(Special Paper) 19 6, 2014 11 (JBE Vol. 19, No. 6, November 2014) http://dx.doi.org/10.5909/jbe.2014.19.6.836 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) HEVC a), a) Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter Hong-rae Lee a) and Kwang-deok Seo a) (UHD: Ultra High Definition) H.264/AVC HEVC (High-Efficiency Video Codec). HEVC H.264/AVC. HEVC 35 H.264/AVC. 35 4 PU (Prediction Unit). 2% BD-rate 7%. Abstract To expedite UHD (Ultra High Definition) video service, the HEVC (High-Efficiency Video Coding) technology has recently been standardized and it achieves two times higher compression efficiency than the conventional H.264/AVC. To obtain the improved efficiency, however, it employs many complex methods which need complicated calculation, thereby resulting in a significantly increased computational complexity when compared to that of H.264/AVC. For example, to improve the coding efficiency of intra frame coding, up to 35 intra prediction modes are defined in HEVC, but this results in an increased encoding time than the H.264/AVC. In this paper, we propose a fast intra prediction mode decision scheme which reduces computational complexity by changing the number of intra prediction mode in accordance with the percentage of PU sizes for a given video resolution, and by classifying the 35 intra prediction modes into 4 categories considering video resolution and quantization parameter. The experimental results show that the total encoding time is reduced by about 7% on average at the cost of only 2% increase in BD-rate. Keyword : HEVC Intra coding, Quantization Parameter, Fast encoder design, Intra Prediction Mode
1 : HEVC (Hong-rae Lee et al. : Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter). UHD (Ultra High Definition),, Full HD. Full HD UHD [1].. ISO/IEC MPEG ITU-T VCEG 2010 1 JCT-VC (Joint Collaborative Team on Video Coding) HEVC (High Efficiency Video Coding) [2] 2013 1. HEVC H.264/AVC [3], 40~50% [4],[5] HEVC,,,., HEVC CTU (Coding Tree Unit). CTU 64 64, 32 32, 16 16 [6]. CTU (Quad-tree) CTU CU (Coding Unit). CU a) (Yonsei University) Corresponding Author : (Kwang-deok Seo) E-mail: kdseo@yonsei.ac.kr Tel: +82-33-760-2788 2013 ( ) (No. NRF-2013R1A1A2011635). 2014. Manuscript received September 2, 2014 Revised October 27,2014 Accepted October 27,2014 64 64, 32 32, 16 16, 8 8, CU PU(Prediction Unit) TU(Transform Unit). PU CU 64 64, 32 32, 16 16, 8 8, 4 4. CU 2N 2N, 2N N, N 2N, N N. TU [7]. H.264/AVC 16 16 4, 8 8 4 4 9 [8] HEVC 35 [9],[10]. [11],[12]. Y. Piao RDO (Rate distortion optimization) RMD (Rough Mode Decision) [13]. RMD N PU 64 64, 32 32, 16 16 3, 8 8, 4 4 8. RMD RDO RMD 35. H.264/AVC 16 16 4, 8 8 4 4 9. HEVC PU. [14]. RMD. 2 HEVC
