loiwtjsf
Dołączył: 30 Sie 2010
Posty: 1562
Przeczytał: 0 tematów
Ostrzeżeń: 0/5 Skąd: ewcinl
|
Wysłany: Pią 10:03, 24 Gru 2010 Temat postu: stivali ugg Based on H. |
|
|
264/AVC the Multiple Reference Frame Selection Algorithm _83
Based on H. 264/AVC the Fast Selection Algorithm for Multiple Reference Frame
; 4 Feng ¨ 4 × 8 model reference frame using the most paternity and l6 × l6 l18 × 8 model level dry fI days, to determine l6 × 8 and 8 × l6 official reference frame model of the most fast selection algorithm in Figure 2 the flow chart Biao. 2Flowchartofthefastselectionalgorithm within the reference frame than the o, determine the best reference frame. In steps 1 and 3 using the texture features and motion information. In step 1, if the entire frame will be further divided into smaller areas for statistics, with better results. Threshold value can be assigned based on the actual situation, this experiment was 0.4. The threshold in step 2 and quantitative parameters of l, according to the quantized 4 × 4 matrix of the DC coefficient is equal to or less than 1 to determine the specific method can refer to [5]. Table 2 Quantitative parameters of the test results for 24 Tab. 2Resultsofpropose0algorithm, vitllQP = 24 Table 3 quantify the parameters of the test results for the 28 Tab. 3Resultsofproposedalgorithm, vitllQJP = 28 Table 4 quantify the parameters of the test results for the 32-Tab. 4Resultsofpropo ~ algorithm,[link widoczny dla zalogowanych], vitllQP = 325 conclusion was proposed based on H. 264 / C Fast multiple reference frame selection algorithms to select the reference frame sequence number of features. Experimental results show that the algorithm can guarantee the same signal to noise ratio and bit rate is almost the same time, obviously to save coding time. 4 results References: The combination of this algorithm and JM7.2 encoder key parameters include: entropy coding using CABAC; motion estimation using full search, the search range is 16.1 / 4 search precision; quantization parameter is 24 , 28 and 32. Experimental test of the five CIF sequences, including Foreman, Flow. er, Paris, Bus and Mobile. Table 2,3 and 4 and 5 are given a reference frame model of experimental results in check. Table APSNR that lower average SNR,[link widoczny dla zalogowanych], & d3itrate that rate changes, Atime that save time,[link widoczny dla zalogowanych], sequence test structure IBBP. Can be seen from the table, with the full search method for the model are compared to the BP structure can not save 30% of the time. SNR and bit rate changes also can be ignored. [1] WiegandT, SullivanG, BjontegaardG, eta1. OverviewoftheH. 264/Cvideocodingstandard [J]. IEEETransCircuitsandSystVideoTeehnol, 2003,13:560 ~ 576. [2] RavasiM, MattavelliM,. ClercC. AcomputationalcomplexitycomparisonofMPEG a 4and Ⅳ rcodecs [S]. JointVideoTeam (muscle) ofIS0/IECh Ⅱ 'DG & ITU-TVCEG, JVT-D153rl-L. 2o02. [3] H. 264/Creferencesoftware,[link widoczny dla zalogowanych], JM7.2 [DB / OL]. [link widoczny dla zalogowanych] 11l1i. de / suehring / tml /. 【4] WiegandT, Schwa = H, JochA, eta1. Rate-constrainedcodercontrolandcomparisonofvideocodingstandards [J]. 1EEETransCircuitsandSystVideoTechnol, 2003,13:688 ~ 703, [5] HuangYuwen, HsiehBingyu, WangTuchih, eta1. AnalysisandreductionofreferenceframesformotionestimationinMPEG a 4C / H. 264 [C]. ICASSP'03,[link widoczny dla zalogowanych], 2003,3: III A 145-148. (Edit Yang Bei)
相关的主题文章:
[link widoczny dla zalogowanych]
[link widoczny dla zalogowanych]
[link widoczny dla zalogowanych]
Post został pochwalony 0 razy
|
|