• 제어로봇시스템학회논문지
• /
• 제8권2호
• /
• pp.151-157
• /
• 2002

A 3-dimenstional automatic solder paste inspection system is described that can be used to find defects occurring in solder paste printing process. This system extracts height and volume information very fast as well as area of solder paste printed, using multiple slit ray projection and Galvano-mirror scanning. Methods are presented on calibration of camera and slit projector, real-time image processing of multiple slit images, determination of reference height, and extraction of paste height information are proposed. Performance of the system was successfully demonstrated through field tests.

• 전자공학회논문지SC
• /
• 제38권3호
• /
• pp.31-39
• /
• 2001

본 논문에서는 CMOS 이미지 센서의 동작 범위(Dynamic Range; DR)를 높이기 위해서， multiple sampling 방법과 조건적 reset 기능을 갖는 새로운 픽셀 회로를 제안한다. 제안된 구조는 한 번의 integration 시간 내에서 픽셀의 출력이 일정한 간격으로 여러 번 sampling되고 sampling된 각 신호는 기준 전압과 비교되며 이 결과에 따라 해당 픽셀을 rest 할지의 여부가 결정된다. 제안된 방법을 사용하면 이미지 센서의 최대 DR은 축적 기간 동안의 총 sampling 회수인 N 배로 증가될 수 있다. 테스트 칩은 0.65-${\mu}m$ CMOS 공정(2-P, 2-M)으로 제작되었으며 이에 대한 측정결과로 본 논문의 알고리듬이 DR의 증가에 효과적임을 확인하였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 ITC-CSCC -2
• /
• pp.1003-1006
• /
• 2002

In the communications systems with diversity, we are commonly faced on needing of new source coding technique, error resilient coding. The error resilient coding addresses the coding algorithm that has the robustness to unreliability of communications channel. In recent years, many error resilient coding techniques were proposed such as data partitioning, resynchronization, error detection, concealment, reference picture selection and multiple description coding (MDC). Especially, the MDC using correlating transform explicitly adds correlation between two descriptions to enable the estimation of one set from the other. However, in the conventional correlating transform method, there is a critical problem that decoder must know statistics of original image. In this paper, we propose an enhanced method, the MDC using whitening transform that is not necessary additional statistical information to decode image because the DCT coefficients to apply whitening transform to an image have uni-variance statistics. Our experimental results show that the proposed method achieves a good trade-off between the coding efficiency and the reconstruction quality. In the proposed method, the PSNR of images reconstructed from two descriptions is about 0.7dB higher than conventional method at the 1.0 BPP and from only one description is about 1,8dB higher at the same rate.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권2호
• /
• pp.771-789
• /
• 2019

Text detection has been a popular research topic in the field of computer vision. It is difficult for prevalent text detection algorithms to avoid the dependence on datasets. To overcome this problem, we proposed a novel unsupervised text detection algorithm inspired by bootstrap learning. Firstly, the text candidate in a novel form of superpixel is proposed to improve the text recall rate by image segmentation. Secondly, we propose a unique text sample selection model (TSSM) to extract text samples from the current image and eliminate database dependency. Specifically, to improve the precision of samples, we combine maximally stable extremal regions (MSERs) and the saliency map to generate sample reference maps with a double threshold scheme. Finally, a multiple kernel boosting method is developed to generate a strong text classifier by combining multiple single kernel SVMs based on the samples selected from TSSM. Experimental results on standard datasets demonstrate that our text detection method is robust to complex backgrounds and multilingual text and shows stable performance on different standard datasets.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권4호
• /
• pp.1166-1186
• /
• 2022

Artistic font design has become an integral part of visual media. However, without prior knowledge of the font domain, it is difficult to create distinct font styles. When the number of characters is limited, this task becomes easier (e.g., only Latin characters). However, designing CJK (Chinese, Japanese, and Korean) characters presents a challenge due to the large number of character sets and complexity of the glyph components in these languages. Numerous studies have been conducted on automating the font design process using generative adversarial networks (GANs). Existing methods rely heavily on reference fonts and perform font style conversions between different fonts. Additionally, rather than capturing style information for a target font via multiple style images, most methods do so via a single font image. In this paper, we propose a network architecture for generating multilingual font sets that makes use of geometric structures as content. Additionally, to acquire sufficient style information, we employ multiple style images belonging to a single font style simultaneously to extract global font style-specific information. By utilizing the geometric structural information of content and a few stylized images, our model can generate an entire font set while maintaining the style. Extensive experiments were conducted to demonstrate the proposed model's superiority over several baseline methods. Additionally, we conducted ablation studies to validate our proposed network architecture.

