• IEIE Transactions on Smart Processing and Computing
• /
• v.6 no.2
• /
• pp.85-92
• /
• 2017

This paper presents a dehazing method based on a fuzzy membership function and variational method. The proposed algorithm consists of three steps: i) estimate transmission through a pixel-based operation using a fuzzy membership function, ii) refine the transmission using an L1-norm-based regularization method, and iii) obtain the result of haze removal based on a hazy image formation model using the refined transmission. In order to prevent color distortion of the sky region seen in conventional methods, we use a trapezoid-type fuzzy membership function. The proposed method acquires high-quality images without halo artifacts and loss of color contrast.

• Proceedings of the KIPE Conference
• /
• 2019.07a
• /
• pp.537-539
• /
• 2019

Recently, large-scale fires have been generated as urban buildings have become more and more density. Especially, the expansion of smoke in buildings due to high-rise is an problem, and the smoke is the main cause of death in fires. Therefore, in this paper, the image-based smoke detection is proposed through deep learning-based artificial intelligence techniques to prevent possible damage if existing detectors are not detected. In addition, the detection model was not configured simply through only the smoke data set, but the data set in the haze form was additionally composed together to compensate for the accuracy.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.113-117
• /
• 2020

Images were taken under various weather such as rain, haze, snow often show low visibility, which can dramatically decrease accuracy of some tasks in computer vision: object detection, segmentation. Besides, previous work to enhance image usually downsample the image to receive consistency features but have not yet good upsample algorithm to recover original size. So, in this research, we jointly implement removal streak in heavy rain image and super resolution using a deep network. We put forth a 2-stage network: a multi-model network followed by a refinement network. The first stage using rain formula in the single image and two operation layers (addition, multiplication) removes rain streak and noise to get clean image in low resolution. The second stage uses refinement network to recover damaged background information as well as upsample, and receive high resolution image. Our method improves visual quality image, gains accuracy in human action recognition task in datasets. Extensive experiments show that our network outperforms the state of the art (SoTA) methods.

• Current Optics and Photonics
• /
• v.7 no.5
• /
• pp.569-573
• /
• 2023

We utilized terahertz time-domain spectroscopy (THz-TDS) to measure the thickness and electrical properties of nickel-chromium (Ni-Cr) films. This technique not only aligns well with traditional methods, such as haze-meter and transmission-densitometer measurements, but it also reveals the electrical properties and thickness of films down to a few tens of nanometers. The complex conductivity of the Ni-Cr thin films was extracted using the Tinkham formula. The experimental values closely aligned with the Drude model, indicating the reliability of our Ni-Cr film's electrical and optical constants. The thickness of Ni-Cr was estimated using the complex conductivity. These findings emphasize the potential of THz-TDS in quality control of metallic nanofilms, pointing toward an efficient and nondestructive test (NDT) for such analyses.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.12
• /
• pp.4420-4438
• /
• 2021

Image dehazing is an ill-posed problem which is far from being solved. Traditional image dehazing methods often yield mediocre effects and possess substandard processing speed, while modern deep learning methods perform best only in certain datasets. The haze removal effect when processed by said methods is unsatisfactory, meaning the generalization performance fails to meet the requirements. Concurrently, due to the limited processing speed, most dehazing algorithms cannot be employed in the industry. To alleviate said problems, a lightweight fast dehazing network based on a multiple scale-patch framework (MSP) is proposed in the present paper. Firstly, the multi-scale structure is employed as the backbone network and the multi-patch structure as the supplementary network. Dehazing through a single network causes problems, such as loss of object details and color in some image areas, the multi-patch structure was employed for MSP as an information supplement. In the algorithm image processing module, the image is segmented up and down for processed separately. Secondly, MSP generates a clear dehazing effect and significant robustness when targeting real-world homogeneous and nonhomogeneous hazy maps and different datasets. Compared with existing dehazing methods, MSP demonstrated a fast inference speed and the feasibility of real-time processing. The overall size and model parameters of the entire dehazing model are 20.75M and 6.8M, and the processing time for the single image is 0.026s. Experiments on NTIRE 2018 and NTIRE 2020 demonstrate that MSP can achieve superior performance among the state-of-the-art methods, such as PSNR, SSIM, LPIPS, and individual subjective evaluation.

• Journal of the Korean earth science society
• /
• v.38 no.2
• /
• pp.129-140
• /
• 2017

Global solar radiation was calculated in this research using ground-base measurement data, meteorological satellite data, and GWNU (Gangneung-Wonju National University) solar radiation model. We also analyzed the accuracy of the GWNU model by comparing the observed solar radiation according to the total cloud cover. Our research was based on the global solar radiation of the GWNU radiation site in 2012, observation data such as temperature and pressure, humidity, aerosol, total ozone amount data from the Ozone Monitoring Instrument (OMI) sensor, and Skyview data used for evaluation of cloud mask and total cloud cover. On a clear day when the total cloud cover was 0 tenth, the calculated global solar radiations using the GWNU model had a high correlation coefficient of 0.98 compared with the observed solar radiation, but root mean square error (RMSE) was relatively high, i.e., $36.62Wm^{-2}$. The Skyview equipment was unable to determine the meteorological condition such as thin clouds, mist, and haze. On a cloudy day, regression equations were used for the radiation model to correct the effect of clouds. The correlation coefficient was 0.92, but the RMSE was high, i.e., $99.50Wm^{-2}$. For more accurate analysis, additional analysis of various elements including shielding of the direct radiation component and cloud optical thickness is required. The results of this study can be useful in the area where the global solar radiation is not observed by calculating the global solar radiation per minute or time.

