二维动画人物的特征提取与三维重建
|
摘要
为了解决快速提取二维动画人物特征,完成立体动画人物形象构建的问题,提出了二维动画人物的特征提取与三维重建方法.通过分析二维动画人物特征提取,具体分析了二维动画人物Harris算子的确定、二维动画人物对极几何关系的确定、基于Harris算子和对极几何角度的特征提纯;分析了基于二维动画人物特征的三维重建方法,主要包括基于二维动画人物特征的三维重建流程、三维重建方法的哈希排序和利用CUDA框架实现三维重建方法.实验可知,所提方法的二维动画人物提取速度最大值为0.401mm/s,而传统二维动画人物提取速度最大值为0.589mm/s.动画人物形象三维构建准确性的对比中,在特殊检测方法中,本文方法对称性更好,准确性也更高.应用对比实验的方法证明,应用新型重建方法后,二维动画人物特征提取速度明显加快,动画人物形象三维构建准确性也大大提升.
To extract the feature of 2D animation characters quickly and complete the image construction of three-dimensional animation characters,a method of feature extraction and 3D reconstruction of 2D animation characters is proposed.By analyzing the feature extraction of 2D animation characters,the determination of Harris operator of 2D animation characters,the determination of the dipole geometry relation of 2D animated characters,and the character purification based on Harris operator and dipole geometry angle are analyzed in detail.Three-dimensional reconstruction method based on the features of 2D animated characters is analyzed,which mainly includes the 3D reconstruction process, Hash ordering of 3D reconstruction methods,and the implementation of 3D reconstruction method using CUDA framework.The experiment shows that the maximum speed of 2D animation character extraction is 0.401 mm/s.The maximum speed of traditional 2D animation character extraction is 0.296 mm/s.In the comparison of 3D construction accuracy of animation characters,in the special detection method,the symmetry of this method is better.The method of contrast experiment proves that after the new reconstruction method,the speed of feature extraction of 2D animation character is obviously accelerated,and the 3D structure of animation character image is obviously accelerated.The accuracy of construction is also greatly improved.
To extract the feature of 2D animation characters quickly and complete the image construction of three-dimensional animation characters,a method of feature extraction and 3D reconstruction of 2D animation characters is proposed.By analyzing the feature extraction of 2D animation characters,the determination of Harris operator of 2D animation characters,the determination of the dipole geometry relation of 2D animated characters,and the character purification based on Harris operator and dipole geometry angle are analyzed in detail.Three-dimensional reconstruction method based on the features of 2D animated characters is analyzed,which mainly includes the 3D reconstruction process, Hash ordering of 3D reconstruction methods,and the implementation of 3D reconstruction method using CUDA framework.The experiment shows that the maximum speed of 2D animation character extraction is 0.401 mm/s.The maximum speed of traditional 2D animation character extraction is 0.296 mm/s.In the comparison of 3D construction accuracy of animation characters,in the special detection method,the symmetry of this method is better.The method of contrast experiment proves that after the new reconstruction method,the speed of feature extraction of 2D animation character is obviously accelerated,and the 3D structure of animation character image is obviously accelerated.The accuracy of construction is also greatly improved.
引文
[1]荆茜.什么是有感觉的设计:图形心理在二维动画角色造型设计中的应用研究[J].当代青年月刊,2015(7):59.JING Q.What is sensory design:research on the application of graphics psychology in the character modeling design of two-dimensional animation animation[J].Journal of Contemporary Youth,2015(7):59.
[2]田瀚,马杰.从圣衣设计变化看经典二维动画造型在三维动画中的延续和发展:以《圣斗士星矢:圣域传说》为例[J].艺术与设计:理论,2016(12):94-96.TIAN H,MA J.On the continuation and development of classical two-dimensional animation modeling in 3D animation from the change of the design of holy clothes:taking Saint Seiya:the Legend of the Holy Land as an example[J].Art and Design:Theory,2016(12):94-96.