RMD. 3 4.. HEVC HEVC PU CU 2N 2N 8 8 CU N N 4 4 PU. PU (luma) 35, (chroma) 6. PU. H.264/AVC HEVC PU. PU HEVC Class B, Class C, Class D test QP (Quantization Parameter) 22, 27, 32, 37 All-Intra. test QP PU. 1, 8 8 PU,. 64 64 PU,. HEVC Class B, Class C, Class D test QP 22, 27, 32, 37 All-Intra. test QP. 2, 3, 4 Class B, Class C, Class D 1%. Class Planar(0), DC(1), Horizontal(10), Vertical(26). Horizontal(10), Vertical(26). 8 8 Vertical(26) 1. PU Table 1. Selection ratio of PU in accordance with video resolution and quantization parameter Class (Video Resolution) B Class (1920 1080) C Class (832 480) D Class (416 240) PU_Size QP=22 QP=27 QP=32 QP=37 8 43.87% 28.44% 20.55% 13.39% 16 24.65% 28.36% 29.08% 27.41% 32 25.78% 34.14% 36.97% 40.74% 64 5.70% 9.07% 13.41% 18.46% 8 67.55% 59.12% 49.40% 38.32% 16 21.46% 26.06% 31.88% 35.75% 32 9.75% 13.21% 16.63% 22.50% 64 1.24% 1.61% 2.09% 3.44% 8 70.11% 63.09% 56.91% 46.41% 16 17.85% 21.83% 25.33% 30.93% 32 11.53% 14.20% 16.59% 20.66% 64 0.51% 0.87% 1.17% 2.01%
1 : HEVC (Hong-rae Lee et al. : Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter) 2. Class B QP (QP=22, 27, 32, 37 ) Table 2. Selection ratio of Prediction mode in accordance with quantization parameter from Class B (QP=22, 27, 32, 37) QP = 22 QP = 27 0 25.67% 24.92% 22.10% 7.27% 0 24.99% 25.47% 27.84% 19.91% 1 17.61% 25.98% 41.92% 57.81% 1 15.62% 20.19% 31.95% 41.10% 2 1.13% 0.59% 0.66% 0.17% 2 0.92% 0.60% 0.76% 0.70% 3 0.67% 0.58% 0.49% 0.08% 3 0.72% 0.65% 0.60% 0.63% 4 0.67% 0.71% 0.53% 0.11% 4 0.73% 0.76% 0.66% 0.65% 5 0.94% 0.96% 0.57% 0.11% 5 1.03% 1.10% 1.07% 1.08% 6 2.30% 2.79% 0.77% 0.14% 6 2.87% 3.45% 1.72% 1.05% 7 2.85% 2.19% 0.83% 0.15% 7 3.41% 4.03% 2.51% 0.65% 8 2.79% 2.94% 1.19% 0.23% 8 3.06% 3.55% 1.97% 0.64% 9 3.73% 2.87% 2.61% 0.96% 9 3.19% 3.05% 2.88% 1.45% 10 4.03% 3.84% 4.11% 5.64% 10 4.17% 3.80% 4.47% 6.68% 11 3.25% 2.27% 2.06% 0.66% 11 2.88% 2.50% 2.35% 1.32% 12 1.92% 1.97% 0.62% 0.11% 12 1.73% 2.00% 0.72% 0.37% 13 1.56% 1.49% 0.59% 0.25% 13 1.46% 1.55% 0.66% 0.51% 14 1.52% 1.07% 0.44% 0.17% 14 1.61% 1.20% 0.48% 0.55% 15 1.19% 0.80% 0.37% 0.09% 15 1.43% 0.88% 0.37% 0.25% 16 0.90% 0.64% 0.32% 0.12% 16 1.00% 0.77% 0.31% 0.21% 17 0.89% 0.55% 0.27% 0.12% 17 1.00% 0.61% 0.26% 0.36% 18 1.31% 0.51% 0.27% 0.13% 18 1.21% 0.56% 0.24% 0.32% 19 0.79% 0.45% 0.27% 0.13% 19 0.87% 0.52% 0.24% 0.33% 20 0.80% 0.49% 0.28% 0.17% 20 0.91% 0.58% 0.23% 0.32% 21 0.89% 0.58% 0.28% 0.10% 21 0.87% 0.64% 0.25% 0.27% 22 1.04% 0.92% 0.33% 0.09% 22 1.12% 0.86% 0.31% 0.27% 23 1.14% 1.04% 0.37% 0.07% 23 1.20% 1.01% 0.43% 0.22% 24 1.43% 1.27% 0.45% 0.14% 24 1.39% 1.36% 0.53% 0.30% 25 2.90% 3.50% 4.08% 11.04% 25 2.62% 2.97% 2.77% 7.02% 26 7.72% 7.46% 9.08% 13.06% 26 9.82% 8.29% 8.49% 10.01% 27 2.12% 1.25% 1.05% 0.42% 27 2.03% 1.45% 1.25% 0.91% 28 1.26% 1.27% 0.54% 0.14% 28 1.26% 1.26% 0.68% 0.43% 29 0.99% 1.08% 0.50% 0.07% 