An Adaptive Fast Intra Mode Decision Algorithm Based on Gradient in HEVC

[[abstract]]為了因應市場需求逐漸提高的視訊品質，JCT-VC推出了劃時代的視訊壓縮編碼標準－ H 265/HEVC，在同樣的畫面品質下，H 265/HEVC比上一代H 264/AVC多了一倍的壓縮率，但顯著的缺點是比H 264/AVC冗長許多的編碼時間；在現代，H 265/HEVC被愈來愈頻繁地使用於消費性產品，所以如何更快速地編解碼來提升效率是非常必要的。 本論文提出一個可應用於畫面內預測的視訊編碼架構，利用H 265/HEVC編碼端預測單元的梯度變化量，對複雜且耗時的最佳的畫面內預測模式選擇過程提出快速決策法來減少冗餘的候選模式，以達到有效地減少編碼時間。依據實驗結果顯示，我們所提出的方法比HEVC測試軟體HM10 0節省的編碼時間可達到約21%，且可以保持相近的壓縮率與重建視訊品質。[[abstract]]In order to meet the market demand for increasing quality of the video JCT-VC launched a landmark video coding standard- the H 265/HEVC The compression ratio of video encoded using the H 265/HEVC is about twice of that using the previous video compression standard the H 264/AVC under the same quality of reconstructed video But the computational complexity of the H 265/HEVC encoder is very high Hence it is very important to improve the efficiency of the encoder of the H 265/HEVC to be extended to the real-time application of video coding This thesis proposes a strategy to improve the efficiency of intra-frame prediction mode decision in the H 265/HEVC standard so that the encoding time is saved The proposed method utilizes the relationship between the angular modes of the intra-frame prediction and the gradient variation of the prediction Unit to reduce the number of the RMD candidates in the intra-frame prediction mode decision Experimental results show that the proposed algorism can reduce the encoding time up to about 21% compared to the HEVC testing model HM10 0 with few degradation of the performance of the bit rate and the quality of reconstructed vide