Viewpoint Feature Histogram

In any case, for manipulation, we require that the robot not only identifies objects, but also recognizes their 6DOF poses for grasping. FPFH is invariant both to object scale (distance) and object pose and so cannot achieve the latter task.

由于FPFH的尺度和姿态的不变性，不能用来估计位姿。

In this work, we decided to leverage the strong recognition results of FPFH, but to add in viewpoint variance while retaining invariance to scale

Our contribution to the problem of object recognition and pose identification is to extend the FPFH to be estimated for the entire object cluster , and to compute additional statistics between the viewpoint direction and the normals estimated at each point.  To do this, we used the key idea of mixing the viewpoint direction directl y into the relative normal angle calculation in the FPFH

在每个点（？）上计算视点方向和估计法线的夹角。

 Figure 6 presents this idea with the new feature consisting of two parts:(1) a viewpoint direction component (see Figure 5) and (2) a surface shape component comprised of an extended FPFH (see Figure 4).

 

Fig. 5. The Viewpoint Feature Histogram is created from the extended Fast Point Feature Histogram as seen in Figure 4 together with the statistics of the relative angles between each surface normal to the central viewpoint direction

 

The viewpoint component is computed by collecting a histogram of the angles that the viewpoint direction makes with each normal. Note, we do not mean the view angle to each normal as this would not be scale invariant, but instead we mean the angle between the central viewpoint direction translated to each normal.

意思是：不是每个法线和视点的夹角，因为它不是不变的。是视点方向转到每个法线  之间的角度------迷。

 

 

直方图怎么看：The computational complexity of VFH is O(n). In our experiments, we divided the viewpoint angles into 128 bins and the α,ϕ and θ angles into 45 bins each or a total of 263 dimensions.

 

参考资料：Radu Bogdan Rusu ; Gary Bradski ; Romain Thibaux ; John Hsu，Fast 3D recognition and pose using the Viewpoint Feature Histogram