• Journal of Broadcast Engineering
• /
• v.19 no.2
• /
• pp.257-267
• /
• 2014

Distributed compressive video sensing (DCVS) is a low cost sampling paradigm for video coding based on the compressive sensing and the distributed video coding. In this paper, we propose using a skip-mode coding in DCVS under the assumption that in case of high temporal correlation, temporal interpolation can guarantee sufficiently good quality of nonkey frame, therefore no need to transmit measurement data in such a nonkey frame. Furthermore, we extend it to use a hierarchical structure for better temporal interpolation. Simulation results show that the proposed skip-mode coding can save the average subrate of whole video sequence while the PSNR is reduced only slightly. In addition, by using the proposed scheme, the computational complexity is also highly decreased at decoder on average by 43.75% for video sequences that have strong temporal correlation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.5
• /
• pp.1187-1193
• /
• 2014

This paper presents an effective motion estimation to improve the performance of the motion compensated temporal filtering (MCTF) which is a core part of the wavelet-based scalable video coding. The proposed scheme makes the motion vector field uniform by the modified median operation and the search strategies using adjacent motion vectors, in order to enhance the pixel connectivity which is significantly relevant to the performance of the MCTF. Moreover, the motion estimation with variable block sizes that reflects the features of frames is introduced for further correlation improvement of the motion vector field. Experimental results illustrate that the proposed method reduces the decomposed energy on the temporal high frequency subband frame up to 30.33% in terms of variance compared to the case of the full search with fixed block sizes.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.3
• /
• pp.23-32
• /
• 2007

The H.264/AVC video coding standard uses rate distortion optimization (RDO) method to improve the compression performance in the intra prediction. The complexity and computational load are increased more than previous standard by using this method, even though this standard selects the best coding mode for the current macroblock. This paper proposes a fast intra mode decision algorithm for H.264/AVC encoder based on dominant edge direction (DED). To apply the idea, this algorithm uses the approximation of discrete cosine transform (DCT) coefficient. By detecting the DED, 3 modes instead of 9 modes are chosen for RDO calculation to decide the best mode in the $4{\times}4$ luma block. As for the $16{\times}16$ luma and $8{\times}8$ chroma block, instead of 4 modes, only 2 modes are searched. Experimental results show that the computation time of the proposed algorithm is decreased to about 72% of the full search method with negligible quality loss.

• Journal of Korea Society of Industrial Information Systems
• /
• v.22 no.5
• /
• pp.17-24
• /
• 2017

Recently Deep-learning, Technologies using Large or Deep Artificial Neural Networks, have Shown Remarkable Performance, and the Increasing Size of the Network Contributes to its Performance Improvement. However, the Increase in the Size of the Neural Network Leads to an Increase in the Calculation Amount, which Causes Problems Such as Circuit Complexity, Price, Heat Generation, and Real-time Restriction. In This Paper, We Propose and Test a Method to Reduce the Number of Network Connections by Effectively Pruning the Redundancy in the Connection and Showing the Difference between the Performance and the Desired Range of the Original Neural Network. In Particular, we Proposed a Simple Method to Improve the Performance by Re-learning and to Guarantee the Desired Performance by Allocating the Error Rate per Layer in Order to Consider the Difference of each Layer. Experiments have been Performed on a Typical Neural Network Structure such as FCN (full connection network) and CNN (convolution neural network) Structure and Confirmed that the Performance Similar to that of the Original Neural Network can be Obtained by Only about 1/10 Connection.

• Journal of Broadcast Engineering
• /
• v.23 no.1
• /
• pp.115-125
• /
• 2018

This paper presents a fast partition decision framework for High Efficiency Video Coding (HEVC) Screen Content Coding (SCC) based on machine learning. Currently, the HEVC performs quad-tree block partitioning process to achieve optimal coding efficiency. Since this process requires a high computational complexity of the encoding device, the fast encoding process has been studied as determining the block structure early. However, in the case of the screen content video coding, it is difficult to apply the conventional early partition decision method because it shows different partition characteristics from natural content. The proposed method solves the problem by classifying the screen content blocks after partition decision, and it shows an increase of 3.11% BD-BR and 42% time reduction compared to the SCC common test condition.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.38 no.5
• /
• pp.509-517
• /
• 2001

Full-search algorithm requires large amount of computation which causes time delay or very complex hardware architecture for real time implementation. In this paper, we propose a fast motion estimator based on bit-plane matching, which reduce the computational complexity and the hardware cost. In the proposed motion estimator, the conventional motion estimation algorithms are applied to the binary images directly extracted from the video sequence. Furthermore, in the proposed VLSI motion estimator, we employ a Pair of processing cores that calculate the motion vector continuously By controlling the data flow in a systolic fashion using the internal shift registers in the processing cores, we avoid using SRAM (local memory) so that we remove the time overhead for accessing the local memory and adopt lower-cost fabrication technology. We modeled and tested the proposed motion estimator in VHDL, and then synthesized the whole system which has been integrated in a 0.6-$\mu$m triple-metal CMOS chip of size 8.15 X 10.84$\textrm{mm}^2$.

• Journal of the Korea Computer Graphics Society
• /
• v.17 no.4
• /
• pp.31-40
• /
• 2011

Projector-based visual systems are widely used for VR and experience display applications. But the illumination irregularity on the screen surface due to the screen material and its light reflection properties sometimes deteriorates the user experience. This phenomenon is particularly troublesome when the participants of the head tracking VR system such as CAVE or the motion generation experience system continually move around the system. One of reason to illumination irregularity is projector-screen specular reflection component to participant's eye's position and it's analysis needs high computation complexity. Similar to calculate specular lighting term using GPU's programmable shader, Our research adjusts every pixel's brightness in runtime with given 3D screen space model to reduce illumination irregularity. For doing that, Angle-based brightness compensate function are considered for specific screen installation and modified it for GPU-friendly compute and access. Two aspects are implemented, One is function access transformation from angular form to product and the other is piecewise linear interpolate approximation.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.2
• /
• pp.274-279
• /
• 2009

H.264 suggests applying a 2-D 6-tap wiener filter to realize the interpolation for half-pixel positions, followed by a bilinear interpolation to get the data of quarter-pixels precision. This method is comparatively simpler; however, it only considers the affection of 4-connection neighborhood ignoring the influence that comes from the changing rate between respective neighborhoods. As a result, it has the characteristics of a Low-pass filter under the risk of losing high-frequency weights. The Cubic interpolation uses the gray-values within the larger regions of points to be sampled for interpolation. Nevertheless, the cubic interpolation is more complicated and computational. We give a deep analysis on the features while applying both bilinear and cubic interpolation in H.264 presenting a proper selection of interpolation algorithm with respect to specific distribution of gray-value in a certain grand block. The experiments point out that load far motion searching and interpolation are reduced when promoting the precision of interpolation simultaneously.

• 한국정보교육학회:학술대회논문집
• /
• 2010.08a
• /
• pp.185-192
• /
• 2010

The purpose of this paper is to suggest design principles for videoconferencing using an English Native Speaker. For this, Design-Based Research which is emerging newly was applied. During the research period of 3 months, several English classes were recorded. After finishing each class, Korean teacher and Native teacher wrote reflective diary and a researcher implemented interviews to students who had this class. Based on these data, this class was divided into 3 steps which is consisted of before class, under class, and after class. And then a research analysed patterns of this class, errors, and interactivity etc. The result of this paper, because a Korean teacher and students in the classroom meet a Native speaker in the distance through the videoconferencing system, teachers who design this class should consider many aspects due to the complexity of this.

• Journal of the Korea Computer Graphics Society
• /
• v.17 no.2
• /
• pp.17-26
• /
• 2011

This paper presents an interactive progressive image transmission method, which enables a remote user to interactively select and transmit preferred regions from an index image. Our enhanced quadtree decomposition using PSNR-based rules and new implicit quadtree coding provide better rate-distortion performance than previous quadtree coders as well as leading bit plane methods. An adaptive traversal of child nodes is introduced for better visual display of restored images. Depth-first traversal combined with breadth-first traversal of the quadtree to accomplish interactive transmission as presented, results in a method that provides competitive performance at a low level of computational complexity. Moreover, our decoding requires only simple arithmetic which is enabling our method to be used for real-time mobile applications.