• The KIPS Transactions:PartD
• /
• v.10D no.3
• /
• pp.459-470
• /
• 2003

It is required to research on high-dimensional index structures for efficiently retrieving high-dimensional data because an attribute vector in data warehousing and a feature vector in multimedia database have a characteristic of high-dimensional data. For this, many high-dimensional index structures have been proposed, but they have so called ‘dimensional curse’ problem that retrieval performance is extremely decreased as the dimensionality is increased. To solve the problem, the cell-based filtering (CBF) scheme has been proposed. But the CBF scheme show a linear decreasing on performance as the dimensionality. To cope with the problem, it is necessary to make use of parallel processing techniques. In this paper, we propose a parallel CBF scheme which uses a horizontally-partitioned technique as declustering. In order to maximize the retrieval performance of the proposed parallel CBF scheme, we construct our parallel CBF scheme under a SN (Shared Nothing) cluster architecture. In addition, we present a data insertion algorithm, a rage query processing one, and a k-NN query processing one which are suitable for the SN cluster architecture. Finally, we show that our parallel CBF scheme achieves better retrieval performance in proportion to the number of servers in the SN cluster architecture, compared with the conventional CBF scheme.

• Journal of information and communication convergence engineering
• /
• v.11 no.4
• /
• pp.298-306
• /
• 2013

The scale invariant feature transform (SIFT) is an effective algorithm used in object recognition, panorama stitching, and image matching. However, due to its complexity, real-time processing is difficult to achieve with current software approaches. The increasing availability of parallel computers makes parallelizing these tasks an attractive approach. This paper proposes a novel parallel approach for SIFT algorithm implementation using a block filtering technique in a Gaussian convolution process on the SIMD Pixel Processor. This implementation fully exposes the available parallelism of the SIFT algorithm process and exploits the processing and input/output capabilities of the processor, which results in a system that can perform real-time image and video compression. We apply this implementation to images and measure the effectiveness of such an approach. Experimental simulation results indicate that the proposed method is capable of real-time applications, and the result of our parallel approach is outstanding in terms of the processing performance.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.6 no.2
• /
• pp.100-107
• /
• 2011

As the use of mobile multimedia devices is increasing in the recent year, the needs for high-performance multimedia processors are increasing. In this regard, we propose a SIMD (Single Instruction Multiple Data) based parallel processor that supports high-performance multimedia applications with low energy consumption. The proposed parallel processor consists of 16 processing elements(PEs) and operates on a 3-stage pipelining. Experimental results for the JPEG encoding algorithm indicate that the proposed parallel processor outperforms conventional parallel processors in terms of performance and energy efficiency. In addition, the proposed parallel processor architecture was developed and verified with verilog HDL and a FPGA prototype system.

• Journal of Korea Multimedia Society
• /
• v.12 no.3
• /
• pp.329-334
• /
• 2009

When an optical disc is being played. the pick-up converts light to analog signal at first. The analog signal is equalized for removing the inter-symbol interference and then the equalized analog signal is converted into the digital signal for extracting the synchronized data and clock signals. There are a lot of algorithms that minimize the BER in extracting the synchronized data and clock when high. density optical disc like BD is being played in low speed. But if the high-density optical disc is played in high speed, it is difficult to adopt the same extraction algorithm to data PLL and PRML architecture used in low speed application. It is because the signal with more than 800MHz should be processed in those architectures. Generally, in the 0.13-${\mu}m$ CMOS technology, it is necessary to have the high speed analog cores and lots of efforts to layout. In this paper, the parallel data PLL and PRML architecture, which enable to process in BD 8x speed of the maximum speed of the high-density optical disc as the extracting data and clock circuit, is proposed. Test results show that the proposed architecture is well operated without processing error at BD 8x speed.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.10
• /
• pp.11-21
• /
• 2011

This paper introduces an SIMD(Single Instruction Multiple Data) based parallel processor that efficiently processes massive data inherent in multimedia. In addition, this paper implements MMX(MultiMedia eXtension)-type instructions on the data parallel processor and evaluates and analyzes the performance of the MMX-type instructions. The reference data parallel processor consists of 16 processors each of which has a 32-bit datapath. Experimental results for a JPEG compression application with a 1280x1024 pixel image indicate that MMX-type instructions achieves a 50% performance improvement over the baseline instructions on the same data parallel architecture. In addition, MMX-type instructions achieves 100% and 51% improvements over the baseline instructions in energy efficiency and area efficiency, respectively. These results demonstrate that multimedia specific instructions including MMX-type have potentials for widely used many-core GPU(Graphics Processing Unit) and any types of parallel processors.