29 1.02% 1.13% 0.65% 0.29% 30 0.91% 0.81% 0.39% 0.06% 30 0.92% 0.83% 0.50% 0.20% 31 0.82% 0.66% 0.41% 0.05% 31 0.73% 0.75% 0.47% 0.21% 32 0.63% 0.57% 0.40% 0.03% 32 0.69% 0.57% 0.44% 0.21% 33 0.65% 0.51% 0.43% 0.06% 33 0.70% 0.54% 0.41% 0.21% 34 0.99% 0.48% 0.43% 0.04% 34 0.84% 0.52% 0.51% 0.36% QP = 32 QP = 37 0 27.08% 27.18% 30.94% 30.00% 0 28.77% 28.68% 31.66% 31.97% 1 15.27% 18.38% 26.53% 30.57% 1 15.32% 17.41% 23.40% 26.45% 2 0.91% 0.69% 0.65% 0.74% 2 1.02% 0.79% 0.70% 1.46% 3 0.64% 0.60% 0.50% 0.74% 3 0.65% 0.64% 0.50% 0.82% 4 0.68% 0.73% 0.65% 0.87% 4 0.65% 0.73% 0.70% 1.02% 5 1.03% 1.00% 1.16% 1.68% 5 1.03% 1.02% 1.17% 1.88% 6 2.89% 3.95% 2.68% 1.52% 6 2.81% 3.82% 3.81% 2.43% 7 3.37% 4.39% 3.02% 0.91% 7 3.20% 4.24% 3.65% 1.36% 8 2.85% 3.12% 2.05% 0.86% 8 2.39% 2.97% 2.11% 1.18% 9 2.72% 2.91% 3.08% 1.50% 9 2.31% 2.51% 2.72% 1.52% 10 3.73% 3.68% 4.24% 6.01% 10 3.14% 3.30% 3.88% 5.45% 11 2.41% 2.68% 2.38% 1.29% 11 2.10% 2.39% 2.36% 1.28% 12 1.52% 1.67% 0.80% 0.43% 12 1.41% 1.48% 0.89% 0.47% 13 1.29% 1.42% 0.73% 0.62% 13 1.09% 1.26% 0.75% 0.65% 14 1.60% 1.26% 0.59% 0.48% 14 1.59% 1.31% 0.72% 0.51% 15 1.37% 0.90% 0.39% 0.19% 15 1.68% 0.97% 0.53% 0.23% 16 1.04% 0.73% 0.34% 0.18% 16 1.02% 0.76% 0.42% 0.24% 17 1.01% 0.69% 0.29% 0.16% 17 0.97% 0.75% 0.33% 0.17% 18 1.21% 0.64% 0.25% 0.16% 18 1.35% 0.63% 0.29% 0.16% 19 0.92% 0.57% 0.25% 0.20% 19 0.95% 0.70% 0.28% 0.16% 20 0.97% 0.63% 0.26% 0.25% 20 1.03% 0.75% 0.31% 0.16% 21 0.95% 0.68% 0.24% 0.28% 21 1.00% 0.80% 0.31% 0.20% 22 1.12% 0.82% 0.31% 0.18% 22 1.08% 0.82% 0.34% 0.19% 23 1.18% 0.95% 0.46% 0.34% 23 1.24% 1.00% 0.47% 0.39% 24 1.34% 1.29% 0.60% 0.43% 24 1.28% 1.37% 0.67% 0.53% 25 2.51% 2.97% 2.72% 4.92% 25 2.24% 2.97% 2.59% 3.18% 26 10.52% 8.62% 9.12% 11.05% 26 10.82% 9.19% 9.56% 11.94% 27 1.82% 1.46% 1.23% 0.98% 27 1.68% 1.39% 1.22% 0.89% 28 1.22% 1.19% 0.69% 0.51% 28 1.27% 1.15% 0.67% 0.54% 29 1.02% 1.05% 0.64% 0.42% 29 1.14% 1.03% 0.65% 0.51% 30 0.94% 0.78% 0.51% 0.24% 30 0.90% 0.80% 0.53% 0.38% 31 0.75% 0.67% 0.47% 0.26% 31 0.73% 0.63% 0.50% 0.34% 32 0.66% 0.57% 0.40% 0.25% 32 0.64% 0.57% 0.41% 0.29% 33 0.67% 0.54% 0.37% 0.32% 33 0.60% 0.54% 0.39% 0.37% 34 0.82% 0.57% 0.47% 0.49% 34 0.92% 0.61% 0.53% 0.66%
3. Class C QP (QP=22, 27, 32, 37 ) Table 3. Selection ratio of Prediction mode in accordance with quantization parameter from Class C (QP=22, 27, 32, 37) QP = 22 QP = 27 0 16.49% 19.58% 25.86% 8.81% 0 18.73% 20.69% 23.34% 20.90% 1 13.31% 17.20% 38.77% 70.95% 1 14.50% 16.65% 30.95% 47.40% 2 1.17% 0.73% 0.97% 0.08% 2 1.25% 0.75% 0.72% 0.11% 3 1.00% 0.55% 0.35% 0.08% 3 1.04% 0.60% 0.36% 0.07% 4 0.93% 0.71% 0.33% 0.00% 4 0.94% 0.67% 0.41% 0.10% 5 1.01% 0.79% 0.33% 0.04% 5 0.88% 0.60% 0.42% 0.09% 6 1.92% 1.27% 2.02% 0.06% 6 2.14% 1.78% 3.96% 0.03% 7 2.68% 2.43% 0.41% 0.04% 7 2.29% 