• Journal of Information Processing Systems
• /
• 제18권4호
• /
• pp.575-586
• /
• 2022

Uneven light in real-world causes visual degradation for underexposed regions. For these regions, insufficient consideration during enhancement procedure will result in over-/under-exposure, loss of details and color distortion. Confronting such challenges, an unsupervised low-light image enhancement network is proposed in this paper based on the guidance of the unpaired low-/normal-light images. The key components in our network include super-resolution module (SRM), a GAN-based low-light image enhancement network (LLIEN), and denoising-scaling module (DSM). The SRM improves the resolution of the low-light input images before illumination enhancement. Such design philosophy improves the effectiveness of texture details preservation by operating in high-resolution space. Subsequently, local lightness attention module in LLIEN effectively distinguishes unevenly illuminated areas and puts emphasis on low-light areas, ensuring the spatial consistency of illumination for locally underexposed images. Then, multiple discriminators, i.e., global discriminator, local region discriminator, and color discriminator performs assessment from different perspectives to avoid over-/under-exposure and color distortion, which guides the network to generate images that in line with human aesthetic perception. Finally, the DSM performs noise removal and obtains high-quality enhanced images. Both qualitative and quantitative experiments demonstrate that our approach achieves favorable results, which indicates its superior capacity on illumination and texture details restoration.

• 방송공학회논문지
• /
• 제24권1호
• /
• pp.87-96
• /
• 2019

저조도 영상의 개선에 관한 연구는 대부분 대비 개선을 목적으로 한다. 저저도 영상에서 밝기 개선, 대조 개선, 및 조명 성분 감쇠 등의 다양한 연구가 진행됐다. 최근에 인공신경망으로 상기 방법들을 대체하는 연구가 진행 중이다. 본 논문에서는 Retinex 이론에 기반하여 조명 광원이 존재하는 저저도 영상으로부터 조명 성분을 감쇠하고, 반사 성분만을 생성하는 기법을 심층신경망으로 대체하는 방법을 제안한다. 실험에서는 102장의 저저도 영상으로 학습시킨 인공신경망으로 반사 영상을 생성하였는데, PSNR=30.8682(db), SSIM=0.4345를 얻었다.

• 로봇학회논문지
• /
• 제18권1호
• /
• pp.66-71
• /
• 2023

As cameras have become primary sensors for mobile robots, vision based Simultaneous Localization and Mapping (SLAM) has achieved impressive results with the recent development of computer vision and deep learning. However, vision information has a disadvantage in that a lot of information disappears in a low-light environment. To overcome the problem, we propose an image enhancement method to perform visual SLAM in a low-light environment. Using the deep generative adversarial models and modified gamma correction, the quality of low-light images were improved. The proposed method is less sharp than the existing method, but it can be applied to ORB-SLAM in real time by dramatically reducing the amount of computation. The experimental results were able to prove the validity of the proposed method by applying to public Dataset TUM and VIVID++.

• ETRI Journal
• /
• 제34권1호
• /
• pp.138-141
• /
• 2012

This letter proposes a real-time digital image stabilization system for cell phone cameras without the need for frame memory. The system post-processes an image captured with a safe shutter speed using an adaptive denoising filter and a global color correction algorithm. This system can transfer the normal brightness of an image previewed under long exposure to the captured image making it bright and crisp with low noise. It is even possible to take photos in low-light conditions. By not needing frame memory, the approach is feasible for integration into the size-constrained image sensors of cell phone cameras.

