摘要
分析了应用于三维曲面重构中的神经网络和自适应模糊神经网络两种算法,并结合具体的应用例子对两种算法在物体三维模型重建及曲面再现中的优劣进行了比较.试验结果表明,在相同的网络模型训练时间及模型充分收敛的前提下,自适应模糊神经网(ANFIS)算法比神经网络(ANN)算法具有更快的收敛速度和更高的重构精度,而且ANFIS更适合对表面复杂的三维物体进行模型重构和曲面再现.
The both of algorithms： neural network（ANN ） and adaptive network based on fuzzy inference system（ANFIS） that are applied to 3D surface reconstruction are analyzed respectively, and the superiority or inferiority of them on recontruction and shape recovery for 3D surface are compared via .some real examples, it is shown that based on the same network training time, the convergence rate and modeling precision of ANFIS are better than ANN, and ANFIS is more fit for the reconstruction and shape recovery of 3D object with complex surface.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第5期200-203,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(50575029)
关键词
逆向工程
曲面重构
神经网络
自适应模糊神经网络
reverse engineering
surface reconstruction
neural network
adaptive network based on fuzzy inference system（ANFIS）
