• Journal of Korea Multimedia Society
• /
• v.8 no.12
• /
• pp.1649-1657
• /
• 2005

In this paper, a fast implementation of JPEG is discussed and its application to multimedia service is presented for mobile wireless internet. A fast JPEG player is developed based on several fast algorithms for mobile handset. In the color transformation, RCT is adopted instead of ICT for JPEG source. For the most time-consuming DCT part, the binDCT can reduce the decoding time. In upsampling and RGB conversion, the transformation from YCbCr to RGB 16 bit is made at one time. In some parts, assembly language is applied for high-speed. Also, an implementation of multimedia in mobile handset is described using MJPEG (Motion JPEG) and QCELP(Qualcomm Code Excited Linear Prediction Coding). MJPEG and QCELP are used for video and sound, which are synchronized in handset. For the play of MJPEG, the decoder is implemented as a S/W upon the MSM 5500 baseband chip using the fast JPEG decoder. For the play of QCELP, the embedded QCELP player in handset is used. The implemented multimedia player has a fast speed preserving the image quality.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.07a
• /
• pp.39-41
• /
• 2011

본 논문에서는 깊이 영상의 해상도와 품질을 향상시키는 방법을 제안한다. 일반적으로 2D-plus-Depth 구조의 3D 콘텐츠에서는 깊이 영상의 품질이 매우 중요하다. 최근 들어 Time-of-Flight (TOF) 방식의 깊이 센서가 깊이 영상 획득에 많이 사용되고 있는데 TOF 깊이 센서가 제공하는 깊이 영상은 저해상도이기 때문에 고해상도 3D 콘텐츠를 제작하기 위해서는 깊이 영상의 해상도를 상향 변환하는 것이 필수적이다. 또한 고품질의 깊이 영상을 얻기 위해서는 물체 간의 경계를 정교하게 보존하는 것이 중요하다. 최근에는 깊이 영상의 해상도 상향 변환을 위해서 Joint Bilateral Upsampling(JBU) 방식이 많이 사용되고 있다. 본 논문은 깊이 영상의 해상도를 높임에 있어서 우선 보간법을 수행하여 영상의 상향 변환 시에 생긴 빈 홀들의 값을 채워준 후 Bilateral Filtering을 수행함으로써 성능을 높인다. 일반적으로 영상을 상향 변환을 할 때 다양한 방법들이 있는데 본 논문에서는 Nearest Neighborhood(NN), Gaussian과 경계 보존 보간법, 경계 보존 보간법과 Fast Curvature Based Interpolation(FCBI)를 결합한 보간법을 사용하였다. 실험 결과 제안 방법이 기존 방법보다 우수한 성능을 가짐을 보여준다. 또한 경계 보존 보간법과 FCBI를 결합한 보간법을 이용해서 상향 변환을 수행한 결과가 다른 보간법들에 의한 결과보다 우수하다는 점을 알 수 있다.

• Journal of KIISE:Software and Applications
• /
• v.30 no.7_8
• /
• pp.687-695
• /
• 2003

In this paper, we present a document image compression system based on segmentation, Quincunx downsampling, (5/3) wavelet lifting and subband-oriented SPIHT coding. We reduced the coding time by the adaptation of subband-oriented SPIHT coding and Quincunx downsampling. And to increase compression rate further, we applied arithmetic coding to the bitstream of SPIHT coding output. Finally, we present the reconstructed images for visual comparison and also present the compression rates and PSNR values under various scalar quantization methods.

• The Journal of Bigdata
• /
• v.2 no.2
• /
• pp.95-104
• /
• 2017

In 2014, Seoul Metropolitan Government launched a response service aimed at responding promptly to civil complaints. The complaints received are categorized based on their content and sent to the department in charge. If this part can be automated, the time and labor costs will be reduced. In this study, we collected 17,700 cases of complaints for 7 years from June 1, 2010 to May 31, 2017. We compared the XGBoost with RandomForest and confirmed the suitability of Korean text classification. As a result, the accuracy of XGBoost compared to RandomForest is generally high. The accuracy of RandomForest was unstable after upsampling and downsampling using the same sample, while XGBoost showed stable overall accuracy.

• The Journal of the Korea Contents Association
• /
• v.16 no.7
• /
• pp.431-438
• /
• 2016

In this paper, we present an upsampling technique for depth map image using color and depth weights. First, we construct a high-resolution image using the bilinear interpolation technique. Next, we detect a common edge region using RGB color space, HSV color space, and depth image. If an interpolated pixel belongs to the common edge region, we calculate weighting values of color and depth in $3{\times}3$ neighboring pixels and compute the cost value to determine the boundary pixel value. Finally, the pixel value having minimum cost is determined as the pixel value of the high-resolution depth image. Simulation results show that the proposed algorithm achieves good performance in terns of PSNR comparison and subjective visual quality.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.07a
• /
• pp.323-324
• /
• 2012

서로 다른 해상도의 좌, 우 영상으로 구성된 3DTV 시스템에서는 낮은 화질의 영상을 높은 화질의 해상도로 확대하여야 한다. 본 논문에서는 3종류의 보간기법을 실험 대상으로 하였으며 부호화기 및 부호화기에 각각 적용한 경우의 PSNR을 기준으로 성능을 분석하였다. 3종류의 보간 기법 중 SVC의 normative upsampling(JVT-R006) 보간기법의 성능이 우수함을 확인 하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.11a
• /
• pp.8-10
• /
• 2011

3D 콘텐츠를 획득하는 여러 가지 방법 중 2D-plus-Depth 구조는 다시점 영상을 얻을 수 있는 장점 때문에 최근 이에 관한 연구가 활발히 진행되고 있다. 이 구조를 통해서 고품질의 3D영상을 얻기 위해서는 무엇보다 고품질의 깊이 영상을 구현하는 것이 중요하다. 깊이 영상을 얻기 위해서 Time-of-Flight(ToF)방식의 깊이 센서가 활용되고 있는데 이 깊이 센서는 실시간으로 깊이 정보를 획득할 수 있지만 낮은 해상도와 노이즈가 발생한다는 단점이 있다. 따라서 깊이 영상의 특성을 보존하는 상환 변환을 하여야지만 고품질의 3D 콘텐츠를 제작할 수 있다. 주로 깊이 영상의 해상도를 높이기 위해서 Joint Bilateral Upsampling(JBU) 방식이 사용되고 있다. 하지만 이 방식은 4배 이상의 고 해상도 깊이 영상을 획득하는 데에는 적합하지 않다. 따라서 고해상도의 깊이 영상을 얻기 위해서 보간법을 수행하여 가이드 영상을 만든 후 Bilateral Filtering(BF)을 처리함으로써 영상의 품질을 향상시킨다. 본 논문에서는 2D-plus-Depth 구조에서 얻은 컬러 영상과 깊이 영상을 결합한 보간법을 통해서 깊이 영상의 특성을 살린 가이드 영상을 구현하는 방법을 제안한다. 실험 결과는 제안 방법이 기존 보간법보다 경계 영역 및 평활한 영역에서 깊이 영상의 특성을 잘 보존하는 것을 보여준다.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.2
• /
• pp.108-113
• /
• 2009

FIR upsampling circuit generates output data L times faster than input data. For efficient implementation of upsampler, FT filter stages Noble identity theory. In this paper, we propose a method that one of the divided filter stages is removed by making Phase delay to integer number. 192taps 2-times upsampler used in CDtoDAT is designed using proposed method. The designed circuit is synthesized using Hynix 0.18um process. By the simulation results, it is shown that the proposed method leads to up to 48%, 43% and 99.9% and 68% reduction in area, power consumption and maximum phase delay error and maximum ripple error compared with conventional method.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.59 no.11
• /
• pp.2108-2113
• /
• 2010

The Long Term Evolution (LTE) system is based on the Orthogonal Frequency Division Multiplexing (OFDM) and relies its channel estimation on the lattice-type pilot samples in the multipath fading channel environment. The estimation of the channel frequency response (CFR) makes use of the least squares estimate (LSE) for each pilot samples, followed by an interpolation both in time- and in frequency-domain to fill up the channel estimates for subcarriers corresponding to data samples. Any interpolation scheme could be adopted for this purpose. Depending on the requirements of the target system, we may choose a simple linear interpolation or a sophisticated one. For any choice of an interpolation scheme, these is a trade-off between estimation accuracy and numerical cost. For those wireless communication systems based on the OFDM and the preamble-type pilot structure, the DFT-based channel estimation and its variants have been successfully. Yet, it may not be suitable for the lattice-type pilot structure, since the pilot samples are not sufficient to provide an accurate estimate and it is known to be sensitive to the location as well as the length of the time-domain window. In this paper, we propose a simple interpolated based on the upsampling mechanism in the multirate signal processing. The proposed method provides an excellent alternative to the DFT-based methods in terms of numerical cost and accuracy. The performance of the proposed technique is verified on a multipath environment suggested on a 3GPP LTE specification.

• Journal of Multimedia Information System
• /
• v.8 no.4
• /
• pp.203-210
• /
• 2021

Deep leaning convolutional neural networks (CNN) have successfully been applied to image super-resolution (SR). Despite their great performances, SR techniques tend to focus on a certain upscale factor when training a particular model. Algorithms for single model multi-scale networks can easily be constructed if images are upscaled prior to input, but sub-pixel convolution upsampling works differently for each scale factor. Recent SR methods employ multi-scale and multi-path learning as a solution. However, this causes unshared parameters and unbalanced parameter distribution across various scale factors. We present a multi-scale single-path upsample module as a solution by exploiting the advantages of sub-pixel convolution and interpolation algorithms. The proposed model employs sub-pixel convolution for the highest scale factor among the learning upscale factors, and then utilize 1-dimension interpolation, compressing the learned features on the channel axis to match the desired output image size. Experiments are performed for the single-path upsample module, and compared to the multi-path upsample module. Based on the experimental results, the proposed algorithm reduces the upsample module's parameters by 24% and presents slightly to better performance compared to the previous algorithm.