2.32% 0.44% 0.12% 8 1.74% 1.58% 0.90% 0.38% 8 1.53% 1.50% 1.18% 0.19% 9 2.82% 2.65% 1.84% 0.44% 9 2.56% 2.66% 2.48% 0.68% 10 4.61% 7.06% 2.65% 0.09% 10 4.30% 8.11% 4.09% 1.40% 11 2.27% 1.85% 1.50% 0.40% 11 2.05% 1.96% 1.77% 1.05% 12 1.52% 1.08% 0.39% 0.18% 12 1.39% 1.08% 0.54% 0.47% 13 1.13% 0.70% 0.50% 0.00% 13 1.04% 0.60% 0.41% 0.05% 14 1.07% 0.56% 0.15% 0.00% 14 0.95% 0.59% 0.26% 0.06% 15 1.10% 0.77% 0.24% 0.02% 15 1.03% 0.62% 0.26% 0.00% 16 1.50% 2.25% 0.20% 0.00% 16 1.19% 2.04% 0.28% 0.02% 17 1.81% 2.28% 0.31% 0.00% 17 1.69% 1.44% 0.48% 5.52% 18 3.75% 3.01% 1.57% 0.00% 18 3.59% 2.62% 2.41% 0.02% 19 3.17% 2.40% 1.25% 10.71% 19 2.81% 2.58% 3.54% 2.53% 20 2.65% 2.56% 1.91% 0.00% 20 2.51% 2.48% 1.47% 1.53% 21 2.41% 3.34% 1.10% 0.04% 21 2.28% 2.85% 2.13% 0.02% 22 3.22% 2.54% 0.82% 0.13% 22 3.17% 2.56% 1.08% 0.05% 23 2.44% 2.80% 2.52% 0.04% 23 2.19% 3.16% 1.72% 0.21% 24 2.00% 1.86% 0.78% 0.09% 24 1.79% 1.74% 1.07% 0.12% 25 2.27% 1.54% 1.33% 0.22% 25 2.07% 1.52% 1.47% 0.19% 26 8.34% 7.25% 6.76% 6.90% 26 9.27% 7.64% 8.24% 16.68% 27 2.07% 1.54% 1.07% 0.06% 27 1.85% 1.58% 1.32% 0.10% 28 1.89% 2.03% 0.93% 0.04% 28 1.73% 1.73% 0.91% 0.03% 29 1.64% 1.52% 0.65% 0.09% 29 1.42% 1.33% 0.47% 0.10% 30 1.48% 1.14% 0.34% 0.00% 30 1.44% 1.11% 0.55% 0.05% 31 1.33% 0.91% 0.46% 0.04% 31 1.20% 0.85% 0.33% 0.07% 32 1.32% 0.67% 0.24% 0.02% 32 1.28% 0.69% 0.37% 0.02% 33 1.00% 0.52% 0.29% 0.00% 33 0.88% 0.50% 0.28% 0.00% 34 0.94% 0.32% 0.24% 0.00% 34 0.99% 0.42% 0.29% 0.02% QP = 32 QP = 37 0 21.90% 22.52% 24.17% 16.60% 0 25.47% 24.66% 25.37% 22.76% 1 15.21% 17.50% 23.66% 32.78% 1 14.84% 16.91% 20.23% 26.10% 2 1.28% 0.71% 0.65% 0.44% 2 1.28% 0.78% 0.80% 1.07% 3 0.85% 0.60% 0.42% 0.53% 3 0.80% 0.56% 0.45% 0.70% 4 0.80% 0.83% 0.49% 0.13% 4 0.69% 0.90% 0.55% 0.84% 5 0.78% 0.64% 0.56% 0.14% 5 0.84% 0.82% 0.54% 0.53% 6 2.67% 2.00% 4.09% 1.85% 6 2.57% 2.45% 3.37% 4.12% 7 2.07% 1.61% 0.49% 0.15% 7 1.80% 2.01% 0.73% 0.46% 8 1.44% 1.59% 1.79% 0.73% 8 1.25% 1.44% 1.64% 0.56% 9 2.43% 2.64% 2.50% 0.62% 9 2.19% 2.49% 2.46% 0.76% 10 3.93% 7.83% 5.22% 2.29% 10 3.56% 6.66% 8.00% 2.47% 11 2.05% 1.99% 1.60% 0.66% 11 2.05% 1.93% 1.93% 0.60% 12 1.33% 1.16% 0.64% 0.37% 12 1.32% 1.14% 0.59% 0.33% 13 0.95% 0.61% 0.57% 0.09% 13 0.88% 0.68% 0.54% 0.13% 14 0.94% 0.62% 0.34% 0.04% 14 0.82% 0.53% 0.46% 0.07% 15 0.98% 0.73% 0.44% 0.03% 15 0.88% 0.64% 0.39% 0.04% 16 1.08% 0.95% 0.42% 0.22% 16 0.95% 0.90% 0.73% 0.27% 17 1.36% 1.55% 1.16% 2.09% 17 1.31% 1.26% 1.12% 1.04% 18 2.83% 2.73% 2.83% 2.56% 18 2.50% 2.22% 2.48% 4.86% 19 2.10% 2.62% 2.99% 6.60% 19 1.77% 2.05% 2.63% 1.45% 20 2.46% 2.23% 1.84% 0.85% 20 2.37% 2.53% 1.99% 1.44% 21 1.99% 2.42% 1.80% 2.82% 21 1.88% 2.11% 1.79% 1.44% 22 2.70% 2.17% 1.41% 0.43% 22 2.72% 2.14% 1.84% 0.96% 23 1.99% 2.50% 2.16% 0.31% 23 1.79% 2.17% 1.83% 0.68% 24 