[3]高伟,郭瑾,郭舒婷.中国传统皮影艺术在数字化动画中的应用与发展[J].辽宁师范大学学报(自然科学版),2015(4):475-481.GAO W,GUO J,GUO S T.The application and development of the traditional China Shadow Art in digital animation[J].Journal of Liaoning Normal University(Natural Science Edition),2015(4):475-481.
[4]王春莉,梁立波,王宝玉.计算机三维重建技术发展与应用[J].沈阳大学学报,2003,15(2):25-26.WANG C L,LIANG L B,WANG B Y.Development and application of computer 3Dreconstruction[J].Journal of Shenyang University,2003,15(2):25-26.
[5]全旭松,刘长春,陈海平,等.面向高功率激光装置精密装校实验室的三维重建技术及应用[J].计算机辅助设计与图形学学报,2017,29(5):854-859.QUAN X S,LIU C C,CHEN H P,et al.3Dreconstructing technique and application of high-power laser facility's optics assembly building[J].Journal of Computer-Aided Design&Computer Graphics,2017,29(5):854-859.
[6]牛晰,袁晓东.基于物体特征有效提取和离散点三维重建的3D扫描系统原型研究[J].计算机测量与控制,2017(5):158-161.NIU X,YUAN X D.Research of a prototype 3Dscanning system based on object features active extraction and 3D reconstruction with discrete points[J].Computer Measurement&Control,2017(5):158-161.
[7]韩硕,童晓冲,卢学良,等.手持线激光扫描视频三维重建中的运动线提取算法[J].计算机应用与软件,2016,33(7):154-158.HAN S,TONG X C,LU X L,et al.Moving line extraction algorithm in 3D reconstruction of handheld line laser scanning video[J].Computer Applications and Software,2016,33(7):154-158.
[8]白洪伟,李进,穆星,等.近景摄影测量在地下三维重建中的应用[J].沈阳大学学报(自然科学版),2016,28(3):223-226.BAI H W,LI J,MU X,et al.Application of close range photogrammetry in underground 3D reconstruction[J].Journal of Shenyang University(Natural Science),2016,28(3):223-226.
[9]赵广军,王旭初,牛彦敏,等.基于SAE深度特征学习的数字人脑切片图像分割[J].计算机辅助设计与图形学学报,2016,28(8):1297-1305.ZHAO G J,WANG X C,NIU Y M,et al.Digital human brain slice image segmentation based on SAE depth feature learning[J].Journal of Computer-Aided Design and Graphics,2016,28(8):1297-1305.
[10]刘丹,诸叶平,李书钦,等.基于实测数据的小麦叶片模拟模型与三维重建:以偃展4110为例[J].山东农业科学,2016,48(7):137-141.LIU D,ZHU Y P,LI S Q,et al.Simulation model and 3D reconstruction of wheat leaves based on measured data:a case study of Yanzhan 4110[J].Shandong Agricultural Sciences,2016,48(7):137-141.
[11]孟偲,曲圣悦,王利军,等.用于C臂快速标定的平面靶标可见标记点的提取与配准[J].高技术通讯,2015,25(4):397-403.MENG S,QU S Y,WANG L J,et al.Extraction and match of visible markers on a planar phantom for fast Carm calibration[J].Chinese High Technology Letters,2015,25(4):397-403.
[12]江静,蔡国榕,李绍滋,等.基于多结构估计的建筑物激光雷达点云特征线提取算法[J].厦门大学学报(自然版),2015,54(3):390-396.JIANG J,CAI G R,LI S Z,et al.Extract feature lines from building LiDAR point cloud based on multi-structure estimators[J].Journal of Xiamen University(Natural Science),2015,54(3):390-396.
