Yoshimitsu Kuroki

Fast parallel processing using GPU in computing L1-PCA bases

In data-analysis problems with a large number of dimensions, the principal component analysis based on L2-norm (L2-PCA) is one of the most popular methods, but L2-PCA is sensitive to outliers. Unlike L2-PCA, PCA-L1 is robust to outliers because it utilizes the L1-norm, which is less sensitive to outliers; therefore, some studies have shown the superiority of PCA-L1 to L2-PCA [2][3]. However, PCA-L1 requires enormous computational cost to obtain the bases, because PCA-L1 employs an iterative algorithm, and initial bases are eigenvectors of autocorrelation matrix. The autocorrelation matrix in the PCA-L1 needs to be recalculated for the each basis besides. In previous works [3], the authors proposed a fast PCA-L1 algorithm providing identical bases in terms of theoretical approach, and decreased computational time roughly to a quarter. This paper attempts to accelerate the computation of the L1-PCA bases using GPU.

4K digital cinema home theater over high throughput wireless transmission system

In this paper, we propose a 4K digital cinema transmission over 1.2Gbps wireless LAN system. The proposed system employs the next generation wireless LAN system based on IEEE802.11ac specification. It reaches more than 33 meter propagation distance by using 80MHz of bandwidth on the 5GHz band. In this system, video data is compressed by JPEG 2000 with error resilience tools. These tools improve error performance against wireless channel, and enable very high throughput communication. Computer simulations are used to evaluate the bit error performances influences to the video quality. Finally, we present RTL design results of the proposed wireless LAN system.

HDTV over MIMO wireless transmission system

In this paper, we propose a HDTV transmission over IEEE802.3ac 4×5 MIMO WLAN System. The proposed system employs the next generation wireless LAN system based on IEEE802.3ac specification. It reaches more than 33 meter propagation distance by using 80MHz of bandwidth on the 5GHz band. In this system, video data is compressed by JPEG 2000 with error resilience tools. These tools improve error performance against wireless channel, and enable very high throughput communication. Computer simulations are used to evaluate the bit error performances influences to the video quality. Finally, we present RTL design results of the proposed wireless LAN system.

Real time 2D-DWT of JPEG 2000 for Digital Cinema using CUDA 4.0

The Specifications of the Digital Cinema system released by Digital Cinema Initiatives decide to encode so-called 2K or 4K images by JPEG 2000. The JPEG 2000 codec provides high compression rates and error tolerance, but this codec is burden for both encoding and decoding. The parallel computing architecture named CUDA receives a lot of attention recently. In this paper, we attempt to realize real-time implementation of the two dimensional discrete wavelet transform (2D-DWT) of JPEG 2000 by using the newest developmental environment CUDA 4.0.

4K Digital Cinema Transmission over 1.2Gbps Wireless LAN System

In this paper, we propose a 4K digital cinema wireless transmission over 1.2 Gbps wireless LAN system. The proposed system employs next generation wireless LAN system based on IEEE802.11TGac. It uses 80 MHz of bandwidth on 5 GHz band. In this system, video data is compressed by JPEG 2000 with added error resilience tools to improve error performance against wireless channel. Computer simulations are used to evaluate the bit error performances influences to the video quality.

Acceleration of Local Intensity Compensation using sparse representation with parallel processing

The novel motion compensation method, called LIC (Local Intensity Compensation), approximates a current block using a linear combination of reference blocks. To reduce the bitrate of high quality images, sparse coefficients are preferable; then, a method applying sparse representation to the LIC has been proposed. A solver of sparse representation called AIHT (Accelerated Iterative Hard Thresholding) uses many vector and matrix operations, and video coding standards encode frames in GOP (Group Of Pictures) structure. In some cases, video coding methods have to encode videos in real time, but the LIC using sparse representation takes a large amount of computational time for the prediction; thus, this paper demonstrates a method to accelerate the LIC using sparse representation as follows: (i) acceleration of the AIHT itself using GPU (Graphics Processing Unit); and (ii) employing parallel processing in multi-core CPU, namely OpenMP, for the independent structure of the GOP. By using OpenMP, multiple GPUs work under different cores. Experimental results show that the methods (i) and (ii) are faster than CPU implementation by about 4 times and about 7.4 times maximum, respectively.

Digital cinema wireless transmission and its windows application

In this paper, we propose a 4K digital cinema transmission over 1.2Gbps wireless LAN system. The proposed system employs the next generation wireless LAN system based on IEEE802.11ac task group. It reaches more than 33 meter propagation distance by using 80MHz of bandwidth on 5GHz band. In this system, a video data is compressed by JPEG 2000 with error resilience tools. These tools improve error performance against wireless channel, and enable very high throughput communication. Computer simulations are used to evaluate the bit error performances influences to the video quality. In addition, we implement the proposed system using wired LAN under Windows XP operating system.

Distributed compressed video sensing of multi-view images using ADMM

A light field camera records information of light (light field) in three dimensional space. It can also generate multi-view images by processing the light field. Distributed compressed video sensing is a technique composed by a distributed video coding and a compressed sensing. This study examines a distributed compressed video sensing that focuses on correlation between the multi-view images and solves sparse problem on the distributed compressed video sensing with ADMM (Alternating Direction Method of Mutipliers) which is one of convex optimization algorithms. In addition, we design dictionaries to represent an original signal sparser and confirm high-quality image reconstruction.

Image recovery by D-SET framework with improved BNN

Decomposed SET, namely Smooth, Edge, and Texture, (D-SET) recovery, is an image decomposition and recovery method which assumes images as the sum of smooth, edge, and texture components. All the components are obtained by solving an optimization problem consists of regularizations based on priori information of original images. In this paper, we apply BNN (block nuclear norm) to the regularization of texture components of D-SET. BNN is originally applied for cartoon-texture image decomposition, and its decomposition method regards a image as the sum of ideal cartoon and sub-texture components. BNN is defined as the sum of singular values of all possibility overlapped sub-blocks of sheared image, and BNN of texture components become small. On the other hand, D-SET uses Shift-Invariant Redundant Discrete Cosine Transform (RDCT) for the texture component. This work tries to obtain better recovery images by applying BNN instead of RDCT, and we also improve BNN itself by adjusting the shearing operations of BNN in a block-wise manner. The proposed method demonstrates more effective image recovery than the conventional D-SET.

Local Intensity Compensation using sparse representation with additional sparse dictionary

The novel motion compensation called LIC (Local Intensity Compensation) predicts a current block using a linear combination of reference blocks. The weight coefficients of the LIC should be sparse in order to predict and to encode frames effectively. Sparse representation aims at obtaining the least possible number of non-zero coefficients and is a powerful solution for the LIC. Our previous work has proposed the LIC using sparse representation and has achieved high prediction accuracy. Video coding standards encode motion vectors and prediction errors of the macro blocks independently. The prediction errors are converted via DCT (Discrete Cosine Transform), and the quantized DCT coefficients are encoded. This means that the video coding standards predicts the current block using two steps, namely motion compensation and the DCT. Sparse representation is applied to optional basis vectors as reference blocks; thus, this study proposes a framework to conduct motion compensation and the DCT simultaneously by adding the DCT basis vectors into the sparse dictionary. Additionally, this study employs macro blocks generated using intra prediction because the intra prediction is also applied to the current block by neighbouring macro blocks. Experimental results demonstrate that the proposed method integrating the two steps shows high prediction accuracy.