Method of vertical parallax reduction combined with Levenberg-Marquardt algorithm
-
摘要: 垂直视差的存在是影响立体视频观视舒适度的主要因素。为了在不影响水平视差的条件下实现对垂直视差的消减,本文引入Levenberg-Marquardt(L-M)非线性算法实现变换矩阵的精确求解。首先用抗缩放、旋转及仿射变换的SIFT(Scale-invariant feature transform)特征匹配算法检测出双目图像对的特征匹配点,然后根据匹配点的坐标位置运用L-M算法计算可消减垂直视差的变换矩阵,将变换矩阵作用于目标图像,计算出该视图每个像素点的新坐标位置。实验结果表明:与利用线性算法求解二维射影变换矩阵的垂直视差消减方法相比,本文提出的求解方法在垂直视差消减上比该算法提高了约0.029 1~0.323 2个像素,对水平视差的影响比该算法降低了约0.118 7~1.139 1个像素。因此本文提出的方法对垂直视差的消减起到了优化作用。
-
关键词:
- SIFT /
- 射影变换 /
- 垂直视差消减 /
- Levenberg-Marquardt算法
Abstract: The existence of vertical parallax is the main factor of affecting the viewing comfort of stereo video. In order to reduce the vertical parallax without affecting the horizontal parallax, Levenberg-Marquardt(L-M) algorithm which is the nonlinear algorithm, is introduced in this paper to achieve the accuracy of the transformation matrix. Firstly, the SIFT algorithm, which is invariant to scaling, rotation and affine transformation, is used to detect the feature matching points from the binocular images. Then according to the coordinate position of matching points, the transformation matrix, which can reduce the vertical parallax, is calculated using Levenberg-Marquardt algorithm. Finally, the transformation matrix is applied to target image to calculate the new coordinate position of each pixel from the view images. The experimental results show that compared with the method that can reduce the vertical parallax using linear algorithm to calculate two-dimensional projective transformation, the proposed method using nonlinear algorithm improves the vertical parallax reduction from about 0.029 1 to 0.323 2 pixel and the effect of horizontal parallax is reduced from about 0.118 7 to 1.139 1 pixel. Therefore, the proposed method can optimize the vertical parallax reduction. -
图 3 breakdancer第7帧的3、4视点实验效果图
Figure 3. Experimental results of the 3 and 4 view points of the 7th frame of breakdancer image
图 4 ballet第1帧的1、2视点实验效果图
Figure 4. Experimental results of the 1 and 2 view points of the 1th frame of ballet image
表 1 不同参数下breakdancer垂直视差消减前后性能评价(单位：像素)
Table 1. Experimental results of breakdancer image under different parameters(Unit:pixel)
下载: 导出CSV
表 2 不同参数下ballet垂直视差消减前后性能评价(单位：像素)
Table 2. Experimental results of ballet image under different parameters(Unit:pixel)
下载: 导出CSV
-
[1] 杜歆,邹泷.视点合成中基于深度的空洞修复[J].浙江大学学报:工学版,2015,49(9):1616-1624.DU X,ZOU L. Depth-based hole filling in view synthesis[J]. J. Zhejiang University(Engineering Science),2015,49(9):1616-1624.(in Chinese) [2] 王晓燕,王世刚,姜秀红,等.亮度优化立体视频视觉舒适度评价[J].中国光学,2015,8(3):394-400.WANG X Y,WANG SH G,JIANG X H,et al.. Evaluation of stereo video visual comfort based on luminance optimization[J]. Chinese Optics,2015,8(3):394-400.(in Chinese) [3] 李实秋,雷建军,周志远,等.基于SIFT匹配的多视点立体图像零视差调整[J].