[2]田瀚,马杰.从圣衣设计变化看经典二维动画造型在三维动画中的延续和发展:以《圣斗士星矢:圣域传说》为例[J].艺术与设计:理论,2016(12):94-96.TIAN H,MA J.On the continuation and development of classical two-dimensional animation modeling in 3D animation from the change of the design of holy clothes:taking Saint Seiya:the Legend of the Holy Land as an example[J].Art and Design:Theory,2016(12):94-96.
[3]高伟,郭瑾,郭舒婷.中国传统皮影艺术在数字化动画中的应用与发展[J].辽宁师范大学学报(自然科学版),2015(4):475-481.GAO W,GUO J,GUO S T.The application and development of the traditional China Shadow Art in digital animation[J].Journal of Liaoning Normal University(Natural Science Edition),2015(4):475-481.
[4]王春莉,梁立波,王宝玉.计算机三维重建技术发展与应用[J].沈阳大学学报,2003,15(2):25-26.WANG C L,LIANG L B,WANG B Y.Development and application of computer 3Dreconstruction[J].Journal of Shenyang University,2003,15(2):25-26.
[5]全旭松,刘长春,陈海平,等.面向高功率激光装置精密装校实验室的三维重建技术及应用[J].计算机辅助设计与图形学学报,2017,29(5):854-859.QUAN X S,LIU C C,CHEN H P,et al.3Dreconstructing technique and application of high-power laser facility's optics assembly building[J].Journal of Computer-Aided Design&Computer Graphics,2017,29(5):854-859.
[6]牛晰,袁晓东.基于物体特征有效提取和离散点三维重建的3D扫描系统原型研究[J].计算机测量与控制,2017(5):158-161.NIU X,YUAN X D.Research of a prototype 3Dscanning system based on object features active extraction and 3D reconstruction with discrete points[J].Computer Measurement&Control,2017(5):158-161.
[7]韩硕,童晓冲,卢学良,等.手持线激光扫描视频三维重建中的运动线提取算法[J].计算机应用与软件,2016,33(7):154-158.HAN S,TONG X C,LU X L,et al.Moving line extraction algorithm in 3D reconstruction of handheld line laser scanning video[J].Computer Applications and Software,2016,33(7):154-158.
[8]白洪伟,李进,穆星,等.近景摄影测量在地下三维重建中的应用[J].沈阳大学学报(自然科学版),2016,28(3):223-226.BAI H W,LI J,MU X,et al.Application of close range photogrammetry in underground 3D reconstruction[J].Journal of Shenyang University(Natural Science),2016,28(3):223-226.
[9]赵广军,王旭初,牛彦敏,等.基于SAE深度特征学习的数字人脑切片图像分割[J].计算机辅助设计与图形学学报,2016,28(8):1297-1305.ZHAO G J,WANG X C,NIU Y M,et al.Digital human brain slice image segmentation based on SAE depth feature learning[J].Journal of Computer-Aided Design and Graphics,2016,28(8):1297-1305.
[10]刘丹,诸叶平,李书钦,等.基于实测数据的小麦叶片模拟模型与三维重建:以偃展4110为例[J].山东农业科学,2016,48(7):137-141.LIU D,ZHU Y P,LI S Q,et al.Simulation model and 3D reconstruction of wheat leaves based on measured data:a case study of Yanzhan 4110[J].Shandong Agricultural Sciences,2016,48(7):137-141.
[11]孟偲,曲圣悦,王利军,等.用于C臂快速标定的平面靶标可见标记点的提取与配准[J].高技术通讯,2015,25(4):397-403.MENG S,QU S Y,WANG L J,et al.Extraction and match of visible markers on a planar phantom for fast Carm calibration[J].Chinese High Technology Letters,2015,25(4):397-403.
[12]江静,蔡国榕,李绍滋,等.基于多结构估计的建筑物激光雷达点云特征线提取算法[J].厦门大学学报(自然版),2015,54(3):390-396.JIANG J,CAI G R,LI S Z,et al.Extract feature lines from building LiDAR point cloud based on multi-structure estimators[J].Journal of Xiamen University(Natural Science),2015,54(3):390-396.