• Journal of Information Technology and Architecture
• /
• v.10 no.3
• /
• pp.387-397
• /
• 2013

This paper proposes an approach to big data analytics for phon call data. The analytical models for phon call data is composed of the PVPF (Parallel Variable-length Phrase Finding) algorithm for identifying verbal phrases of natural language and the word count algorithm for measuring the usage frequency of keywords. In the proposed model, we identify words using the PVPF algorithm, and measure the usage frequency of the identified words using word count algorithm in MapReduce. The results can be interpreted from various viewpoints. We design and implement the model based HDFS (Hadoop Distributed File System), verify the proposed approach through a case study of phon call data. So we extract useful results through analysis of keyword correlation and usage frequency.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.2
• /
• pp.149-158
• /
• 1996

In this paper we describe designing and implementing a digital neural chip and a parallel neural machine for simulating large scale neural netsorks. The chip is a single-chip multiprocessor which has four digiral neural processors (DNP-II) of the same architecture. Each DNP-II has program memory and data memory, and the chip operates in MIMD (multi-instruction, multi-data) parallel processor. The DNP-II has the instruction set tailored to neural computation. Which can be sed to effectively simulate various neural network models including on-chip learning. The DNP-II facilitates four-way data-driven communication supporting the extensibility of parallel systems. The parallel neural machine consists of a host computer, processor boards, a buffer board and an interface board. Each processor board consists of 8*8 array of DNP-II(equivalently 2*2 neural chips). Each processor board acn be built including linear array, 2-D mesh and 2-D torus. This flexibility supports efficiency of mapping from neural network models into parallel strucgure. The neural system accomplishes the performance of maximum 40 GCPS(giga connection per second) with 16 processor boards.

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.33A no.12
• /
• pp.110-119
• /
• 1996

A novel high speed VLSI architecture and its VLSI realization methodologies for a motion estimation processor based on full search block matching algorithm are presentd. The presented architecture is designed in order to be suitable for highly parallel and pipelined processing with identical PE's and adjustable in performance and hardware amount according to various application areas. Also, the throughput is maximized by enhancing PE utilization up to 100% and the chip pin count is reduced by reusing image data with embedded image memories. Also, the uniform and identical data processing structure of PE's eases VLSI implementation and the clock rate of external I/O data can be made slower compared to internal clock rate to resolve I/O bottleneck problem. The logic and spice simulation results of the proposed architecture are presented. The performances of the proposed architecture are evaluated and compared with other architectures. Finally, the chip layout is shown.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.11
• /
• pp.5389-5396
• /
• 2012

In this paper, an optimized residual data decoder architecture is proposed to improve the performance in H.264/AVC. The proposed architecture is an integrated architecture that combined parallel inverse transform architecture and parallel inverse quantization architecture with common operation units applied new inverse quantization equations. The equations without division operation can reduce execution time and quantity of operation for inverse quantization process. The common operation unit uses multiplier and left shifter for the equations. The inverse quantization architecture with four common operation units can reduce execution cycle of inverse quantization to one cycle. The inverse transform architecture consists of eight inverse transform operation units. Therefore, the architecture can reduce the execution cycle of inverse transform to one cycle. Because inverse quantization operation and inverse transform operation are concurrency, the execution cycle of inverse transform and inverse quantization operation for one $4{\times}4$ block is one cycle. The proposed architecture is synthesized using Magnachip 0.18um CMOS technology. The gate count and the critical path delay of the architecture are 21.9k and 5.5ns, respectively. The throughput of the architecture can achieve 2.89Gpixels/sec at the maximum clock frequency of 181MHz. As the result of measuring the performance of the proposed architecture using the extracted data from JM 9.4, the execution cycle of the proposed architecture is about 88.5% less than that of the existing designs.

• Korean Journal of Remote Sensing
• /
• v.28 no.6
• /
• pp.683-691
• /
• 2012

High resolution satellite images are now widely used for a variety of mapping applications including photogrammetry, GIS data acquisition and visualization. As the spectral and spatial data size of satellite images increases, a greater processing power is needed to process the images. The solution of these problems is parallel systems. Parallel processing techniques have been developed for improving the performance of image processing along with the development of the computational power. However, conventional CPU-based parallel computing is often not good enough for the demand for computational speed to process the images. The GPU is a good candidate to achieve this goal. Recently GPUs are used in the field of highly complex processing including many loop operations such as mathematical transforms, ray tracing. In this study we proposed a technique for parallel processing of high resolution satellite images using GPU. We implemented a spectral radiometric processing algorithm on Landsat-7 ETM+ imagery using CUDA, a parallel computing architecture developed by NVIDIA for GPU. Also performance of the algorithm on GPU and CPU is compared.