의 제3법칙을 적용한 Terra 위성의 공전주기는 98.75분으로 계산되며, 위성 자료의 공전주기는 98.87분으로 나타난다. 검증 결과 약 0.12초의 오차가 발생하며, 정확한 위성 고도와 높은 해상도의 위성 자료를 통해 오차의 감소가 가능하다. 이를 통해 시각화 된 동적 시계열 이미지는 시간에 따른 위성 궤도의 정보를 추출 할 수 있다. 이는 수재해 정보시스템의 모니터링을 위해 사용 가능하고, 시간에 따른 위성 궤도 정보를 통하여 필요한 시간대의 위성 위치 정보, 해당 지점의 관측 자료를 효율적으로 수집하여 자료 수집을 위한 시간 단축이 가능하며, 사용자 또는 관리자를 위한 모니터링 수행 또한 효율적인 운영이 가능할 것으로 사료된다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.6
• /
• pp.107-112
• /
• 2022

In an embedded system such as a drone, it is difficult to store, transfer and analyze the entire hyper-spectral image to a server in real time because it takes a lot of power and time. Therefore, the hyper-spectral image data is transmitted to the server through dimension reduction or compression pre-processing. Feature selection method are used to send only the bands for analysis purpose, and these algorithms usually take a lot of processing time depending on the size of the image, even though the efficiency is high. In this paper, by improving the temporal disadvantage of the band selection algorithm, the time taken 24 hours was reduced to around 60-180 seconds based on the 40000*682 image resolution of 8GB data, and the use of 7.6GB RAM was significantly reduced to 2.3GB using 45 out of 150 bands. However, in terms of pixel classification performance, more than 98% of analysis results were derived similarly to the previous one.

• Korean Journal of Materials Research
• /
• v.32 no.8
• /
• pp.345-353
• /
• 2022

In this study, using deep learning, super-resolution images of transmission electron microscope (TEM) images were generated for nanomaterial analysis. 1169 paired images with 256 × 256 pixels (high resolution: HR) from TEM measurements and 32 × 32 pixels (low resolution: LR) produced using the python module openCV were trained with deep learning models. The TEM images were related to DyVO₄ nanomaterials synthesized by hydrothermal methods. Mean-absolute-error (MAE), peak-signal-to-noise-ratio (PSNR), and structural similarity (SSIM) were used as metrics to evaluate the performance of the models. First, a super-resolution image (SR) was obtained using the traditional interpolation method used in computer vision. In the SR image at low magnification, the shape of the nanomaterial improved. However, the SR images at medium and high magnification failed to show the characteristics of the lattice of the nanomaterials. Second, to obtain a SR image, the deep learning model includes a residual network which reduces the loss of spatial information in the convolutional process of obtaining a feature map. In the process of optimizing the deep learning model, it was confirmed that the performance of the model improved as the number of data increased. In addition, by optimizing the deep learning model using the loss function, including MAE and SSIM at the same time, improved results of the nanomaterial lattice in SR images were achieved at medium and high magnifications. The final proposed deep learning model used four residual blocks to obtain the characteristic map of the low-resolution image, and the super-resolution image was completed using Upsampling2D and the residual block three times.

• Geophysics and Geophysical Exploration
• /
• v.25 no.3
• /
• pp.140-161
• /
• 2022

Seismic data with missing traces are often obtained regularly or irregularly due to environmental and economic constraints in their acquisition. Accordingly, seismic data interpolation is an essential step in seismic data processing. Recently, research activity on machine learning-based seismic data interpolation has been flourishing. In particular, convolutional neural network (CNN) and generative adversarial network (GAN), which are widely used algorithms for super-resolution problem solving in the image processing field, are also used for seismic data interpolation. In this study, CNN-based algorithm, U-Net and GAN-based algorithm, and conditional Wasserstein GAN (cWGAN) were used as seismic data interpolation methods. The results and performances of the methods were evaluated thoroughly to find an optimal interpolation method, which reconstructs with high accuracy missing seismic data. The work process for model training and performance evaluation was divided into two cases (i.e., Cases I and II). In Case I, we trained the model using only the regularly sampled data with 50% missing traces. We evaluated the model performance by applying the trained model to a total of six different test datasets, which consisted of a combination of regular, irregular, and sampling ratios. In Case II, six different models were generated using the training datasets sampled in the same way as the six test datasets. The models were applied to the same test datasets used in Case I to compare the results. We found that cWGAN showed better prediction performance than U-Net with higher PSNR and SSIM. However, cWGAN generated additional noise to the prediction results; thus, an ensemble technique was performed to remove the noise and improve the accuracy. The cWGAN ensemble model removed successfully the noise and showed improved PSNR and SSIM compared with existing individual models.