红外与激光工程,2015(2):764-768.LI SH Q,LEI J J,ZHOU ZH Y,et al.. Zero-disparity adjustment of multiview stereoscopic images based on SIFT matching[J]. Infrared and Laser Engineering,2015(2):764-768.(in Chinese) [4] 邵枫,姜求平,蒋刚毅,等.基于显著性分析的立体图像视觉舒适度预测[J].光学 精密工程,2014,22(6):1631-1638.SHAO F,JIANG Q P,JIANG G Y,et al.. Prediction of visual discomfort of stereoscopic images based on saliency analysis[J]. Opt. Precision Eng.,2014,22(6):1631-1638.(in Chinese) [5] LOOP C,ZHANG Z. Computing rectifying homographies for stereo vision[C]. IEEE Computer Society Conference on Computer Vision and Pattern Recognition,Fort Collins,Colorado,USA,1999:125-131. [6] 刘利亮,安平,王贺,等.基于公共点提取的多视图像校正[J].信号处理,2011,6(27):857-863.LIU L L,AN P,WANG H,et al.. Multi-view images rectification based on common points extraction[J]. Signal Processing,2011,6(27):857-863.(in Chinese) [7] ZHOU B L,ZHAO Y,WANG J Y,et al.. Improvement of visual fatigue based on vertical parallax elimination[C]. 2013 9th International Conference on Natural Computation,Shengyang,China,2013:1350-1354. [8] DAVID G L. Distinctive image features from scale-invariant keypoints[J]. International J. Computer Vision,2004(60):91-110. [9] 王灿进,孙涛,陈娟.局部不变特征匹配的并行加速技术研究[J].液晶与显示,2014,29(2):266-274.WANG C J,SUN T,CHEN J. Speeding up local invariant feature matching using parallel technology[J]. Chinese J. Liquid Crystals and Displays,2014,29(2):266-274.(in Chinese) [10] HARTLEY R. Theory and practice of projective rectification[J]. International J. Computer Vision,1999(35):115-127. [11] 程明明,王贺,安平,等.基于特征点匹配的多视图像校正[J].液晶与显示,2010,8(24):593-597.CHENG M M,WANG H,AN P,et al.. Multi-view images rectification based on feature points matching[J]. Chinese J. Liquid Crystals and Displays,2010.8(24):593-597.(in Chinese) [12] 姚佳宝,田秋红,陈本永.一种基于L-M算法的RANSAC图像拼接算法[J].浙江理工大学学报:自然科学版,2015,33(4):552-557.YAO J B,TIAN Q H,CHEN B Y. RANSAC image stitching algorithm based on L-M algorithm[J]. J. Zhejiang Sci-Tech University(Natural Sciences),2015,33(4):552-557.(in Chinese) [13] 项文强,张华,王姮,等.基于L-M算法的BP网络在变压器故障诊断中的应用[J].电力系统保护与控制,2011,39(8):100-103.XIANG W Q,ZHANG H,WANG H,et al..Application of BP neural network with L-M algorithm in power transformaer fault diagnosis[J]. Power System Protection and Control,2011,39(8):100-103.(in Chinese) [14] SHAO X T,SHAN L,ZONG M L. Levenberg-Marquardt algorithm based nonlinear optimization of camera calibration for relative measurement[C]. 2015 34th Chinese Control Conference(CCC),IEEE,Hangzhou,China,2015:4868-4872. [15] 邹泷.3D视频中的虚拟视点合成方法研究[D]. 浙江大学,2015.ZOU L. Research on virtual view synthesis for 3D video[D]. Zhejiang University,2005.(in Chinese) [16] 高强,金杰,沈丽丽,等.使用EEG分析水平视差对3D图像舒适性的影响[J].红外与激光工程,2014(12):4140-4145.GAO Q,JIN J,SHEN L L,et al.. Effect analysis of horizontal parallax on 3D images comfortableness using EEG[J]. Infrared and Laser Engineering,2014(12):4140-4145.(in Chinese) -
下载:
图(4) / 表(2)
计量
- 文章访问数: 1738
- HTML全文浏览量: 375
- PDF下载量: 822
- 被引次数: 0
