• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1408-1414
• /
• 2019

Deep artificial neural network with transfer learning is applied to compressed sensing cardiovascular MRI. Transfer learning is a method that utilizes structure, filter kernels, and weights of the network used in prior learning for current learning or application. The transfer learning is useful in accelerating learning speed, and in generalization of the neural network when learning data is limited. From a cardiac MRI experiment, with 8 healthy volunteers, the neural network with transfer learning was able to reduce learning time by a factor of more than five compared to that with standalone learning. Using test data set, reconstructed images with transfer learning showed lower normalized mean square error and better image quality compared to those without transfer learning.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.39 no.10
• /
• pp.817-824
• /
• 2015

Experiments were carried out to investigate the flow and mass transfer characteristics of an orifice nozzle. Measurements of primary and suction flow rates, dissolved oxygen concentration, and electric power were obtained. Vertically injected mixed-jet images were captured by a direct visualization technique with a high speed camera unit. The mass ratio, volumetric mass transfer coefficient, and mass transfer performance were calculated using the measured data. As the primary flow pressure increases, the mass ratio decreases slightly, while the volumetric mass transfer coefficient and electric power increase. As the primary flow pressure increases and the mass ratio decreases, the mass transfer rate increases because of the fine bubbles and wider distribution of the bubbles.

• The Journal of the Korea Contents Association
• /
• v.10 no.4
• /
• pp.76-83
• /
• 2010

In this paper, we propose a method that converts a inputted real image to a image feeling like printmaking. That is, this method converts a inputted real image to man made rubber printmaking style image using image processing techniques such as spatial filters, image bit-block transfer, etc. The process is as follows. First, after detecting edges in source image, we get the first image by deleting noise lines and points, then by sharpening. Secondly, we get second image using the similar method to the first image. Finally, we blend the first and the second image by logical AND operation This processing enables us to represent rubber panel and knife effects. Also, the proposed method shows that double edge detecting is effective in enhancing line-width and removing the tiny lines.

• Korean Journal of Remote Sensing
• /
• v.27 no.5
• /
• pp.587-599
• /
• 2011

As one measure of image interpretability, NIIRS(National Imagery Interpretability Rating Scale) has been used. Unlike MTF(Modulation Transfer Function), SNR(Signal to Noise Ratio), and GSD(Ground Sampling Distance), NIIRS can describe the quality of overall image at user's perspective. NIIRS is observed with human observation directly or estimated by edge analysis. For edge analysis specially manufactured artificial target is used commonly. This target, formed with a tarp of black and white patterns, is deployed on the ground and imaged by the satellite. Due to this, the artificial target-based method needs a big expense and can not be performed often. In this paper, we propose a new edge analysis method that enables to estimate NIIRS accurately. In this method, natural targets available in the image are used and characteristics of the target are considered. For assessment of the algorithm, various experiments were carried out. The results showed that our algorithm can be used as an alternative to the artificial target-based method.

• Journal of the Korean Society of Radiology
• /
• v.5 no.6
• /
• pp.421-426
• /
• 2011

This research was accomplished to assess dose effects on image quality at computed radiography (CR). The ultimate target of the research was finding optimized exposure that provides necessary image quality for the clinical chest diagnosis. Modulation transfer function (MTF), normalized noise power spectrum (NNPS), and Noise equivalent quanta (NEQ) corresponding to the different doses were measured for the assessment of image quality. The preparation of "edge test device" used in MTF measurement and experimental geometry setup were followed by the recommendations of International Electrotechnical Commission (IEC). The experimental results show the necessary image quality can be achieved even at a half of the automatic exposure control (AEC) setting dose for chest diagnosis. It means that the patient exposure can be reduced dramatically by using optimized dose.