• 한국통신학회논문지
• /
• 제34권11A호
• /
• pp.870-881
• /
• 2009

의료 IT 기술 발전과 함께 의료 디지털 도서관, 3D PACS, 3D 의료진단기기 등의 헬스케어 정보 관리 기술이 급격히 발전되면서 이에 대한 보안 이슈가 제기되고 있다. 본 논문에서는 헬스케어 정보 관리 시스템에서 3D 의료영상 데이터의 저작권 보호, 인증, 인덱싱 및 진단 정보 은닉 등을 위한 다중 워터마킹 기법을 제안한다. 제안한 기법에서는 POCS 워터마킹 기반으로 의료진의 디지털 서명 및 정보 검색 인덱싱을 위한 강인한 워터마크를 꼭지점 정규곡률 분포에 삽입하고, 진단 정보와 인증 기준 메시지를 위한 연약한 워터마크를 꼭지점 거리 차이에 삽입한다. 이 때 강인성, 연약성 및 비가시성에 대한 각각의 볼록 집합들을 설계한 다음, 3D 의료영상 데이터들을 이들 집합으로 반복 투영함으로써 다중 워터마크를 삽입한다. 실험 결과부터 제안한 기법이 다양한 3D 기하학 및 메쉬 변형에 대한 강인성과 연약성을 모두 만족함을 확인하였다.

• Current Optics and Photonics
• /
• 제5권5호
• /
• pp.524-531
• /
• 2021

In an integral-imaging microscopy (IIM) system, a microlens array (MLA) is the primary optical element; however, surface errors impede the resolution of a raw image's details. Calibration is a major concern with regard to incorrect projection of the light rays. A ray-tracing-based calibration method for an IIM camera is proposed, to address four errors: MLA decentering, rotational, translational, and subimage-scaling errors. All of these parameters are evaluated using the reference image obtained from the ray-traced white image. The areas and center points of the microlens are estimated using an "8-connected" and a "center-of-gravity" method respectively. The proposed approach significantly improves the rectified-image quality and nonlinear image brightness for an IIM system. Numerical and optical experiments on multiple real objects demonstrate the robustness and effectiveness of our proposed method, which achieves on average a 35% improvement in brightness for an IIM raw image.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권9호
• /
• pp.2483-2504
• /
• 2023

Most existing low-light enhancement algorithms either use a large number of training parameters or lack generalization to real-world scenarios. This paper presents a novel lightweight and robust pixel-wise polynomial approximation-based deep network for low-light image enhancement. For mapping the low-light image to the enhanced image, pixel-wise higher-order polynomials are employed. A deep convolution network is used to estimate the coefficients of these higher-order polynomials. The proposed network uses multiple branches to estimate pixel values based on different receptive fields. With a smaller receptive field, the first branch enhanced local features, the second and third branches focused on medium-level features, and the last branch enhanced global features. The low-light image is downsampled by the factor of 2b-1 (b is the branch number) and fed as input to each branch. After combining the outputs of each branch, the final enhanced image is obtained. A comprehensive evaluation of our proposed network on six publicly available no-reference test datasets shows that it outperforms state-of-the-art methods on both quantitative and qualitative measures.

• 대한인간공학회지
• /
• 제12권1호
• /
• pp.75-86
• /
• 1993

Hypermedia systems allow the retrieval and representation of multimedia in- formation such as text, graphics, image and voice/sound using navigation and browsing mechanisms. In this study we developed a hypermedia system which provides hierarchical group, local map and cluster view for effective navigation in hyperspace. The system also supports hot link, reference link, move-to link and multiple link to browse the multimedia information space effectively. This system has been designed to integrate the navigation, browsing and searching function of the hypermedia system in hyman factor perspective and provides the user-friendly user interface mechanism.

• 대한방사선기술학회지:방사선기술과학
• /
• 제44권4호
• /
• pp.327-333
• /
• 2021

Magnetic resonance imaging(MRI) uses strong magnetic field to image the cross-section of human body and has excellent image quality with no risk of radiation exposure. Because of above-mentioned advantages, MRI has been widely used in clinical fields. However, the noise generated in MRI degrades the quality of medical images and has a negative effect on quick and accurate diagnosis. In particular, examining a object with a detailed structure such as brain, image quality degradation becomes a problem for diagnosis. Therefore, in this study, we acquired T2 weighted 3D data of multiple sclerosis disease using BrainWeb simulation program, and used quantitative evaluation factors to find appropriate slice thickness among 1, 3, 5, and 7 mm. Coefficient of variation and contrast to noise ratio were calculated to evaluate the noise level, and root mean square error and peak signal to noise ratio were used to evaluate the similarity with the reference image. As a result, the noise level decreased as the slice thickness increased, while the similarity decreased after 5 mm. In conclusion, as the slice thickness increases, the noise is reduced and the image quality is improved. However, since the edge signal is lost due to overlapped signal, it is considered that selecting appropriate slice thickness is necessary.