1.70% 1.77% 0.97% 0.34% 24 1.42% 1.74% 0.99% 0.26% 25 1.80% 1.41% 1.61% 0.30% 25 1.60% 1.46% 1.57% 0.49% 26 10.09% 7.69% 9.69% 24.89% 26 10.59% 7.92% 9.15% 23.37% 27 1.61% 1.62% 1.47% 0.27% 27 1.40% 2.08% 1.60% 0.66% 28 1.46% 1.81% 1.01% 0.18% 28 1.35% 1.71% 0.96% 0.34% 29 1.40% 1.38% 0.77% 0.24% 29 1.32% 1.34% 0.81% 0.55% 30 1.38% 0.97% 0.61% 0.19% 30 1.12% 1.02% 0.69% 0.22% 31 1.20% 0.84% 0.61% 0.21% 31 1.31% 0.89% 0.59% 0.26% 32 1.37% 0.78% 0.34% 0.05% 32 1.37% 0.70% 0.52% 0.11% 33 0.86% 0.51% 0.35% 0.01% 33 0.95% 0.58% 0.33% 0.04% 34 1.01% 0.45% 0.33% 0.03% 34 1.04% 0.57% 0.34% 0.02%
1 : HEVC (Hong-rae Lee et al. : Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter) 4. Class D QP (QP=22, 27, 32, 37 ) Table 4. Selection ratio of Prediction mode in accordance with quantization parameter from Class D (QP=22, 27, 32, 37) QP = 22 QP = 27 0 18.01% 22.43% 24.93% 6.23% 0 20.74% 24.56% 27.87% 12.52% 1 12.71% 23.01% 42.26% 61.13% 1 13.66% 19.99% 40.22% 56.01% 2 1.90% 0.68% 1.39% 0.00% 2 1.94% 0.69% 0.75% 0.00% 3 1.40% 0.39% 0.43% 0.00% 3 1.36% 0.54% 0.47% 0.00% 4 1.40% 0.67% 0.23% 0.00% 4 1.14% 1.03% 0.29% 0.00% 5 1.63% 1.14% 0.68% 0.00% 5 1.43% 1.04% 0.72% 0.00% 6 2.32% 1.13% 0.29% 0.00% 6 2.19% 1.31% 0.26% 0.00% 7 2.00% 1.26% 0.22% 0.00% 7 1.75% 1.17% 0.35% 0.25% 8 1.72% 1.50% 0.35% 0.00% 8 1.45% 1.26% 0.50% 0.00% 9 2.15% 1.41% 1.60% 0.00% 9 1.71% 1.33% 1.66% 0.26% 10 4.03% 5.10% 2.22% 0.59% 10 3.46% 4.96% 2.07% 1.32% 11 4.05% 7.92% 4.70% 5.59% 11 3.25% 7.77% 4.50% 1.85% 12 2.53% 3.48% 0.97% 0.00% 12 2.16% 3.69% 1.08% 0.13% 13 1.76% 0.90% 0.27% 0.00% 13 1.66% 1.13% 0.26% 0.00% 14 1.73% 0.84% 0.08% 0.00% 14 1.88% 0.69% 0.13% 0.00% 15 1.51% 0.44% 0.17% 0.00% 15 1.48% 0.68% 0.29% 0.00% 16 1.21% 0.97% 0.07% 0.00% 16 1.19% 1.11% 0.22% 0.00% 17 1.37% 0.62% 0.06% 0.00% 17 1.32% 0.68% 0.03% 0.00% 18 1.74% 0.48% 0.17% 0.00% 18 1.83% 0.60% 0.10% 0.00% 19 1.62% 0.81% 0.08% 0.00% 19 1.56% 0.61% 0.16% 0.00% 20 1.61% 0.91% 0.04% 0.00% 20 1.44% 1.36% 0.10% 0.00% 21 1.55% 0.85% 0.26% 0.00% 21 1.49% 0.87% 0.23% 0.00% 22 2.17% 1.24% 0.28% 0.00% 22 2.11% 1.22% 0.18% 0.00% 23 2.22% 1.55% 0.30% 0.00% 23 2.06% 1.54% 0.27% 0.00% 24 2.75% 4.43% 4.59% 0.00% 24 2.51% 3.80% 4.02% 0.00% 25 2.51% 2.28% 6.26% 0.00% 25 2.15% 2.33% 4.35% 0.00% 26 7.05% 5.87% 3.89% 1.47% 26 8.29% 6.08% 5.29% 2.51% 27 1.99% 1.33% 1.44% 0.00% 27 1.87% 1.38% 1.53% 0.00% 28 2.09% 1.47% 0.44% 0.00% 28 1.84% 1.30% 0.42% 0.13% 29 1.83% 1.07% 0.08% 0.00% 29 1.75% 1.29% 0.29% 0.00% 30 2.20% 1.36% 0.17% 0.00% 30 2.02% 1.34% 0.21% 0.00% 31 1.62% 0.91% 0.21% 0.00% 31 1.57% 1.07% 0.34% 0.00% 32 1.24% 1.05% 0.32% 0.00% 32 1.25% 0.93% 0.29% 0.00% 33 1.17% 0.33% 0.29% 0.00% 33 1.26% 0.37% 0.27% 0.00% 34 1.19% 0.20% 0.30% 0.00% 34 1.25% 0.27% 0.28% 0.00% QP = 32 QP = 37 0 25.34% 27.47% 30.44% 30.67% 0 