获得三维曲面相对大小高斯曲率的方法与实现

Methods and Implementation of Getting the Relative Size Gaussian Curvature in 3D Surface

本方法在不使用任何参照物的前提下,从对象图像中获得曲面的三维表面重建信息。通过物体自身的多角度旋转并拍摄多枚照片,并从多枚照片中获得的数据使用神经网络的学习数据。从而最终获得相对的高斯曲率大小值、曲率的符号以及曲面的类型。从而达到无任何参照物的前提下,获得对象物体的三维形状复原信息。

Under the premise of no reference frame, this method gets the 3D surface rebulding information of the hook face from the object image. It takes many photos by the multiple rotation of the object itself, and uses the learning data of the neural network by the data of the photos. Finally, it can get the relative size value of Gaussian curvature, the symbols of the curvature and the types of the hook face,and then to get the reversion information of the 3D shape of the object without any reference frame.