• ETRI Journal
• /
• v.25 no.5
• /
• pp.297-304
• /
• 2003

A multimedia SoC often requires a large internal buffer, because it must store the whole search window to reduce the huge I/O bandwidth of motion estimation. However, the silicon area of the internal buffer increases tremendously as the search range becomes larger. This paper proposes a new method that greatly reduces the internal buffer size of a multimedia SoC while the computational cost, I/O bandwidth, and image quality do not change. In the proposed method, only the overlapped parts of search windows for consecutive macroblocks are stored in the internal buffer. The proposed method reduces the internal buffer. The proposed method reduces the internal buffer size to 1/5.0 and 1/8.8 when the search range is ${\pm}64{\times}{\pm}$64 and ${\pm}128{\times}{\pm}$128, respectively.

• Journal of The Institute of Information and Telecommunication Facilities Engineering
• /
• v.10 no.2
• /
• pp.53-61
• /
• 2011

This research presents an adaptive fast motion estimation (ME) computation on the stage of uneven multi-hexagon grid search (UMHGS) algorithm included in an unsymmetrical-cross multi-hexagon-grid search (UMHexagonS) in H.264 standard. The proposed adaptive method is based on statistical analysis and previously obtained motion vectors to reduce the computational complexity of ME. For this purpose, the algorithm is decomposed into three processes: skipping, terminating, and reducing search areas. Skipping and terminating are determined by the statistical analysis of the collected minimum SAD (sum of absolute difference) and the search area is constrained by the slope of previously obtained motion vectors. Simulation results show that 13%-23% of ME time can be reduced compared with UMHexagonS, while still maintaining a reasonable PSNR (peak signal-to-noise ratio) and average bitrates.

• Journal of Advanced Navigation Technology
• /
• v.16 no.3
• /
• pp.545-552
• /
• 2012

In this paper, we propose Modified Multi CAMShift Algorithm based on Kalman filter and Weighted Search Windows(KWMCAMShift) that extracts skin color area and tracks several human body parts for real-time human tracking system. We propose modified CAMShift algorithm that generates background model, extracts skin area of hands and head, and tracks the body parts. Kalman filter stabilizes tracking search window of skin area due to changing skin area in consecutive frames. Each occlusion areas is avoided by using weighted window of non-search areas and main-search area. And shadows are eliminated from background model and intensity of shadow. The proposed KWMCAMShift algorithm can estimate human pose in real-time and achieves 96.82% accuracy even in the case of occlusions.

• Journal of Broadcast Engineering
• /
• v.10 no.3
• /
• pp.357-371
• /
• 2005

In this Paper, we propose a new motion search algorithm called CDTS (Cross-Diamond-Triangle Search algorithm) that uses optimal search pattern according to the position of a search area to improve the performance of CDS(Cross-Diamond Search algorithm) as well as CDHSs(Cross-Diamond-Hexagonal Searches algorithms). We analyze motion distributions in various test video sequences to apply optimal search pattern according to a position of search area. Based on the result of this analysis, we propose a new triangle-shaped search pattern whose structure is asymmetric while previous search patterns are generally symmetric in conventional algorithms. In CDTS, we apply cross- and diamond-shaped search patterns to central search areas, and triangle- and diamond-shaped patterns to the other areas. Applying CDTS to test video sequences, the proposed scheme can reduce search points more than CDS and CDHSs by 16.22$\%$ and 3.09$\%$, respectively, without any visual quality degradation.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.389-391
• /
• 1999

영상 압축 기법에는 여러 가지 알고리즘을 적용되고 있다. 이런 알고리즘들에는 주파수 영역 중복을 제거하기 위한 DCT, 시간 중복성 제거를 위한 움직임 측정, 압축기법에 의해서 만들어진 정보를 부호화하는 VLC들이 있다. 이런 부호화 알고리즘들은 부호화기를 구현하는데 많은 시간을 요구하며 특히 움직임 추정은 부호화기의 절반에 가까운 시간을 소비한다. 움직임 측정 기술의 복잡도는 search algorithm, cost function, search range parameter의 요인으로 나타낼 수 있다. 본 논문에서는 기존의 Full Search 알고리즘 대신에 three-step 알고리즘을 사용하여 움직임 측정 시간을 줄였다. Full Search 알고리즘은 search area에서 모든 지역에 대해 cost function을 사용하여 이전 블록과 얼마나 유사한지를 조사한다. 따라서 이전 블록과 가장 유사한 부분을 찾는 좋은 방법이지만 그만큼 시간이 많이 사용한다. Three-step 알고리즘은 search area의 일정 지역에 대해 cost function를 사용하여 이전 블록과의 유사성을 찾는 fast 알고리즘이다. Three-step 알고리즘을 사용한 경우 기존의 full search 알고리즘을 사용할 때 보다 60% 정도의 시간이 단축되었다. 그리고 생성되는 압축 데이터의 크기는 full search 알고리즘을 사용할 때 보다 많이 차지한다. 생성되는 H.263파일의 화질에서는 Three-step 알고리즘을 사용한 경우일지라도 full search 알고리즘을 사용한 경우와 거의 비슷한 화질을 보여준다.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.7
• /
• pp.1255-1262
• /
• 2009

