未知光源位置环境中物体形状恢复的神经网络方法研究
Shape Recovery in Unknown Environment by Neural Networks
-
摘要: 用神经网络方法解决未知光源位置环境中物体三维形状恢复的问题.对漫反射表 面,用神经网络方法由已知表面形状物体及其对应图像的灰度值进行学习,所得权值可视为 环境光源参数.由此可恢复同样光源环境中其它物体的三维形状.实验证明,神经网络方法可 以解决未知光源位置环境(包括多个光源)中漫反射表面物体的三维形状恢复问题.Abstract: In this paper, we propose a new method based on neural networks to recover the 3D shape of an object with Lambertian surface. The method can be applied under multiple light sources of unknown distances and sizes. A mulfiqayer neural network is used to learn the mapping between the image intensities and the corresponding surface normals. A sphere is used as the calibration object in training the neural network by a backpropagation algorithm. The weights of the network can be regarded as the environment parameters. The efficiency of the method is verified by simulated and real experiments.
-
Key words:
- Neural network /
- shape recovery /
- lambertian surface
-
计量
- 文章访问数: 3044
- HTML全文浏览量: 99
- PDF下载量: 929
- 被引次数: 0
下载:
