基于神经网络的三维图象恢复与重建新算法  

New Method for Three-dimensional Reconstruction Based on Neural Network

作　　者：袁友伟[1] 贺素良[1] 鄢腊梅[1] 

机构地区：[1]株洲工学院计算机系,株洲412008

出　　处：《计算机工程》2001年第10期69-70,共2页Computer Engineering

基　　金：湖南省自然科学基金资助项目(00JJY2082)

摘　　要：简介了兰伯特Lambertian反射模型用于三维图象的恢复与重建的算法,由于其具有很大的局限性,为此,给出了一种基于神经网络与模糊处理相结合的新型反射模型,用于三维图象的恢复与重建,且新算法不需知道光源方向,经实验证明,具有计算快捷且图象识别精度高的特点。In this paper, the limitation of the conventional Lambertian reflectance model is addressed and a new neural-based reflectance model is proposed. The new neural learning algorithm is to optimize a proper reflectance model and to recover the object surface by a simple shape-form-shading(SFS) variational method with this neural-based model and fuzzy method model. An example is also given to prove that the SFS technique is robust for most objects, even when the lighting conditions are uncertain.

关 键 词：模糊神经网络 三维图象重建 图象恢复 图象处理学 计算机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]