查詢結果分析
相關文獻
頁籤選單縮合
題名 | A Fast Motion Estimation Vector Algorithm Using Successive Refinement by Subsampling=利用減取樣逐步篩選的快速移動向量估計法則 |
---|---|
作者姓名(中文) | 蔡明傑; 黃光輝; | 書刊名 | 大同學報 |
卷期 | 28 1998.11[民87.11] |
頁次 | 頁187-195+458 |
分類號 | 312.1 |
關鍵詞 | 減取樣; 快速移動向量估計法則; 視訊壓縮; 快速區塊比對法則; |
語文 | 英文(English) |
中文摘要 | 隨著多媒體系統的風行,視訊處理之重要性越來越高。但因為數位視訊訊號資料量相當龐大,使得資料在儲存及傳輸感到困難。因此,採用視訊壓縮已是不得不然的趨勢。所以利用視訊壓縮來減少資料量,不但可減損儲存空間,亦可以減少資料的傳輸時間。 在本論文中,我們提出一種新的快速區塊比對法則,它根據將移動向量的候選集合逐步篩選的原理,可避免局部最佳化 (local-minimum) 的問題。在與其它的搜尋法則比較上,根據H.261規格的模擬結果,所提出的搜尋法則之效果較其它的快速搜尋法來得好,甚至也比全面搜尋法則 (Full Search) 來得好。 本法則採用了一種階層式架構,在第一、二層中,利用減取樣 (Subsampling) 來獲得子區塊 (Subblock) 。在第一層中,利用子區塊用少量的計算時間來做全面搜尋來獲得移動向量的候選集合。在第二層中,我們將篩選這些候選的移動向量,來獲得更精煉的移動向量候選集合。在最後一層中,真正的移動向量將會從少數的候選向量中選出。因為在第一層中,我們利用較少像素 (pixel) 的子區塊來做全面搜尋,不但可避免其它快速搜尋法則最易犯的局部最佳化問題並可大量減少計算時間。 |
英文摘要 | The video image processing has become a key technology as the multimedia industry is getting more and more important. Because of the large amount of video data, the storage space is required to be large and transmission time becomes long consequently. Therefore, the utilization of video compression is inevitable. By means of video compression, both the storage space and transmission time of video can be saved. In this paper, a new fast block matching algorithm (BMA)[1][2] is proposed to alleviate the local minimum problem based on successive refinement of the motion vector candidates. From H.261 simulation, the performance of the proposed algorithm can compete with the best algorithms of the previous studies, even the full search (FS) algorithm. The proposed algorithm employs a hierarchical structure. At the first two layers, we obtain the subblocks by subsampling. At the former layer, a full search is performed with the subblocks to obtain a candidate set for the motion vector within a short computation time. At the latter, we refine the candidate set for the motion vector, And at the next layer, only a single motion vector is selected from a few candidates. Since a full search is performed with a subblock which has less pixel at the first layer, the proposed algorithm cam improve the local minimum problem in the existing fast search algorithms and also reduce the computation time. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。