• Journal of Korea Multimedia Society
• /
• v.13 no.6
• /
• pp.817-824
• /
• 2010

Linear Discriminant Analysis (LDA) has been successfully applied for dimension reduction in face recognition. However, LDA requires the transformation of a face image to a one-dimensional vector and this process can cause the correlation information among neighboring pixels to be disregarded. On the other hand, 2D-LDA uses 2D images directly without a transformation process and it has been shown to be superior to the traditional LDA. Nevertheless, there are some problems in 2D-LDA. First, it is difficult to determine the optimal number of feature vectors in a reduced dimensional space. Second, the size of rectangular windows used in 2D-LDA makes strong impacts on classification accuracies but there is no reliable way to determine an optimal window size. In this paper, we propose a new algorithm to overcome those problems in 2D-LDA. We adopt an ensemble approach which combines several classifiers obtained by utilizing various window sizes. And a practical method to determine the number of feature vectors is also presented. Experimental results demonstrate that the proposed method can overcome the difficulties with choosing an optimal window size and the number of feature vectors.

• Proceedings of the IEEK Conference
• /
• 2002.07a
• /
• pp.603-606
• /
• 2002

EZW, a wavelet transformation based compression algorithm, utilizes the correlations of the wavelet coefficients between subbands. Due to the reason that, it is based on Zero Tree's parent-descendant dependency of the subbands, subband coding is possible, but it is not efficient for applications, which require multi-resolution restorations. In this paper we compare and analyze the performance between spatial band coding, which apply the correlation between adjusting pixels, and EZW So as to prove that spatial band coding is superior to EZW in the sense that it exhibit higher PSNR values and that it is apt to molti-resolution restoration.

• Proceedings of KOSOMES biannual meeting
• /
• 2009.06a
• /
• pp.53-54
• /
• 2009

The influence of atmospheric aerosol on satellite ocean color data were evaluated using SeaWiFS monthly standard mapped image products. The atmospheric optical thickness (AOT) was increased in spring and summer, and it showed the strong positive correlation with remote sensing reflectance, normalized waterleaving radiance /solar irradiance, at 555 nm (Rrs555) which is a component of the satellite chlorophyll estimation. Such the high AOT and high Rrs555 pixels showed overestimation of satellite chlorophyll in spring, especially in the area which showed large phytoplankton absorption which 1s expressed by low remote sensing reflectance at 443, 490 and 510 nm (Rrs 443, Rrs490 and Rrs510).

• 한국가시화정보학회:학술대회논문집
• /
• 2002.11a
• /
• pp.51-52
• /
• 2002

Electro-osmotic flow in a PDMS microchannel of $66{\mu}m\;\times\;200{\mu}m\;\times\;3cm$ has been investigated using a micro PIV system. The field of view was $1056{\mu}m\;\times\;200{\mu}m$ and instantaneous velocity fields were obtained using two-frame cross-correlation method with $64\;\times\;64\;pixels^2$ interrogation window. In this study, we focused on the effect of applied electric field on the variation of internal flow with varying the electric field and seeding particles. The electro-osmotic flow shows a flat velocity profile and the mean velocity is proportional to the applied electric field.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.5
• /
• pp.915-927
• /
• 1997

In this paper, we propose an error concealment technique using directional interpolation in block-based image compression. In the proposed method the edge direction is determined by finding the maximum correlation coefficients of boundary pixels of blocks neighboring the errored block in spatial domain. Then the errored block is interpolated linearly or bilinearly along the determined edge direction. The proposed method can conceal the block error, the macro block error, and the slice error adaptively. Also, the parameters for the directional interpolation are represented by closed forms. When applied to compressed images, the proposed method shows superior subjective and objective quality to conventional error concealment methods.

• Journal of KIISE:Software and Applications
• /
• v.37 no.6
• /
• pp.417-429
• /
• 2010

Reversible watermarking inserts watermark into digital media in such a way that visual transparency is preserved and then enables to restore the original media from the marked one without any loss of media quality. This watermarking can be applied to quality-sensitive imaging such as medical imaging, military imaging, remote-sensing imaging, and precious artwork, where the original media should be preserved during image processing and analysis. In this paper, a reversible image watermarking technique that embeds message bits by modifying the differential histogram of adjacent pixels is presented. In order to satisfy both high embedding capacity and visual quality, the proposed technique exploits the fact that adjacent pixels in the image have highly spatial correlation. Also, we prevent overflow/underflow problem and salt-and-pepper artifacts by employing a predicted error compensation scheme. Through experiments using various test images, we prove that the presented technique provides perfect reversibility and high embedding capacity, while maintaining the induced-distortion low.

• Proceedings of the IEEK Conference
• /
• 2003.11a
• /
• pp.381-384
• /
• 2003

Among digital halftoning methods, error diffusion is a procedure for generating high quality bilevel images from continuous-tone images but blurs the edge information in the bilevel images. To solve the problem, we propose the edge enhanced error diffusion using the edge information of the original images. The edge enchanted weights is computed by adding local characteristic weights and input pixels multiplied a constant. Also, we combined the edge enhanced method with the adaptive error diffusion using human spatial and frequency perception characteristic. The performance of the proposed method is compared with conventional method by measuring the edge correlation. The halftoned images applied the proposed method get more fine quality due to the enchanced edge and better quality in halftoned image. And the detailed edge is preserved in the halftoned images by the proposed method.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.656-659
• /
• 2020

초해상화 딥러닝 기법은 학습 시 수렴하기까지 최소 수백 번의 에폭을 필요로 하며 오랜 시간이 걸린다. 최근, 영상 인식용 딥러닝 모델에서는 학습 수렴 속도를 향상시키기 위해 픽셀, 채널간 불필요한 상호연관 정보를 제거하는 Deconvolution 기술이 제안되었다. 본 논문에서는 최초로 Deconvolution 기술을 초해상화 딥러닝 방법에 적용하여 학습 수렴 속도 증가를 시도했다. 영상 인식 딥러닝 기법과 다르게 초해상화 딥러닝 기법은 이미지 특성 추출 부분과 이미지 복원 부분의 정보를 보존하는 것이 중요하기 때문에, EDSR을 Baseline 모델로 사용하여 양쪽 끝의 레이어는 기존의 Convolution 연산을 그대로 유지하고, 중간 레이어의 ResBlock 내의 Convolution 연산만 Deconvolution 연산으로 바꿔서 구성하였다. 초해상화 벤치마크 데이터셋을 사용한 실험 결과, 수렴속도가 빨라지지 않는 결과를 도출했다. 본 논문에서는 Deconvolution 기술이 Baseline 모델의 성능을 개선하지 못하는 이유를 초해상화 분야에서 기본적으로 적용되는 Residual Learning 기법 때문으로 분석했다.

• The Journal of Korea Robotics Society
• /
• v.3 no.4
• /
• pp.300-307
• /
• 2008

This paper is concerned with face recognition for human-robot interaction (HRI) in robot environments. For this purpose, we use Tensor Subspace Analysis (TSA) to recognize the user's face through robot camera when robot performs various services in home environments. Thus, the spatial correlation between the pixels in an image can be naturally characterized by TSA. Here we utilizes face database collected in u-robot test bed environments in ETRI. The presented method can be used as a core technique in conjunction with HRI that can naturally interact between human and robots in home robot applications. The experimental results on face database revealed that the presented method showed a good performance in comparison with the well-known methods such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) in distant-varying environments.

• Journal of Korea Multimedia Society
• /
• v.8 no.3
• /
• pp.313-321
• /
• 2005

Digital halftoning is a process to produce a binary image so that the original image and its binary counterpart appear similar when observed from a distance. Among digital halftoning methods, error diffusion is a procedure for generating high quality bilevel images from continuous-tone images but blurs the edge information in the bilevel images. To solve this problem, we propose the improved error diffusion using local spatial information of the original images. Based on the fact that the human vision perceives not a pixel but local mean of input image, we compute edge enhancement information(EEI) by appling the ratio of a pixel and its adjacent pixels to local mean. The weights applied to local means is computed using the ratio of local activity measure(LAM) to the difference between input pixels of 3$\times$3 blocks and theirs mean. LAM is the measure of luminance changes in local regions and is obtained by adding the square of the difference between input pixels of 3$\times$3 blocks and theirs mean. We add the value to a input pixel of quantizer to enhance edge. The performance of the proposed method is compared with conventional methods by measuring the edge correlation. The halftone images by using the proposed method show better quality due to the enhanced edge. And the detailed edge is preserved in the halftone images by using the proposed method. Also the proposed method improves the quality of halftone images because unpleasant patterns for human visual system are reduced.