首页 > 论文 > 激光与光电子学进展 > 56卷 > 23期(pp:231004--1)

利用Tikhonov正则化改进移动最小二乘的图像变形算法

Moving Least Squares Based Image Deformation Algorithm Improved with Tikhonov Regularization

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

针对移动最小二乘算法在图像变形过程中,求解的线性方程组系数矩阵会出现不可逆、求解不稳定的问题,通过引入Tikhonov正则化,运用L-曲线法求解正则参数,对系数矩阵施加约束项从而得到精确解,避免病态方程组的形成;针对在实现图像变形过程中,定位特征点人工量大且特征点不足的问题,运用Dlib库自动提取68个覆盖人脸五官和轮廓的特征点。仿真实验结果表明,与原算法相比,提出的改进算法可以使图像产生清晰、准确的变形。

Abstract

In the image deformation process of the moving least squares based algorithm, the coefficient matrix of the solved linear equations is irreversible and unstable. In this study, we apply a constraint term to the coefficient matrix to obtain the exact solution and avoid the formation of ill-conditioned equations by introducing Tikhonov regularization and using the L-curve method to solve the regular parameters. To overcome the limitation of a large number of manually localized feature points and an insufficient number of feature points in the process of image deformation, the Dlib library is employed to automatically extract 68 feature points covering facial features and contours. Simulation results demonstrate that, compared to the original algorithm, the proposed algorithm can produce clear and accurate image deformation.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TN919.85

DOI:10.3788/LOP56.231004

所属栏目:图像处理

基金项目:国家自然科学基金、中央高校基本科研业务费专项资金;

收稿日期:2019-04-23

修改稿日期:2019-05-27

网络出版日期:2019-12-01

作者单位    点击查看

崔小曼:江南大学物联网工程学院, 江苏 无锡 214122
于凤芹:江南大学物联网工程学院, 江苏 无锡 214122

联系人作者:于凤芹(13961720781@163.com)

备注:国家自然科学基金、中央高校基本科研业务费专项资金;

【1】Yang J Z, Liu Z K, Yu N H, et al. An image warping method based on control points and its applications [J]. Journal of Image and Graphics. 2001, 6(11): 1070-1074.
杨金钟, 刘政凯, 俞能海, 等. 基于控制点的图象变形方法及其应用 [J]. 中国图象图形学报. 2001, 6(11): 1070-1074.

【2】Wang S J, Liang X Y. Calculation method of image morphing and its applications [J]. Journal of Computer-Aided Design & Computer Graphics. 2011, 23(8): 1304-1310.
王守觉, 梁先扬. 图像变形计算方法及其应用 [J]. 计算机辅助设计与图形学学报. 2011, 23(8): 1304-1310.

【3】Wolberg G. Image morphing: a survey [J]. The Visual Computer. 1998, 14(8/9): 360-372.

【4】Liu T. Study on image deformation based on moving least squares [D]. Dalian: Dalian University of Technology. 2008.
刘婷. 移动最小二乘图像变形方法研究 [D]. 大连: 大连理工大学. 2008.

【5】Zhu X C. Face feature extraction and hairstyle classification based on image Zhejiang: [D]. Zhejiang Sci-Tech University. 2017.
朱希成. 基于图像的人脸特征提取与发型分类 [D]. 浙江: 浙江理工大学. 2017.

【6】Luo D, Tao Y, Yang G. Face texture mapping and deformation with constrains of facial feature [J]. Computer Engineering and Applications. 2018, 54(6): 188-192, 240.
罗岱, 陶洋, 杨刚. 基于面部特征约束的人脸纹理映射及变形 [J]. 计算机工程与应用. 2018, 54(6): 188-192, 240.

【7】Schaefer S. McPhail T, Warren J. Image deformation using moving least squares [J]. ACM Transactions on Graphics. 2006, 25(3): 533-540.

【8】Zuo C W, Nie Y F, Zhao M L. The selection about the radius of influence in MLS [J]. Chinese Journal of Engineering Mathematics. 2005, 22(5): 833-838.
左传伟, 聂玉峰, 赵美玲. 移动最小二乘方法中影响半径的选取 [J]. 工程数学学报. 2005, 22(5): 833-838.

【9】Zhou C H, Wang Z L, Liu S K. Method of image restoration directly based on spatial varied point spread function [J]. Acta Optica Sinica. 2017, 37(1): 0110001.
周程灏, 王治乐, 刘尚阔. 基于空间变化点扩展函数的图像直接复原方法 [J]. 光学学报. 2017, 37(1): 0110001.