• Journal of the Society of Naval Architects of Korea
• /
• v.57 no.3
• /
• pp.140-151
• /
• 2020

It is important to identify the location of ship hull blocks with exact block identification number when scheduling the shipbuilding process. The wrong information on the location and identification number of some hull block can cause low productivity by spending time to find where the exact hull block is. In order to solve this problem, it is necessary to equip the system to track the location of the blocks and to identify the identification numbers of the blocks automatically. There were a lot of researches of location tracking system for the hull blocks on the stockyard. However there has been no research to identify the hull blocks on the stockyard. This study compares the performance of 5 Convolutional Neural Network (CNN) models with multi-view image set on the classification of the hull blocks to identify the blocks on the stockyard. The CNN models are open algorithms of ImageNet Large-Scale Visual Recognition Competition (ILSVRC). Four scaled hull block models are used to acquire the images of ship hull blocks. Learning and transfer learning of the CNN models with original training data and augmented data of the original training data were done. 20 tests and predictions in consideration of five CNN models and four cases of training conditions are performed. In order to compare the classification performance of the CNN models, accuracy and average F1-Score from confusion matrix are adopted as the performance measures. As a result of the comparison, Resnet-152v2 model shows the highest accuracy and average F1-Score with full block prediction image set and with cropped block prediction image set.

• Journal of radiological science and technology
• /
• v.46 no.1
• /
• pp.15-21
• /
• 2023

This study was purpose to assessment of the resolution characteristics by using ATS 535H Basic quality assurance (QA) phantom for ultrasound. The ultrasound equipment was used Logiq P6 (Ultrasound, GE Healthcare System, Chicago, IL, USA). And the ultrasound transducer were used Convex 4C (4~5.5 MHz), Linear 11L (10~13 MHz), Sector 3SP (3~5.5 MHz) probe. As for the noise power spectrum (NPS) comparison results by using ATS 535H Basic QA ultrasound phantom and Convex 4C, Linear 11L, Sector 3SP probe. The NPS value of the Convex 4C probe image was 0.0049, Linear 11L probe image was 0.0049, Sector 3SP probe image was 0.1422 when the frequency is 1.0 mm^-1. The modulation transfer function (MTF) comparison results by using ATS 535H Basic QA ultrasound phantom and Linear 11L probe the MTF value of the 3 cm focus image was 0.7511 and 4 cm focus image was 0.9001 when the frequency is 1.0 mm^-1. This study was presented characteristics of spatial resolution a quantitative evaluation methods by using ultrasound medical images for QA of ultrasound medical QA phantom. The quality control (QC) for equipment maintenance can be efficiently used in the clinic due to the quantitative evaluation of the NPS and MTF as the standard methods. It is meaningful in that it is applied mutatis mutandis and presented the results of physical resolution characteristics of the ultrasound medical image.

• The Journal of Korea Robotics Society
• /
• v.15 no.1
• /
• pp.1-7
• /
• 2020

In this paper, we introduce the methodology that utilizes deep learning-based front-end to enhance underwater feature matching. Both optical camera and sonar are widely applicable sensors in underwater research, however, each sensor has its own weaknesses, such as light condition and turbidity for the optic camera, and noise for sonar. To overcome the problems, we proposed the opti-acoustic transformation method. Since feature detection in sonar image is challenging, we converted the sonar image to an optic style image. Maintaining the main contents in the sonar image, CNN-based style transfer method changed the style of the image that facilitates feature detection. Finally, we verified our result using cosine similarity comparison and feature matching against the original optic image.

• The Journal of the Korea Contents Association
• /
• v.11 no.1
• /
• pp.83-91
• /
• 2011

The attribute of cartoon and comics has been known the combination of image and text. It is improved the we find the shape of comics at the first comics on the newspaper. But we can read the comic without word. These works don't give the difficult to read and it may transfer to readers by the image of figure on comics. Therefore how can the image reach the readers by the communication. This study is the research of the visual image communication and the attribute of wordless comics.

• Journal of the Optical Society of Korea
• /
• v.12 no.2
• /
• pp.79-87
• /
• 2008

We have developed a miniature fisheye lens with $190^{\circ}$ field of view operating simultaneously in the visible and the near infrared wavelengths. The modulation transfer function characteristic for the visible wavelength is sufficient for a mega-pixel-grade image sensor. The lens also has a fair resolution in the infrared wavelength region. The calibrated $f-{\theta}$ distortion is less than 5%, and the relative illumination is over 90%. In consequence, a sharp wide-angle image can be obtained which is uniform in brightness over the entire range of field angles. The real image heights for the visible and the near infrared wavelengths have been fitted to polynomial functions of incidence angle with sub-pixel accuracies. Combined with the near equidistance projection scheme of the lens, this lens can be advantageously employed in various image-processing applications requiring a wide-angle lens.