• 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.

• International Journal of Computer Science & Network Security
• /
• v.21 no.12spc
• /
• pp.463-468
• /
• 2021

Single image super-resolution (SISR) is an image processing technique, and its main target is to reconstruct the high-quality or high-resolution (HR) image from the low-quality or low-resolution (LR) image. Currently, deep learning-based convolutional neural network (CNN) image super-resolution approaches achieved remarkable improvement over the previous approaches. Furthermore, earlier approaches used hand designed filter to upscale the LR image into HR image. The design architecture of such approaches is easy, but it introduces the extra unwanted pixels in the reconstructed image. To resolve these issues, we propose novel deep learning-based approach known as Lightweight deep CNN-based approach for Single Image Super-Resolution (LDCSIR). In this paper, we propose a new architecture which is inspired by ResNet with Inception blocks, which significantly drop the computational cost of the model and increase the processing time for reconstructing the HR image. Compared with the other state of the art methods, LDCSIR achieves better performance in terms of quantitively (PSNR/SSIM) and qualitatively.

• 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 Institute of Electronics Engineers of Korea SP
• /
• v.45 no.6
• /
• pp.117-124
• /
• 2008

The approaches to overcome the limited spatial resolution and the limited dynamic range of image sensors have been studied independently. A high resolution image is reconstructed from multiple low resolution observations and a wide dynamic range image is reconstructed from differently exposed multiple low dynamic range in es based on signal processing approach. In practical situations, it is reasonable to address them in a unified context because the recorded image suffers from limitations of both spatial resolution and dynamic range. In this paper, the image acquisition process including limited spatial resolution and limited dynamic range is modelled. With the image acquisition model, the response function of the imaging system is estimated and the single image of which spatial resolution and dynamic range are simultaneously enhanced is obtained. Experimental results indicate that the proposed algorithm outperforms the conventional approaches that perform the high resolution and wide dynamic range reconstruction sequentially with respect to both objective and subjective criteria.

• Proceedings of the KSRS Conference
• /
• 2008.03a
• /
• pp.137-139
• /
• 2008

This study presents an approach to reconstruct high-resolution imagery for multispectral imagery of low-resolution using panchromatic imagery of high-resolution. The proposed scheme reconstructs a high-resolution image which agrees with original spectral values. It uses a linear model of high-and low- resolution images and consists of two stages. In this study, an 1m RGB image was generated from 4m IKONOS multispectral data.

• IEIE Transactions on Smart Processing and Computing
• /
• v.1 no.1
• /
• pp.17-26
• /
• 2012

This paper presents an image processing-based validation method for unrecognizable numbers in severely distorted license plate images which have been degraded by various factors including low-resolution, low light-level, geometric distortion, and periodic noise. Existing vehicle license plate recognition (LPR) methods assume that most of the image degradation factors have been removed before performing the recognition of printed numbers and letters. If this is not the case, conventional LPR becomes impossible. The proposed method adopts a novel approach where a set of reference number images are intentionally degraded using the same factors estimated from the input image. After a series of image processing steps, including geometric transformation, super-resolution, and filtering, a comparison using cross-correlation between the intentionally degraded reference and the input images can provide a successful identification of the visually unrecognizable numbers. The proposed method makes it possible to validate numbers in a license plate image taken under low light-level conditions. In the experiment, using an extended set of test images that are unrecognizable to human vision, the proposed method provides a successful recognition rate of over 95%, whereas most existing LPR methods fail due to the severe distortion.

• ETRI Journal
• /
• v.32 no.3
• /
• pp.390-394
• /
• 2010

In this paper, we propose a new super-resolution technique based on interpolation of the high-frequency subband images obtained by discrete wavelet transform (DWT) and the input image. The proposed technique uses DWT to decompose an image into different subband images. Then the high-frequency subband images and the input low-resolution image have been interpolated, followed by combining all these images to generate a new super-resolved image by using inverse DWT. The proposed technique has been tested on Lena, Elaine, Pepper, and Baboon. The quantitative peak signal-to-noise ratio (PSNR) and visual results show the superiority of the proposed technique over the conventional and state-of-art image resolution enhancement techniques. For Lena's image, the PSNR is 7.93 dB higher than the bicubic interpolation.

• 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 Electronics Engineers of Korea SP
• /
• v.38 no.2
• /
• pp.169-178
• /
• 2001

The demand for high-resolution images is gradually increasing, whereas many imaging systems have been designed to allow a certain level of aliasing during image acquisition. Thus, digital image processing approaches have recently been investigated to reconstruct a high-resolution image from aliased low-resolution images. However, since the sub-pixel motion information is assumed to be accurate in most conventional approaches, the satisfactory high-resolution image cannot be obtained when the sub-pixel motion information is inaccurate. Therefore, in this paper we propose a new algorithm to reduce the distortion in the reconstructed high-resolution image due to the inaccuracy of sub-pixel motion information. For this purpose, we analyze the effect of inaccurate sub-pixel motion information on a high-resolution image reconstruction, and model it as zero-mean additive Gaussian errors added respectively to each low-resolution image. To reduce the distortion we apply the modified multi-channel image deconvolution approach to the problem. The validity of the proposed algorithm is both theoretically and experimentally demonstrated in this paper.

• Korean Journal of Remote Sensing
• /
• v.39 no.4
• /
• pp.395-407
• /
• 2023

Super-resolution (SR) has great significance in image processing because it enables downstream vision tasks with high spatial resolution. Recently, SR studies have adopted deep learning networks and achieved remarkable SR performance compared to conventional example-based methods. Deep-learning-based SR models generally require low-resolution (LR) images and the corresponding high-resolution (HR) images as training dataset. Due to the difficulties in obtaining real-world LR-HR datasets, most SR models have used only HR images and generated LR images with predefined degradation such as bicubic downsampling. However, SR models trained on simple image degradation do not reflect the properties of the images and often result in deteriorated SR qualities when applied to real-world images. In this study, we propose an image degradation model for HR satellite images based on the modulation transfer function (MTF) of an imaging sensor. Because the proposed method determines the image degradation based on the sensor properties, it is more suitable for training SR models on remote sensing images. Experimental results on HR satellite image datasets demonstrated the effectiveness of applying MTF-based filters to construct a more realistic LR-HR training dataset.