【10】Yang J J, Zheng J L. Stability of moving least squares approximation [J]. Acta Mathematicae Applicatae Sinica. 2012, 35(4): 637-648.
杨建军, 郑健龙. 移动最小二乘法的近似稳定性 [J]. 应用数学学报. 2012, 35(4): 637-648.

【11】Li J, Gao F, Yi X, et al. A methodological and experimental investigation on time-domain diffuse fluorescence tomography of analytic based on two-dimensional circular scheme [J]. Chinese Journal of Lasers. 2010, 37(11): 2743-2748.
李娇, 高峰, 易茜, 等. 基于二维圆域解析模型的时域扩散荧光层析原理与实验研究 [J]. 中国激光. 2010, 37(11): 2743-2748.

【12】Tikhonov A N. Solution of incorrectly formulated problems and the regularization method [J]. Soviet Mathematics Doklady. 1963, 4: 1035-1038.

【13】Tikhonov A N. Regulation of ill-posed problems [J]. Doklady Akademii Nauk Sssr. 1963, 151(1): 49-52.

【14】Zheng C L, He D D, Fei Q G. Improved digital image correlation method based on gray gradient denoised by regularization method [J]. Acta Optica Sinica. 2018, 38(8): 0812002.
郑成林, 何顶顶, 费庆国. 基于灰度梯度正则化去噪的改进数字图像相关法 [J]. 光学学报. 2018, 38(8): 0812002.

【15】Wang C, Zhang B, Cao L X, et al. An improved inversion algorithm to measure particle size distribution [J]. Acta Optica Sinica. 2019, 39(2): 0212009.
王晨, 张彪, 曹丽霞, 等. 颗粒粒径分布测量反演算法的改进 [J]. 光学学报. 2019, 39(2): 0212009.

【16】Ma C Y, Yang S L, Li S P. A regularization method of ill-conditioned linear equations [J]. Journal of Gansu Sciences. 2010, 22(4): 33-35.
马成业, 杨胜良, 黎锁平. 求解病态线性方程组的一个正则化方法 [J]. 甘肃科学学报. 2010, 22(4): 33-35.

【17】International organisation for standardisation. Organisation internationale de normalisation: ISO/IEC JTC1/SC29/WG11 [S]. Japan: JISC. 1998.

【18】Schaefer S. McPhail T, Warren J. Image deformation using moving least squares [J]. ACM Transactions on Graphics. 2006, 25(3): 533-540.

【19】Yuan Z C, Shen Y Z, Zhou Z B. Regularized total least-squares solution to ill-posed error-in-variable model [J]. Journal of Geodesy and Geodynamics. 2009, 29(2): 131-134.
袁振超, 沈云中, 周泽波. 病态总体最小二乘模型的正则化算法 [J]. 大地测量与地球动力学. 2009, 29(2): 131-134.

【20】Hu B, Xu H L, Wang Z W, et al. The method for choosing regularization parameters based on a model function and the L-curve [J]. Journal of Jiangxi Normal University(Natural Science Edition). 2014, 38(6): 569-573.
胡彬, 徐会林, 王泽文, 等. 基于模型函数与L-曲线的正则化参数选取方法 [J]. 江西师范大学学报(自然科学版). 2014, 38(6): 569-573.

【21】Chen S, Zhan B Y. Application of L-curve method on robust ridge estimation model [J]. GNSS World of China. 2014, 39(5): 41-45.
陈帅, 詹本勇. 基于L曲线法的抗差岭估计模型 [J]. 全球定位系统. 2014, 39(5): 41-45.

【22】Zhang L B, Huang Y, Zhu T. Region-of-interest denoising of high spatial resolution remote sensing image based on generalized cross validation [J]. Acta Optica Sinica. 2010, 30(s1): s100410.
张立保, 黄颖, 朱童. 基于广义交叉验证的高分辨率遥感图像感兴趣区去噪 [J]. 光学学报. 2010, 30(s1): s100410.

引用该论文

Cui Xiaoman,Yu Fengqin. Moving Least Squares Based Image Deformation Algorithm Improved with Tikhonov Regularization[J]. Laser & Optoelectronics Progress, 2019, 56(23): 231004

崔小曼,于凤芹. 利用Tikhonov正则化改进移动最小二乘的图像变形算法[J]. 激光与光电子学进展, 2019, 56(23): 231004

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF