多维矢量矩阵变换域的运动矢量估计

桑爱军; 孙敏; 陈绵书; 钟江江; 陈贺新

doi:10.3788/CO.20150802.0191

多维矢量矩阵变换域的运动矢量估计

doi: 10.3788/CO.20150802.0191

吉林大学通信工程学院，吉林长春 130022

基金项目: 吉林省自然科学基金资助项目(No.20130101045JC);吉林省科技发展计划国际科技合作资助项目(No.20130413053GH);国家自然科学基金资助项目(No.61171078)

详细信息

通讯作者:
桑爱军(1973—)，女，山东莱州人，博士，教授，硕士生导师，1994年于华中理工大学获得学士学位，1997年于大连理工大学获得硕士学位，2002年于吉林大学获得博士学位，主要从事数字图像处理等方面的研究。E-mail:sangaj@jlu.edu.cn

中图分类号: TP941.1
计量
- 文章访问数: 1450
- HTML全文浏览量: 437
- PDF下载量: 514
- 被引次数: 0
出版历程
- 收稿日期: 2014-12-11
- 录用日期: 2015-02-18
- 刊出日期: 2015-04-25

Motion vector estimation in multi-dimensional vector matrix transform domain

College of Communication Engineering,Jilin University,Changchun 130022,China

摘要

摘要: 为了实现对视频中运动目标的运动矢量估计,建立了多维矢量矩阵变换域运动估计系统,在变换域对运动视频中运动目标形成的多维能量集中平面进行研究。首先,介绍了多维矢量矩阵理论、变换理论以及运动目标在变换域形成能量集中平面的理论推导;然后,采用平面拟合的方法,求取运动矢量的大小;最后,分析对比了几种方法的效果和迭代速度。实验结果表明:多维矢量矩阵变换域的运动平面拟合方法估计的运动矢量精度达到10^-2 pixel,提供了一种变换域运动矢量估计的高精度方法。
- 视频编码 /
- 多维矢量矩阵 /
- 运动估计 /
- 迭代拟合
Abstract: To realize the motion vector estimation targeted at moving objects in video, the motion estimation system based on transformation domain of multi-dimensional vector matrix is established. The research is then carried out in transformation domain to study on multi-dimensional energy concentration plane formed by moving objects in moving videos. Firstly, the theory of multi-dimensional vector matrix and transformation are introduced. Then the theoretical derivation of energy concentration plane in transformation domain formed by moving objects is presented. The next step is to calculate the motion vector using plane fitting method. Finally,the performance and iteration speed among a few methods are analyzed. Experimental results show that the motion vector accuracy obtained by moving plane fitting in multi-dimensional vector matrix transformation domain can reach 10^-2 pixel. A new high accuracy motion vector estimation method in transformation domain is provided.
- video coding /
- multi-dimensional vector matrix /
- motion estimation /
- iterative fitting

HTML全文

图 1 运动估计的框图

Figure 1. Block diagram of MVM-DCT

下载: 全尺寸图片幻灯片

图 2 均匀平移图像序列的前8帧

Figure 2. First 8 frames of image sequence formed by uniform translation

下载: 全尺寸图片幻灯片

图 3 变换后系数

Figure 3. Coefficients after transformation

下载: 全尺寸图片幻灯片

图 4 迭代拟合的流程图

Figure 4. Iterative fitting flowchart

下载: 全尺寸图片幻灯片

图 5 两种运动矢量的折叠面

Figure 5. Folding surface of two kinds of motion vector

下载: 全尺寸图片幻灯片

图 6 平面拟合效果图

Figure 6. Renderings of plane fitting

下载: 全尺寸图片幻灯片

图 7 剔除数据后平面拟合效果图

Figure 7. Renderings of plane fitting after excluding data

下载: 全尺寸图片幻灯片

图 8 进一步剔除数据后平面拟合效果图

Figure 8. Renderings of plane fitting after further excluding data

下载: 全尺寸图片幻灯片

图 9 迭代拟合试验结果

Figure 9. Iterative fitting test results

下载: 全尺寸图片幻灯片

表 1 迭代拟合试验结果

Table 1. Iterative fitting test results

精度(pixel) 1 2 3 4 9

运动矢量(2,0) 0.17 0.09 0.08 0.07 0.03

运动矢量(3,0) 0.10 0.04 0.03 0.02 0.02

H.264国际标准 0.25

下载: 导出CSV

参考文献(14)

[1]	[1] 朱明,杨航,贺柏根,等.联合梯度预测与导引滤波的图像运动模糊复原[J].中国光学,2013,12(6):850-854. ZHU M,YANG H,HE B G,et al.. Image motion blurring restoration of joint gradient prediction and guided filter[J]. Chinese Optics,2013,12(6):850-854.(in Chinese)
[2]	[2] 许廷发,李俊涛,张一舟,等.真彩色传递双波段图像融合[J].中国光学,2014,6(7):402-810. XU T F,LI J T,ZHANG Y ZH,et al.. True color transfer for dual band image fusion[J]. Chinese Optics, 2014,6(7):402-810.(in Chinese)
[3]	[3] 郭立强,朱明.彩色图像处理的可交换Clifford代数方法[J].中国光学,2013,12(6):884-891. GUO L Q,ZHU M. Commutative Clifford algebra method for color image processing[J]. Chinese Optics, 2013,12(6):884-891.(in Chinese)
[4]	[4] 毕厚杰.新一代视频压缩编码标准--H.264/AVC[M].北京:人民邮电出版社,2005. BI H J. A New Generation of Video Coding Standard--H.264/AVC[M]. Beijing:People Post Press.2005.(in Chinese)
[5]	[5] BENJAMIN B,WOO-JIN H,JENS-RAINER O,et al.. High Efficiency Video Coding(HEVC) text specification draft 10[C], JCT-VC-L1003 of JCT-VC Geneva,CH,Jan.,2013.
[6]	[6] 沈燕飞.高效视频编码[J].计算机学报,2013,11(36):2340-2355. HEN Y F. High efficiency video coding[J]. Chinese J. Computers, 2013,11(36):2340-2355.(in Chinese)
[7]	[7] 王大伟,严俊,王彬,等.视频序列的运动估计技术综述[J].电视技术, 2012(3):20-23. WANG D W,YAN J,WANG B,et al.. Survey of motion estimation used in video sequence[J]. ,2012(3):20-23.(in Chinese)
[8]	[8] 刘龙.基于运动矢量时-空特性的快速运动估计算法研究[J].通信学报,2013,34(1):121-127. LIU L. Fast motion estimation based on the special and temporal characteristic[J]. J. Communications,2013,34(1):121-127.(in Chinese)
[9]	[9] 刘英哲.H.264中一种基于搜索范围自适应调整运动估计算法[J].电子与信息学报,2013,35(6):1382-1387. LIU Y ZH. Motion estimation algorithm based adaptive search range adjustment for H.264[J]. J. Electronics & Information Technology,2013,35(6):1382-1387.(in Chinese)
[10]	[10] 桑爱军,王艇,栾晓利,等.2M维矢量余弦整数变换核矩阵[J].光学精密工程,2013,7(21):1891-1897. SANG A J,WANG T,LUAN X L,et al..[J]. Opt. Precision Eng.,2013,7(21):1891-1897.(in Chinese)
[11]	[11] SANG A J,CHEN M S,CHEN H X,et al.. Multi-dimensional vector matrix theory and its application in color image coding[J]. Imaging Science J.,2010,58(3):171-176.
[12]	[12] 桑爱军,崔海廷,王墨林,等.高维变换域中的系数集中分析及研究[J].吉林大学学报(工学版),2013,43(增刊):96-100. SANG A J,CUI H T,WANG M L,et al.. Coefficient concentration analysis and research in high-dimensional transfortmation domain[J]. J. Jilin University(Engineering and Technology Edition),2013,43(Supp.):96-100.(in Chinese)
[13]	[13] 桑爱军,吴杨,崔海廷,等.基于多维矢量矩阵的离散余弦变换快速算法[J].光学精密工程,2013(21):799-806. SANG A J,WU Y,CUI H T,et al.. Fast DCT algorithms based on multi-dimensional vector matrix[J]. Opt. Precision Eng.,2013,3(21):799-806.(in Chinese)
[14]	[14] NIKOLA BOZINOVIC,JANUSZ KONRAD. Motion analysis in 3D DCT domain and its application to video coding, signal processing[J]. Image Communication,2005,6(20):510-528.