• Journal of the Korea Computer Graphics Society
• /
• v.29 no.5
• /
• pp.31-40
• /
• 2023

As display resolution increases, many apps also tend to include high-resolution texture maps. Recent advancements in deep-learning-based image super-resolution techniques make it possible to automate high-resolution texture generation. However, there is still a lack of comprehensive analysis of the application of these techniques to texture maps. In this paper, we selected three recent super-resolution techniques, namely BSRGAN, Real-ESRGAN, and SwinIR (classical and real-world image SR), and applied them to upscale texture maps. We then conducted a quantitative and qualitative analysis of the experimental results. The findings revealed various artifacts after upscaling, which indicates that there are still limitations in directly applying super-resolution techniques to texture-map upscaling.

• Proceedings of the KSRS Conference
• /
• v.1
• /
• pp.426-429
• /
• 2006

Flood caused by Typhoon and severe rain during summer is the most destructive natural disasters in Korea. Almost every year flood has resulted in a big lost of national infrastructure and loss of civilian lives. It usually takes time and great efforts to estimate the flood-related damages. Government also has pursued proper standard and tool for using state-of-art technologies. High resolution satellite imagery is one of the most promising sources of ground truth information since it provides detailed and current ground information such as building, road, and bare ground. Once high resolution imagery is utilized, it can greatly reduce the amount of field work and cost for flood related damage assessment. The classification of high resolution image is pre-required step to be utilized for the damage assessment. The classified image combined with additional data such as DEM and DSM can help to estimate the flooded areas per each classified land use. This paper applied object-oriented classification scheme to interpret an image not based in a single pixel but in meaningful image objects and their mutual relations. When comparing it with other classification algorithms, object-oriented classification was very effective and accurate. In this paper, IKONOS image is used, but similar level of high resolution Korean KOMPSAT series can be investigated once they are available.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.5
• /
• pp.120-124
• /
• 2017

The temporal super resolution is a method for increasing the frame rate. Electro-optical sensors are used in various surveillance and reconnaissance weapons systems, and the spatial resolution and temporal resolution of the required electro-optical sensors vary according to the performance requirement of each weapon system. Because most image sensors capture images at 30~60 frames/second, it is necessary to increase the frame rate when the target moves and changes rapidly. This paper proposes a method to increase the frame rate using color channel extrapolation. Using a DMD, one frame of a general camera was adjusted to have different consecutive exposure times for each channel, and the captured image was converted to a single channel image with an increased frame rate. Using the optical flow method, a virtual channel image was generated for each channel, and a single channel image with an increased frame rate was converted to a color channel image. The performance of the proposed temporal super resolution method was confirmed by the simulation.

• Journal of Korean Society for Geospatial Information Science
• /
• v.22 no.3
• /
• pp.155-160
• /
• 2014

This paper deals with comparison of classification accuracy between three land cover classification results having difference in resolution and they were classified with eight classes including building, road, forest, etc. Airborne hyperspectral image used in this study was acquired at 1000m, 2000m, 3000m elevation and had 24 bands(0.5m spatial resolution), 48 bands(1.0m), 96 bands(1.5m). Assessment of classification accuracy showed that the classification using 48 bands hyperspectral image had outstanding result as compared with other images. For using hyperspectral image, it was verified that 1m spatial resolution image having 48 bands was appropriate to classify land cover and qualitative improvement is expected in thematic map creation using airborne hyperspectral image.

• Journal of Broadcast Engineering
• /
• v.19 no.1
• /
• pp.109-117
• /
• 2014

This paper presents a superresolution restoration technique that can restore high-resolution images from differently blurred low resolution images rather than using the motion information between low-resolution images. In order to restore the super-resolution image the rotatable aperture mask lens system is proposed. The proposed technique does not need to estimate point spread function at each frame. In addition, it does not require image registration because there is no global translational motion between low resolution images. By using a rotatable rectangular aperture, two consecutive captured images provide sufficiently exclusive information for superresolution. Therefore, the proposed method can reduce the registration error between the low-resolution image as well as the calculation amount for superresolution restoration. The existing lens system of the camera can be extended to obtain a superresolution image by only adding an rotatable rectangular aperture mask. Finally, in order to verify the performance of the proposed system, experimental results are performed. By comparing with the existing superresolution methods, the proposed method showed the significant improvements in the sense of spatial resolution.

• Journal of the Institute of Convergence Signal Processing
• /
• v.22 no.4
• /
• pp.156-163
• /
• 2021

Attention mechanisms have been used in deep learning-based computer vision systems, including single image super-resolution (SISR) networks. However, existing SISR networks with attention mechanism focused on real image super-resolution, so it is hard to know whether they are available for CG or VR images. In this paper, we attempt to apply a recent attention module, called balanced attention mechanism (BAM) module, to 12 state-of-the-art SISR networks, and then check whether the BAM module can achieve performance improvement in CG or VR image super-resolution. In our experiments, it has been confirmed that the performance improvement in CG or VR image super-resolution is limited and depends on data characteristics, size, and network type.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.2
• /
• pp.45-50
• /
• 2010

An image interpolation is a technique to use for enhancement of image resolution, it have two problems which are image quality degradation of the interpolated result image and high computation complexity. In this paper, to solve the problem, we propose an image interpolation algorithm using loss information estimation and implement the proposed method on portable device. From reduction image of obtained low resolution image, the proposed method can computes error to use image interpolated and estimate loss information by interpolation of the computed error. The estimated loss information is added to interpolated high resolution image with weight factor. We verified that the proposed method has improved FSNR as 2dB than conventional algorithms by experiments. Also, we implemented the proposed method on portable device and checked up real-time action. The proposed algorithm may be helpful for various application for image enlargement and reconstruction.

• Journal of Korea Water Resources Association
• /
• v.43 no.2
• /
• pp.153-165
• /
• 2010

Large-Scale Particle Image Velocimetry (LSPIV) is an extension of particle image velocimetry (PIV) for measurement of flows spanning large areas in laboratory or field conditions. LSPIV is composed of six elements - seeding, illumination, recording, image transformation, image processing, postprocessing - based on PIV. Possible error elements at each step of Mobile LSPIV (MLSPIV), which is a mobile version of LSPIV, in field applications are identified and summarized the effect of the errors which were quantified in the previous studies. The total number of elemental errors is 27, and five error sources were evaluated previously, seven elemental errors are not effective to the current MLSPIV system. Among 15 elemental errors, four errors - sampling time, image resolution, tracer, and wind - are investigated through an experiment at a laboratory to figure out how those errors affect to velocity calculation. The analysis to figure out the effect of the number of images used for image processing on the velocity calculation error shows that if over 50 images or more are used, the error due to it goes below 1 %. The effect of the image resolution on velocity calculation was investigated through various image resolution using digital camera. Low resolution image set made 3 % of velocity calculation error comparing with high resolution image set as a reference. For the effect of tracers and wind, the wind effect on tracer is decreasing remarkably with increasing the flume bulk velocity. To minimize the velocity evaluation error due to wind, tracers with high specific gravity is favorable.

• Journal of Korea Multimedia Society
• /
• v.22 no.11
• /
• pp.1242-1250
• /
• 2019

Super resolution technique aims to convert a low-resolution image with coarse details to a corresponding high-resolution image with refined details. In the past decades, the performance is greatly improved due to progress of deep learning models. However, universal solution for various objects is a still challenging issue. We observe that learning super resolution with a general dataset has poor performance on faces. In this paper, we propose a super resolution fusion scheme that works well for both general- and face datasets to achieve more universal solution. In addition, object-specific feature extractor is employed for better reconstruction performance. In our experiments, we compare our fusion image and super-resolved images from one- of the state-of-the-art deep learning models trained with DIV2K and FFHQ datasets. Quantitative and qualitative evaluates show that our fusion scheme successfully works well for both datasets. We expect our fusion scheme to be effective on other objects with poor performance and this will lead to universal solutions.

• Journal of Korea Multimedia Society
• /
• v.21 no.12
• /
• pp.1425-1430
• /
• 2018

In this paper, we propose a CNN based deep learning algorithm for semantic segmentation of images. In order to improve the accuracy of semantic segmentation, we combined pixel level object classification and image level object classification. The image level object classification is used to accurately detect the characteristics of an image, and the pixel level object classification is used to indicate which object area is included in each pixel. The proposed network structure consists of three parts in total. A part for extracting the features of the image, a part for outputting the final result in the resolution size of the original image, and a part for performing the image level object classification. Loss functions exist for image level and pixel level classification, respectively. Image-level object classification uses KL-Divergence and pixel level object classification uses cross-entropy. In addition, it combines the layer of the resolution of the network extracting the features and the network of the resolution to secure the position information of the lost feature and the information of the boundary of the object due to the pooling operation.