• 한국의류학회지
• /
• 제26권2호
• /
• pp.193-204
• /
• 2002

Many fashion forecasting companies propose the fashion colors in every season. Modern fashion consumer respond to fashionable trends with utmost sensitivity. Therefore to satisfy the consumer with an trendy image, the fashion design must be found first, as image matters, followed by an analysis of each design element's effect on the total image composition. In previous studies of fashion image, has been discussed the positive correlation between fashion design elements of color, fabric, and form as the central issue. In this thesis, two of the fashion design elements, color and fabric are simultaneously considered to classify the image of fabric in fashion. For the color variables, 10 hues are selected from Munsell's system of color notation, and 12 tones from PCCS color notation., which are currently used in the domestic fashion industry. Texture variables used in this survey are classified by luster, prominence-depression of surface, thickness, and density of fabric. Graduate students from 20 to 50 years old and the specialists in fashion companies participated in the survey. The results of this survey are as follows: 1. The fashion fabric image is classified as 5 main images: 'elegant', 'comfortable', 'characteristic', 'light'and 'simple'. 2. The influence of hue, tone and texture is significant to the fashion fabric image. Following colors, yellow-red, red hues and light grayish, dark grayish tones convey the elegant image. The texture property for the elegant image is luster, thin and low density. Properties of fabric conveying the comfortable image are yellow-red and green-yellow hue, soft, light tones, matte and high density. Furthermore, hue turned out to be a insignificant variables for the unique image, whereas dark grayish, grayish tone, luster and prominent texture convey a unique image. For light image, properties of fabric are blue-green, purple hues, light, bright tones with thin, low density texture. Properties of fabric conveying the simple image are blue-green, purple-blue, green-yellow hues, and strong, vivid tones, with luster and flat texture.

• 펄프종이기술
• /
• 제31권4호
• /
• pp.30-40
• /
• 1999

The possibility of the characterization of sheet formation by image analysis with transmitted light was evaluated. Specific perimenter, average perimeter and variation could not be used to predict the sheet formation because there were no corrleation. Although image analysis method still have a lot of problems , it was found that the contrast intensity obtained by image analysis with transmitted light can be used to predict the sheet formation. In the case of highly filled sheet, the intensity of transmitted light was too low to characterize the sheet formation . However, it was possible to characterize the formation of unfilled heavy weight paper($\leq$200g/㎡).

• 전기전자학회논문지
• /
• 제25권1호
• /
• pp.139-147
• /
• 2021

본 논문에서는 저조도 영상을 개선하기 위한 영상 분해 기반 심층 학습 방법 및 분해 채널 특성에 따른 손실함수를 제안한다. 기존 기법들의 문제점인 색신호 왜곡 및 할로 현상을 제거하기 위해, 입력 영상의 휘도 채널을 반사 성분과 조도 성분으로 분해하고, 반사 성분, 조도 성분 및 색차 신호를 신호 특성에 적합한 심층학습 과정을 적용하는 분해 기반 다중 구조 심층 학습 방법을 제안한다. 더불어, 분해 채널들의 특성에 따른 혼합 놈 기반의 손실함수를 정의하여 복원 영상의 안정성을 증대하고 열화 현상을 제거하기 위한 기법에 대해 기술한다. 실험 결과를 통해 제안한 방법이 다양한 저조도 영상을 효과적으로 개선하였음을 확인할 수 있었다.

• 한국철도학회논문집
• /
• 제13권3호
• /
• pp.283-289
• /
• 2010

디지털 카메라의 발전으로 인해 점차 영상을 사용한 철도의 안전관리기법이 그 사용범위를 넓히고 있다. 그러나 선로의 특성상 많은 저조도 환경에서의 영상 취득 과정에서는 심한 잡음이 영상의 화질을 떨어뜨릴 뿐만 아니라 추가적인 영상처리의 오류를 발생시킨다. 최근의 3D 잡음제거 방식은 시간적으로 연속된 영상간의 픽셀을 참조함으로 2D 잡음제거보다 더 나은 잡음 제거 결과를 얻을 수 있으나 움직임 부분에서는 오히려 모션 블러와 같은 열화가 나타나게 된다. 본 논문에서는 저조도 영상에서 적응적 가중평균필터를 이용하여 보다 정확한 움직임 검출을 구현하며, 3D 잡음제거 방식에 2D잡음 제거 방식의 결과를 적응적으로 사용하여 객관적 화질과 주관적 화질을 개선하였다.

• 정보저장시스템학회논문집
• /
• 제6권1호
• /
• pp.1-5
• /
• 2010

Holographic data storage system(HDSS) has many noise sources - crosstalk, scattering and inter pixel interference, etc. Generally the intensity of a light generated from the laser source has Gaussian distribution and this ununiformity of light also can make the data page to have a low SNR. A beam apodizer is used to make the laser as a flat-top beam but the intensity distribution is not strictly uniform. The intensity of light can be controlled using image mask. In this paper the intensity distribution of light used for HDSS is controlled by a digital image mask. The digital image mask is changed arbitrarily in real-time with suggested algorithm for the HDSS.

• IEIE Transactions on Smart Processing and Computing
• /
• 제1권1호
• /
• 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.