• Journal of Korean Society for Atmospheric Environment
• /
• v.27 no.2
• /
• pp.201-208
• /
• 2011

Recently, a real-time, pocket-sized aethalometer (microAeth$^{(R)}$ model AE51) has been developed by Magee Scientific Inc. for measuring the concentration of black carbon in the atmosphere. In this study, two aethalometers, models AE-16 and AE-51, which measure the optical absorption of carbon particles at infrared 880 nm, were operated at time interval of 5-min between January 9 and February 10, 2010 at an urban site of Gwangju, to compare the accuracy of black carbon (BC) concentrations reported from the AE-51 model and to investigate reasonable sampling time of filter media in the AE-51. The air samples in the AE-51 and AE-16 models are collected on T60 (Teflon coated glass fiber) filter media (filter spot area: 0.07 $cm^2$) and quartz fiber roll-tape filter (filter spot area: 1.67 $cm^2$), respectively. Real-time measurement results indicate that when the filters were clean, the AE-51 BC was greater than or similar to the AE-16 BC data. However as the filter spots become darker, the AE-16 BC concentrations were higher than the AE-51 BC data and the difference in the BC concentrations from two AE models becomes gradually increased. Relative error in the AE-51 and AE-16 BC concentrations showed significance difference depending on used time of the filter in the AE-51 model, weather pattern, levels of air pollution, etc, ranging from 11.5% (used time of the filter in AE-51: 1,595 min) to 52.5% (used time of the filter in AE-51: 2,085 min). When considering the used time of one filter ticket in the AE-51 model and difference (or relative error %) between AE-16 and AE-51 BC concentrations, it is recommended that the standard sampling time per one filter ticket within the AE-51 model be less than approximately 24 hr (1,440 min) under the normal weather conditions except for severe haze and mist events.

• Journal of the Korean earth science society
• /
• v.43 no.2
• /
• pp.283-302
• /
• 2022

Despite the nationwide COVID-19 lockdown in China since January 23, 2020, haze days with high PM₁₀ levels of 88-98 ㎍ m^-3 occurred on February 1 and 2, 2020. During these haze days, the East Asian region was affected by a warm and stagnant air mass with positive air temperature anomalies and negative zonal wind anomalies at 850 hPa. The Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) was used to analyze the variation of regional PM₁₀ aerosol transport in Korea due to decreased anthropogenic emissions in East Asia. The base experiment (BASE), which applies the basic anthropogenic emissions in the WRF-Chem model, and the control experiment (CTL) applied by reducing the anthropogenic emission to 50%, were used to assess uncertainty with ground-based PM₁₀ measurements in Korea. The index of agreement (IOA) for the CTL simulation was 0.71, which was higher than that of BASE (0.67). A statistical analysis of the results suggests that anthropogenic emissions were reduced during the COVID-19 lockdown period in China. Furthermore, BASE and CTL applied to zero-out anthropogenic emissions outside Korea (BASE_ZEOK and CTL_ZEOK) were used to analyze the variations of regional PM₁₀ aerosol transport in Korea. Regional PM₁₀ transport in CTL was reduced by only 10-20% compared to BASE. Synthetic weather variables may be another reason for the non-linear response to changes in the contribution of regional transport to PM₁₀ in Korea with the reduction of anthropogenic emissions in East Asia. Although the regional transport contribution of other inorganic aerosols was high in CTL (80-90%), sulfate-nitrate-ammonium (SNA) aerosols showed lower contributions of 0-20%, 30-60%, and 30-60%, respectively. The SNA secondary aerosols, particularly nitrates, presumably declined as the Chinese lockdown induced traffic.

• Journal of Convergence for Information Technology
• /
• v.12 no.3
• /
• pp.214-221
• /
• 2022

The purpose of this study is to produce works with motifs from organic forms of nature and to try to drive various forms. As a research method, literature was investigated through rsearch in domestic academic jounals, and four hair sculptures that can be used easily in practice were produced by combining the organic forms and design concepts of nature. As a result of the study, first, the work was produced by applying the law of harmony and balance with the motif of the crescent moon. Second, the flow of wave waves was expressed by applying the law of repetition with the motif of waves. Third, the haze rising from the waterfall was expressed by applying the law of dissonance. Fourth, l tried to give the image of the sun announcing a new beginning by emphasizing the rising sun. As a result, it can be seen that the organic form of nature gives an embarrassing inspiration to create design. The limitation of this study is that the accuracy of producing works only with mannequins and producing the same image in the actual model should be limited, and as a follow-up study, research that can be used in the actual model should be continuously conducted.