29.88% 32.09% 36.50% 40.82% 1 14.83% 17.81% 33.29% 58.74% 1 15.27% 17.48% 24.06% 37.41% 2 1.90% 0.80% 0.81% 1.03% 2 1.77% 0.92% 1.32% 1.58% 3 1.13% 0.71% 0.74% 0.17% 3 0.87% 0.83% 0.75% 0.74% 4 0.96% 1.28% 0.67% 0.17% 4 0.92% 1.00% 1.18% 2.43% 5 1.29% 1.17% 1.19% 0.09% 5 1.09% 1.07% 1.81% 1.33% 6 1.78% 1.24% 1.45% 0.00% 6 1.50% 1.27% 2.12% 1.08% 7 1.59% 1.38% 1.29% 0.17% 7 1.41% 1.04% 1.18% 0.77% 8 1.41% 1.49% 0.68% 0.17% 8 1.05% 1.30% 1.29% 1.38% 9 1.39% 1.12% 1.49% 0.98% 9 1.13% 0.98% 1.56% 2.27% 10 3.13% 3.69% 3.40% 0.64% 10 2.27% 3.04% 3.64% 1.69% 11 2.74% 7.47% 4.62% 1.55% 11 2.28% 6.43% 5.53% 3.12% 12 1.94% 3.11% 1.34% 0.09% 12 1.70% 1.92% 1.26% 0.39% 13 1.40% 1.29% 0.26% 0.00% 13 1.09% 1.03% 0.50% 0.04% 14 1.45% 0.71% 0.15% 0.00% 14 1.39% 0.87% 0.33% 0.00% 15 1.19% 1.07% 0.31% 0.00% 15 1.00% 1.07% 0.37% 0.00% 16 0.95% 1.26% 0.25% 0.00% 16 0.91% 1.16% 0.33% 0.00% 17 1.22% 0.69% 0.11% 0.00% 17 1.25% 0.71% 0.14% 0.10% 18 1.78% 0.71% 0.20% 0.00% 18 1.51% 0.88% 0.30% 0.00% 19 1.48% 0.82% 0.40% 0.00% 19 1.54% 1.00% 0.28% 0.10% 20 1.29% 1.13% 0.06% 0.00% 20 1.26% 1.29% 0.35% 0.00% 21 1.52% 1.04% 0.32% 0.00% 21 1.39% 0.96% 0.45% 0.00% 22 1.83% 1.36% 0.22% 0.00% 22 1.45% 1.14% 0.39% 0.00% 23 1.75% 1.55% 0.34% 0.00% 23 1.85% 1.62% 0.58% 0.10% 24 2.04% 2.91% 2.00% 0.00% 24 1.66% 1.99% 1.48% 0.04% 25 2.00% 1.90% 4.07% 0.00% 25 1.53% 1.39% 1.78% 0.04% 26 8.90% 6.44% 5.42% 5.24% 26 10.19% 6.73% 6.18% 3.99% 27 1.64% 1.13% 1.59% 0.09% 27 1.47% 1.18% 1.03% 0.09% 28 1.60% 1.46% 0.50% 0.00% 28 1.66% 1.81% 0.67% 0.00% 29 1.61% 1.39% 0.39% 0.00% 29 1.36% 1.31% 0.33% 0.00% 30 1.86% 1.42% 0.29% 0.00% 30 1.55% 1.18% 0.41% 0.10% 31 1.59% 0.95% 0.40% 0.00% 31 1.56% 1.09% 0.36% 0.09% 32 1.15% 0.84% 0.45% 0.17% 32 1.06% 0.93% 0.41% 0.00% 33 1.21% 0.66% 0.36% 0.00% 33 1.02% 0.59% 0.29% 0.10% 34 1.11% 0.55% 0.51% 0.00% 34 1.14% 0.67% 0.85% 0.20%
. Horizontal(10). 64 64 Vertical(26) ( 2 ), Vertical(26) ( 3, 4 ). QP 8 8 2~4(Class B~D) 1% 16 16 3(Class C) 1%, 2(Class B) 4(Class D) 1%. 32 32 64 64 2~4 (Class B~D) 1%. Class PU 1%. 1 PU 1%. 8 8 Class D 70% Class B 40% 1% 4(Class D) 1, 3(Class C) 6, 2(Class B) 12 1. RMD 1, 2, 3, 4 Fig. 1. Proposed candidate SET 1, 2, 3, 4 for RMD prediction
1 : HEVC (Hong-rae Lee et al. : Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter). 16 16 32 32 QP, 3(Class C) 1% 16 16 12, 32 32 19 2(Class B) 4(Class D) 1% 2~6 Planar(0), DC(1). 64 64 1%. RMD 35 PU. HEVC RMD 35 4 PU. 5 RMD 4. SET 1 4 SET 2 Horizontal(10), Vertical(26) 2 18. 2, 3, 4 Horizontal(10), Vertical(26) 1% 18. SET 3 Horizontal(10), Vertical (26) 3 18 1. SET 4 5. RMD Table 5. Candidate sets of the proposed RMD prediction mode SET 1 RMD DC(0), Plnanar(1), Horizontal(10), Vertical(26) SET 2 DC(0), Plnanar(1), 8,9,10,11,12,18,24,25,26,27,28 SET 3 SET 4 DC(0), Plnanar(1), 7,8,9,10,11,12,13,17,18,19,23,24,25,26,27,28,29 35 modes 35. 