• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.11a
• /
• pp.781-784
• /
• 2017

Face recognition systems are characterized by low invasiveness of acquisition, and increasingly better reliability. Such systems may not be applied effectively, when the images are in low resolution (LR) as in the case that photos are taken from long distances, typically public surveillance. In theory, the high resolution (HR) image reconstructed from an LR face image, applying a super resolution (SR) method, can be used for face recognition. However, existing face SR algorithms may not give satisfactory results in face recognition. This article investigates the very low resolution face recognition problem and introduces a partial differential equation (PDE)-based SR method for a face recognition system of convolutional neural network (CNN).

• Journal of Information Processing Systems
• /
• v.19 no.4
• /
• pp.427-438
• /
• 2023

Plenty of works have indicated that single image super-resolution (SISR) models relying on synthetic datasets are difficult to be applied to real scene text image super-resolution (STISR) for its more complex degradation. The up-to-date dataset for realistic STISR is called TextZoom, while the current methods trained on this dataset have not considered the effect of multi-scale features of text images. In this paper, a multi-scale and attention fusion model for realistic STISR is proposed. The multi-scale learning mechanism is introduced to acquire sophisticated feature representations of text images; The spatial and channel attentions are introduced to capture the local information and inter-channel interaction information of text images; At last, this paper designs a multi-scale residual attention module by skillfully fusing multi-scale learning and attention mechanisms. The experiments on TextZoom demonstrate that the model proposed increases scene text recognition's (ASTER) average recognition accuracy by 1.2% compared to text super-resolution network.

• Proceedings of the KSRS Conference
• /
• v.1
• /
• pp.409-412
• /
• 2006

Today's commercial high resolution satellite imagery such as IKONOS and QuickBird, offers the potential to extract useful spatial information for geographical database construction and GIS applications. Recognizing this potential use of high resolution satellite imagery, KARI is performing a project for developing Korea multipurpose satellite 3(KOMPSAT-3). Therefore, it is necessary to develop techniques for various GIS applications of KOMPSAT-3, using similar high resolution satellite imagery. As fundamental studies for this purpose, we focused on the extraction of 3D spatial information and the update of existing GIS data from QuickBird imagery. This paper examines the scheme for rectification of high resolution image, and suggests the convenient semi-automatic algorithm for extraction of 3D building information from a single image. The algorithm is based on triangular vector structure that consists of a building bottom point, its corresponding roof point and a shadow end point. The proposed method could increase the number of measurable building, and enhance the digitizing accuracy and the computation efficiency.

• Imaging Science in Dentistry
• /
• v.50 no.4
• /
• pp.331-337
• /
• 2020

Purpose: As cone-beam computed tomography (CBCT) has become the most widely used 3-dimensional (3D) imaging modality in the dental field, storage space and costs for large-capacity data have become an important issue. Therefore, if 3D data can be stored at a clinically acceptable compression rate, the burden in terms of storage space and cost can be reduced and data can be managed more efficiently. In this study, a deep learning network for super-resolution was tested to restore compressed virtual CBCT images. Materials and Methods: Virtual CBCT image data were created with a publicly available online dataset (CQ500) of multidetector computed tomography images using CBCT reconstruction software (TIGRE). A very deep super-resolution (VDSR) network was trained to restore high-resolution virtual CBCT images from the low-resolution virtual CBCT images. Results: The images reconstructed by VDSR showed better image quality than bicubic interpolation in restored images at various scale ratios. The highest scale ratio with clinically acceptable reconstruction accuracy using VDSR was 2.1. Conclusion: VDSR showed promising restoration accuracy in this study. In the future, it will be necessary to experiment with new deep learning algorithms and large-scale data for clinical application of this technology.

• Journal of Multimedia Information System
• /
• v.8 no.2
• /
• pp.101-110
• /
• 2021

In computer vision, single-image super resolution has been an area of research for a significant period. Traditional techniques involve interpolation-based methods such as Nearest-neighbor, Bilinear, and Bicubic for image restoration. Although implementations of convolutional neural networks have provided outstanding results in recent years, efficiency and single model multi-scalability have been its challenges. Furthermore, previous works haven't placed enough emphasis on real-number scalability. Interpolation-based techniques, however, have no limit in terms of scalability as they are able to upscale images to any desired size. In this paper, we propose a convolutional neural network possessing the advantages of the interpolation-based techniques, which is also efficient, deeming it suitable in practical implementations. It consists of convolutional layers applied on the low-resolution space, post-up-sampling along the end hidden layers, and additional layers on high-resolution space. Up-sampling is applied on a multiple channeled feature map via bicubic interpolation using a single model. Experiments on architectural structure, layer reduction, and real-number scale training are executed with results proving efficient amongst multi-scale learning (including scale multi-path-learning) based models.

