全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...

一种快速精确的人脸三维形状重构新方法

DOI: 10.11834/jig.200308336

Keywords: 计算机图象处理(520?6040),从明暗恢复形状,兰伯特(Lambertian),反射模型,BP神经网络,误差补偿

Full-Text   Cite this paper   Add to My Lib

Abstract:

针对传统SFS(ShapfromShading)的不足,提出了一种新的基于BP神经网络的明暗恢复形状的方法,该方法是基于兰伯特(Lambertian)反射模型的改进算法,利用了BP神经网络强的非线性映射能力,将Lambertian表面反射模型与光滑表面模型相结合,然后再利用一些已知条件,构成SFS问题的正则化模型;变换不同的照明条件,将模型平移或旋转获得多幅图象,以增加约束条件;计算出误差补偿参数去修正邻域内的三维误差.由于考虑了邻域的平均值,使算法的稳定性和精确性都得到了加强.实例表明,该算法较传统的算法更快和更精确

References

[1]  [2]Chao Siu-Yeung, Tommy W S Chow. A neural-learning-based reflectance model for 3-D shape reconstruction [J ]. IEEE Transactions on Industrial Electronics, 2000, 47 ( 12 ): 1346 ~1350.
[2]  [3]赵保军,史彩成,沈胜宏等.一种基于CMAC的图象恢复算法[J].中国图象图形学报.2001,6A(8):780~783.
[3]  [6]陈刚,戚飞虎.实用人脸系统的本征脸法实现[J].计算机研究与发展.2001,38(2):170~175.
[4]  [8]Lisboa P G J,刑春颍,阳影.现代神经网络应用[M].北京:电子工业出版社,1996:54~58.
[5]  [9]Samuel Pierre, Hassance Said, Wilfreied G Probst. An artifical neural network approach for routing in distributed computer networks[J]. Engineering Application of Artificial Intelligence,2001,14 (4) : 51~ 60.
[6]  [10]Turk M, Pentland A. Eigenface for recognition[J]. Jouranl of Cognitive Neuroscience, 1991,3(1) :71~86.
[7]  [1]YUAN You-wei, YAN La-mei. Active networks for efficient and distributed network management based on an artificial neural networks [A]. In: DCABES 2001 Proceedings [C]. Wuhan,China, Wu han University of technology press, 2001:172 ~ 175.
[8]  [4]廖熠,赵荣椿.从明暗恢复形状(SFS)的几类典型算法分析与评价[J].中国图象图形学报,2001,6A(10):953~956.
[9]  [5]Zhou Y T, Chellappa R C. Artifical neural networks for computer vision[M]. New York: Springer-Verlag, 1992.
[10]  [7]Dibyendu Nandy, Jezekiel Ben-Arie. Shape form recognition:A novel approach for face shape from shading [J ]. IEEE Transactions on Image Processing, 2001,10(2) : 206~ 207.

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133