1.., HEVC HM12.0 [15] HEVC. 6 7. 6. Table 6. Test environment CPU RAM OS Compiler Intel Core(TM) i7-4770 CPU @ 3.40Ghz 8.00 GB Microsoft Windows 7 64-bit Visual C++ 2010 sp1 x86 Test Model HM 12.0 Configure All intra main QP 22, 27, 32, 37 7. Table 7. Test sequence Class (Video Resolution) B Class (1920 1080) C Class (832 480) D Class (416 240) Sequence Frames Frame rate Kimono 240 24 ParkScene 240 24 Cactus 500 50 BasketballDrive 500 50 BQTerrace 600 60 BasketballDrill 500 50 BQMall 600 60 PartyScene 500 50 RaceHorsesC 300 30 BasketballPass 500 50 BQSquare 600 60 BlowingBubbles 500 50 RaceHorses 300 30
8 QP PU 5. HEVC Time (1). Class (Video Resolution) B Class (1920 1080) C Class (832 480) D Class (416 240) 8. QP PU Table 8. Mode selection of considering each video resolution, QP and PU ratio PU_Size QP=22 QP=27 QP=32 QP=37 8 SET 3 SET 3 SET 2 SET 2 16 SET 2 SET 2 SET 2 SET 2 32 SET 2 SET 2 SET 3 SET 3 64 SET 1 SET 1 SET 2 SET 2 8 SET 4 SET 4 SET 3 SET 3 16 SET 3 SET 3 SET 3 SET 3 32 SET 1 SET 1 SET 2 SET 2 64 SET 1 SET 1 SET 2 SET 2 8 SET 4 SET 4 SET 3 SET 3 16 SET 3 SET 3 SET 2 SET 2 32 SET 2 SET 2 SET 2 SET 2 64 SET 1 SET 1 SET 1 SET 1 9. Class B 6.7% 3.8%. Class C 3.6% 2.9%. Class D 3.8% 2.2%. Class C D QP 32 37 6% Class B 6%. 8 8 PU Class C D HEVC 35 SET 4 Class B 9. Table 9. Test results Class Sequence BD-rate Y QP Time 22-6.83% Kimono 2.8% 27-7.09% 32-7.35% 37-7.81% 22-6.86% ParkScene 1.2% 27-6.62% 32-7.07% 37-7.45% 22-7.19% 27-6.35% Cactus 4.5% B 32-6.94% 37-7.54% 22-6.26% BasketballDrive 5.9% 27-6.59% 32-6.90% 37-6.21% 22-5.12% BQTerrace 4.7% 27-5.92% 32-6.19% 37-5.93% Overall 3.8% -6.71% Class Sequence BD-rate Y QP Time 22-1.52% BasketballDrill 6.1% 27-1.78% 32-4.70% 37-5.49% 22-2.47% BQMall 1.4% 27-2.69% 32-4.72% 37-5.49% C 22-1.58% PartyScene 1.1% 27-1.87% 32-5.29% 37-4.89% 22-2.66% RaceHorses 2.8% 27-2.92% 32-4.96% 37-5.09% Overall 2.9% -3.63% Class Sequence BD-rate Y QP Time 22-1.14% BasketballPass 2.3% 27-1.50% 32-6.25% 37-6.32% 22-1.18% BQSquare 1.8% 27-1.72% 32-6.52% 37-6.25% D 22-1.07% BlowingBubbles 1.8% 27-1.52% 32-6.26% 37-6.19% 22-1.56% RaceHorses 2.9% 27-1.84% 32-6.28% 37-6.58% Overall 2.2% -3.89% Overall 3.0% -4.7%
1 : HEVC (Hong-rae Lee et al. : Scheme for Reducing HEVC Intra Coding Complexity Considering Video Resolution and Quantization Parameter) SET 1~3. 8 8 PU Class C D 8 8 PU Class C D Class B. Class B Frame rate 50 Cactus, BasketballDrive, BQTerrace 5% 6.4% Frame rate 24 Kimono 2.8% 7%, ParkScene 1.2% 7.3%. 