• The Journal of the Korea Contents Association
• /
• v.11 no.1
• /
• pp.236-244
• /
• 2011

Recently, the number of interventional procedures has increased dramatically as an alternative of invasive surgical procedure. The need for the quality control program of the angiographic units has also increased, because of concerns about the increased patient dose and the importance of image quality of angiographic units for the successful procedures. The purpose of this study was to propose an optimal guideline for the quality control program of the angiographic units. We reviewed domestic and international standards about medical imaging system and we evaluated the quality of 61 angiographic units in Korea with the use of NEMA 21 phantom. According to the results of our study, we propose a guideline for the quality control program of the angiographic units. Quality control program includes tube voltage test, tube current test, HVL test, image-field geometry test, spatial resolution test, low-contrast iodine detectability test, wire resolution test, phantom entrance dose test. Proposed reference levels are as follows: PAE < $\pm$ 10% in tube voltage test, PAE < $\pm$ 15% in tube current test, minimum 2.3 mmAl at 80 kVp in HVL test, minimum 'acceptable' level at image-field geometry test, 0.8 lp/mm for detector size of 34-40cm, 1.0 lp/mm for detector size of 28-33cm, 1.2 lp/mm for detector size of 22-27cm in spatial resolution test, minimum 200mg/cc in low contrast iodine detectability test, phantom entrance dose should be under 10R/min, 0.012 inch wire should be seen at static wire resolution test, and 0.022 inch wire should be seen at moving wire resolution test.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 1998.10a
• /
• pp.59-65
• /
• 1998

A thematic land cover map of Pusan port area was developed using Landsat satellite TM(Thematic Mapper) image. Two types of digital data which are road and sea water layer are extracted from existing paper map were overlayed over the developed land cover map. SPIN-2(KNR-1000) image was utilized to make a facility map of JaSungDae port. SPIN-2 image, which has a cell resolution of 1.56 m showed higer accuracy than TM image, which has a cell resolution of 30 m for facility mapping. Overall, the techniques of digital mapping using satellite image are very useful, effective and efficient.

• Journal of Veterinary Clinics
• /
• v.39 no.3
• /
• pp.114-120
• /
• 2022

The purpose of this study is to examine the status of quality control using multipurpose phantom of ultrasound equipment used in hospital of veterinary college in South Korea by using ATS-539 multipurpose phantom so as to examine quantitative and objective new image evaluation method. Specialists discussed and analyzed multipurpose phantom images acquired by using convex transducer of 10 ultrasound imaging devices, currently used in 9 veterinary colleges, at 4.0-6.0 MHz. Total 8 items that can be measured with ATS-539 multipurpose phantom including dead zone, vertical and horizontal measurement, axial/lateral resolution, sensitivity, focal zone, functional resolution and gray scale/dynamic range were evaluated. For qualitative evaluation, valid decisions were made based on dead zone, axial/lateral resolution, and gray scale/dynamic range which are resolution index, and coefficient of variation (COV) and blind referenceless image spatial quality evaluator (BRISQUE) were found to increase objectivity. As a result of experiment, all the targeted ultrasonic devices were found appropriate from qualitative evaluation items of dead zone, axial/lateral resolution, and gray scale/dynamic range. In other evaluation items, they were found to be appropriate from focal zone and vertical measurement of quantitative evaluation while inappropriate from horizontal measurement, sensitivity, and functional resolution. COV value was 0.12 ± 0.04, and BRISQUE value was 47.77 ± 2.77, both analysis results show that the noise level of all ultrasonic devices was located within tolerance range. Upon image examination using ATS-539 multipurpose phantom, they were 100% appropriate with inspection standards of dead zone, axial/lateral resolution, and gray scale/dynamic range, and besides, focal zone and functional resolution can be used as evaluation items. In the field of veterinary medicine, 8 standard items using ATS-539 multipurpose phantom and image evaluation items using COV and BRISQUE can be used as standards for quality control of ultrasonography machine.

• Journal of Korea Multimedia Society
• /
• v.21 no.11
• /
• pp.1271-1278
• /
• 2018

The NDT using ultrasonic is largely divided into A-Scan and C-Scan methods. Since A-Scan method is subject to subjective judgement by trained personnel, C-Scan method has been introduced, which presents the weld area in two dimensions by placing the transducers two dimensionally used in the A-Scan method. Therefore, it is necessary to develop equipment that can provide weld quality without the help of a welding expert and the presentation of effective C-Scan images. Thus, in this paper, the algorithms that express a low resolution 2-dimensional gray image formed by C-Scan method as a high-resolution color C-Scan image and automatically determine the weld quality from the generated C-Scan color image. The high resolution color C-Scan images proposed in this paper allow the exact shape of the weld point to be expressed, and an objective algorithm to use this image to automatically determine weld quality.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.254-257
• /
• 2009

In this paper, we propose a learning-based super-resolution algorithm. In the proposed algorithm, a multi-resolution wavelet approach is adopted to perform the synthesis of local high-frequency features. To obtain a high-resolution image, wavelet coefficients of two dominant LH- and HL-bands are estimated based on wavelet frames. In order to prepare more efficient training sets, the proposed algorithm utilizes the LH-band and transposed HL-band. The training sets are then used for the estimation of wavelet coefficients for both LH- and HL-bands. Using the estimated high frequency bands, a high resolution image is reconstructed via the wavelet transform. Experimental results demonstrate that the proposed scheme can synthesize high-quality images.