The present study explained the phenomenon that low frequency oscillation is synchronized with discrete data obtained from a wide area system, and a sync search method. When a disturbance occurs in an power system, various controllers operate in order to maintain synchronization. If the system's damping is poor, low frequency oscillations continue for a long time and the oscillations are synchronized with one another at specific frequency. The present study estimated dominant modes, magnitude and phase of signals by applying parameter estimation methods to discrete signals obtained from an power system, and performed sync search among wide area signals by comparing the estimated data. Sync search was performed by selecting those with the same frequency and damping constant from dominant oscillation modes included in a large number of signals, and comparing their magnitude and phase. In addition, we defined sync indexes in order to indicate the degree of sync between areas in a wide area system. Furthermore, we proposed a wide area sync search method by normalizing mode magnitude in discrete data obtained from critical generator of the wide area. By applying the sync search method and sync indexes proposed in this study to two area systems, we demonstrated that sync scanning can be performed for discrete signals obtained from power systems.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.3
• /
• pp.594-600
• /
• 2017

This paper proposes a global search based motion estimation algorithm for high performance HEVC encoder and its hardware architecture. To eliminate temporal redundancy, motion estimation in HEVC inter-view prediction uses global search and fast search algorithm to search for a predicted block having a high correlation with the current PU in an interpolated reference picture. The global search method predicts the motion of all candidate blocks in a given search area, thus ensuring optimal results, but has a disadvantage of large computation time. Therefore we propose a new algorithm that reduces computational complexity by reusing SAD operation in global search to reduce computation time of inter prediction. As a result of applying the proposed algorithm to standard software HM16.12, the computation time was reduced by 61%, BDBitrate by 11.81%, and BDPSNR by about 0.5% compared with the existing search algorithm. As a result of hardware design, the maximum operating frequency is 255 MHz and the total number of gates is 65.1K.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.1
• /
• pp.49-55
• /
• 2007

To improved the multi-resolution fast hierarchical motion estimation by using de-interlacing algorithm that is effective in term of both performance and VLSI implementation, is proposed so as to cover large search area field-based as well as frame based image processing in SoC design. In this paper, we have simulated a various picture mode M=2 or M=3. As a results, the proposed algorithm achieved the motion estimation performance PSNR compare with the full search block matching algorithm, the average performance degradation reached to -0.7dB, which did not affect on the subjective quality of reconstructed images at all. And acquiring the more desirable to adopt design SoC for the fast hierarchical motion estimation, we exploit foreground and background search algorithm (FBSA) base on the dual arithmetic processor element(DAPE). It is possible to estimate the large search area motion displacement using a half of number PE in general operation methods. And the proposed architecture of MHME improve the VLSI design hardware through the proposed FBSA structure with DAPE to remove the local memory. The proposed FBSA which use bit array processing in search area can improve structure as like multiple processor array unit(MPAU).

• Journal of Korea Multimedia Society
• /
• v.8 no.7
• /
• pp.889-897
• /
• 2005

This paper presents a new fast motion estimation algorithm for multi-reference frames. We first analyze the experimental results of the successive elimination algorithm, which is a fast version of full search algorithm, being applied to Multi-reference frames. Based on the analysis, a new scheme for alleviating its computational burden is introduced. In the proposed method, the motion vector for the immediately previous reference frame is found by applying the successive elimination algorithm, while the motion vector for other reference frames is estimated by extrapolation of the already obtained motion vector. Adaptively restricting the motion search area to the local area centered on the estimated motion vector, the proposed method provides dramatic computational complexity reduction but slight quality degradation. The proposed method is evaluated by experiments for some image sequences.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.405-406
• /
• 2008

This paper presents a high performance 2D PE architecture for H.264 Motion Estimation(ME). While existing 2D PE architectures reuse the overlapped data of adjacent search windows scanned in 1 or 3-way, the new architecture scan adjacent windows and multiple paths instead of single raster and zigzag scanning of adjacent windows in 4 way(up,down,left,right). By reducing the redundant access factor by 1.4, the new 4-way search window improve the memory bandwidth by 70-58% compared with 1/3-way search window. With Altera Stratix-III implementation, the high performance 2D PE architecture deals with SD ($720{\times}480$) video of 2 reference frame, $48{\times}48$ search area and $16{\times}16$ macroblock by 30fps at 97.1MHz.