8 8 PU 8 8 PU Class B Frame rate.. HEVC RMD 35 4 RMD PU 4. HM 12.0, 8 8 PU, Class B Flame rate 2.0% 7.1%. UHD Flame rate QP. (References) [1] U. Parka, H. Choi, J. Kang, and J. Kim, Scalable Video Coding Using Large Block and its Performance Analysis, Journal of Broadcast Engineering, Vol. 18, No. 2, pp. 249-260, Mar. 2013. [2] JCT-VC, High Efficiency Video Coding(HEVC) text specification draft 8, JCTVC-J1003, Joint Collaborative Team on Video Coding (JCT-VC) 10th Meeting, Stockholm, Sweden, July 2012. [3] ITU-T and ISO/IEC JTC 1, Advanced video coding for generic audiovisual services, ITU-T Recommendation H.264 and ISO/IEC 14496-10 (MPEG4-AVC), Forth edition: Sep. 2008. [4] B. Li, G. Sullivan, and J. Xu, Comparison of compression performance of HEVC working draft 5 with AVC high profile, JCTVC-H0360, San Jose, CA, Feb. 2012. [5] J. Ohm, G. Sullivan, H. Schwarz, T. Tan, and T. Wiegand, Comparison of the Coding Efficiency of Video Coding Standards-Including High Efficiency Video Coding (HEVC), IEEE Transactions on Circuits and Systems for Video Technology, Vol. 22, No. 12, pp. 1669-1684, Dec. 2012. [6] G. Sullivan, and R. Baker, Efficient Quadtree Coding of Images and Video, IEEE Transactions on Image Processing, Vol. 3 No. 3 pp. 327-331, May 1994. [7] G. Sullivan, J. Ohm, W. Han, and T. Wiegand, Overview of the High Efficiency Video Coding (HEVC) Standard, IEEE Transactions on circuits and systems for video technology, Vol. 22, No. 12, Dec. 2012 [8] B. La, M. Eom, and Y. Choe, Fast Mode Decision for Intra Prediction in H.264/AVC Encoder, IEEE International Conference on Image Processing, Vol. 5, pp. 321-324 Oct. 2007. [9] J. Min, S. Lee, I. Kim, W. Han, J. Lainema, and K. Ugur, Unification of the directional intra prediction methods in TMuC, document JCTVC-B100, Mar. 2012. [10] F. Bossen, T. Tan, and J. Takiue, simplified angular intra prediction, document JCTVC-B093, Jul. 2010. [11] F. Bossen, B. Bross, K. Suhring, and D. Flynn HEVC Complexity and Implementation Analysis, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 22, No. 12 pp. 1685-1696, Dec. 2012. [12] J. Vanne, M. Vitanen, T. Hamalainen, and A. Hallpuro, Comparative Rate-Distortion-Complexity Analysis of HEVC and AVC Video Codecs, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 22, No. 12, pp. 1885-1898, Dec. 2012. [13] JCT-VC, Encoder improvement of unified intra prediction, Document JCTVC-C207, Guangzhou, OCT. 2010. [14] D. Gwon, S. Lee, and H. Choi, HEVC Intra Prediciton Performance Analysis and Complexity Reduction Method, Proceeding of the Korean Society of Broadcast Engineer Conference, Vol. 2012, No. 11, pp. 176-179, Nov. 2012. [15] I. Kim, K. McCann, K. Sugimoto, B. Bross, and W. Han, High Efficiency Video Coding (HEVC) Test Model 12 (HM 12) Encoder Description, JCTVC-M1002, 13th JCT-VC Meeting, April 2013.
- 2010 2 : - 2012 8 : - 2012 9 ~ : - :,, - 1996 2 : KAIST - 1998 2 : KAIST - 2002 8 : KAIST - 2002 8 ~ 2005 2 : LG - 2012 9 ~2013 8 : Courtesy Professor, Univ. of Florida, USA - 2005 3 ~ : - :,,,