一種適用于遮擋問題的運(yùn)動(dòng)補(bǔ)償幀插值方法
Motion Compensated Frame Interpolation Method Applied for Occlusion Issue
-
摘要: 針對(duì)二維運(yùn)動(dòng)估計(jì)中常出現(xiàn)的遮擋問題，提出一種基于網(wǎng)格模型的運(yùn)動(dòng)補(bǔ)償自適應(yīng)幀插值技術(shù)，將圖像幀中分為3種不同區(qū)域，根據(jù)各自的特點(diǎn)分別進(jìn)行運(yùn)動(dòng)估計(jì)和運(yùn)動(dòng)補(bǔ)償插值。在準(zhǔn)確檢測(cè)遮擋區(qū)域前提下，單獨(dú)對(duì)遮擋區(qū)域處理有效地減小了運(yùn)動(dòng)估計(jì)誤差。提出采用特征窗口運(yùn)動(dòng)補(bǔ)償匹配的方法進(jìn)行網(wǎng)格節(jié)點(diǎn)運(yùn)動(dòng)矢量估計(jì)，消除了傳統(tǒng)塊匹配方法導(dǎo)致的塊效應(yīng)，得到了亞像素的匹配精度。實(shí)驗(yàn)證明，該方法簡單易行，有效地解決了遮擋區(qū)域的運(yùn)動(dòng)估計(jì)問題，插值圖像具有較好的視覺效果。Abstract: A motion compensated adaptive frame interpolation technique based on mesh model is proposed considering occlusion issue existed in two-dimensional motion estimation. The whole image is divided into three types of regions in order to perform motion estimation and motion compensated frame interpolation respectively. A more accurate occlusion detec-tion method is presented to improve the precision of detection, which is the main premise of the adaptive interpolation. The motion estimation error introduced by occlusion regions is reduced effectively. Mesh node motion estimation based on feature window matching is used to achieve sub-pixel resolution and overcome block artifacts produced by block matching. Experimental results indicate that the proposed algorithm is easy to implement, and solves motion estimation problem existed in occlusion regions effectively. The interpolated image shows better visual performance.
-
Thoma R, Bierling M. Motion compensating interpolation considering covered and uncovered background[J].Signal Processing: Image Communication.1989, 1(1):191-212[2]Sezan M.[J].Lagendijk R. Frame rate conversion of motion picture films for progressive scan HDTV.Signal Processing VI, Proc. EUSIPCO92, Brussels, Belgium.1992,:-[3]Sullivan G J, Baker R L. Motion compensation for video compression using control grid interpolation, in Proc. ICASSP, Toronto, Ont., Canada, 1991: 2713-2716.[4]Wang Y, Lee O. Use of two-dimensional deformable mesh structures for video coding, part I-the synthesis problem: Mesh-based function approximation and mapping[J].IEEE Trans. on Circuits Syst. Video Technol.1996, 6(6):636-646[5]Wang Y, Lee O, Vetro A. Use of two-dimensional deformable mesh structures for video coding,part Ⅱ-the analysis problem and a region-based coder employing an active mesh representation[J].IEEE Trans. on Circuits Syst. Video Technol.1996, 6(6):647-659[6]Altunbasak Y, Tekalp A. Occlusion-adaptive content-based 2-D mesh design and tracking for object-based coding. IEEE Trans. on Image Processing, 1997, IP-6(9): 1270-1280.[7]吳楓,高鵬,高文. 基于網(wǎng)格模型的運(yùn)動(dòng)估計(jì)技術(shù). 電子學(xué)服, 2000(5): 47-51[8]Horn B K P, Schunck B G. Determining optical flow[J].Artif. Intell.1981, 17(2):185-203[9]Shi J, Tomasi C. Good features to track. TR93-1399, Cornell University, 1993. -
計(jì)量
- 文章訪問數(shù): 2537
- HTML全文瀏覽量: 109
- PDF下載量: 1162
- 被引次數(shù): 0
下載:
