新一代視頻編碼幀內預測模式選擇的優(yōu)化
發(fā)布時間：2018-12-10 19:36
【摘要】：提出了一種基于圖像相關性和最優(yōu)模式概率統(tǒng)計的幀內預測優(yōu)化算法,用于降低視頻編碼的復雜度。首先,介紹了新一代視頻編碼標準(HEVC)幀內預測算法中的Angular預測模式、Planar預測模式、LM預測模式,以及幀內率失真代價最優(yōu)化(RDO)計算方法。采用絕對誤差和(SAD)作為代價函數(shù)處理殘差,初步篩選得到最佳候選預測模式,然后利用簡化率失真代價模型與最有可能預測模式(MPM)判斷得到RDO候選預測模式,利用RDO得到最佳預測模式。最后在HEVC測試模型HM4.0的平臺上對改進算法進行驗證,并采用不同分辨率的視頻序列進行了仿真實驗。實驗結果表明:在峰值信噪比(PSNR)影響可以忽略的情況下(平均降低0.06dB),提出的幀內預測優(yōu)化算法比HM4.0中方法的壓縮時間平均減少了30.18%,碼率平均增加了1.97%。與文獻[20]提出的幀內預測編碼方法相比,其復雜度平均減少了11.45%,碼率平均減少了0.46%,PSNR平均增加了0.01dB,壓縮性能均有所提高。
[Abstract]:An intraframe prediction optimization algorithm based on image correlation and optimal mode probability statistics is proposed to reduce the complexity of video coding. Firstly, the Angular prediction mode, Planar prediction mode, LM prediction mode and (RDO) calculation method of intra-frame rate-distortion cost optimization are introduced in the new generation video coding standard (HEVC) intra prediction algorithm. Using absolute error and (SAD) as cost function to deal with residual error, the best candidate prediction model is obtained, and then the simplified rate-distortion cost model and the most probable prediction model (MPM) are used to judge the RDO candidate prediction model. The best prediction model is obtained by using RDO. Finally, the improved algorithm is verified on the platform of HEVC test model HM4.0, and the simulation experiment is carried out with different resolution video sequences. The experimental results show that under the condition that the PSNR (PSNR) effect can be neglected (average reduction of 0.06dB), the average compression time of the proposed intra-prediction optimization algorithm is 30.18 less than that of the HM4.0 method. The average bit rate increased by 1.97. Compared with the intra prediction coding method proposed in reference [20], its complexity is reduced by 11.455.The average bit rate is reduced by 0.46 and PSNR is increased by 0.01dB, and the compression performance is improved.
【作者單位】： 北京航空航天大學儀器科學與光電工程學院測控與信息技術系;
【基金】：國家自然科學基金資助項目(No.61375025,No.61075011,No.60675018) 教育部留學回國人員科研啟動基金資助項目
【分類號】：TN919.81
[Abstract]:An intraframe prediction optimization algorithm based on image correlation and optimal mode probability statistics is proposed to reduce the complexity of video coding. Firstly, the Angular prediction mode, Planar prediction mode, LM prediction mode and (RDO) calculation method of intra-frame rate-distortion cost optimization are introduced in the new generation video coding standard (HEVC) intra prediction algorithm. Using absolute error and (SAD) as cost function to deal with residual error, the best candidate prediction model is obtained, and then the simplified rate-distortion cost model and the most probable prediction model (MPM) are used to judge the RDO candidate prediction model. The best prediction model is obtained by using RDO. Finally, the improved algorithm is verified on the platform of HEVC test model HM4.0, and the simulation experiment is carried out with different resolution video sequences. The experimental results show that under the condition that the PSNR (PSNR) effect can be neglected (average reduction of 0.06dB), the average compression time of the proposed intra-prediction optimization algorithm is 30.18 less than that of the HM4.0 method. The average bit rate increased by 1.97. Compared with the intra prediction coding method proposed in reference [20], its complexity is reduced by 11.455.The average bit rate is reduced by 0.46 and PSNR is increased by 0.01dB, and the compression performance is improved.
【作者單位】： 北京航空航天大學儀器科學與光電工程學院測控與信息技術系;
【基金】：國家自然科學基金資助項目(No.61375025,No.61075011,No.60675018) 教育部留學回國人員科研啟動基金資助項目
【分類號】：TN919.81
【參考文獻】
相關期刊論文 前3條
1 蘇睿;劉貴忠;張彤宇;劉寶蘭;;利用變換域信息快速實現(xiàn)H.264幀內預測編碼的新算法[J];電子與信息學報;2007年01期
2 蔡曉霞;崔巖松;鄧中亮;常志峰;;下一代視頻編碼標準關鍵技術[J];電視技術;2012年02期
3 金惠羨;;淺談下一代編碼壓縮技術——HEVC[J];數(shù)字通信世界;2011年11期
【共引文獻】
相關期刊論文 前10條
1 崔遙;劉軍;;HEVC關鍵技術介紹[J];科技創(chuàng)新與應用;2012年33期
2 劉毅;羅軍;黃啟俊;常勝;;HEVC整數(shù)DCT變換與量化的FPGA實現(xiàn)[J];電視技術;2013年11期
3 張玢;;高效視頻編碼標準中的關鍵技術概述[J];電腦知識與技術;2013年18期
4 原菲;司占軍;顧，
本文編號：2371092
本文鏈接：http://sikaile.net/kejilunwen/wltx/2371092.html
最近更新
教材專著
