Free-Shaped Object Recognition Method from Partial Views Using Weighted Cone Curvatures
This work presents a method for free-shaped object recognition from its partial views. Consecutive database reductions are achieved in three stages by using effective discriminant features. These features are extracted from the spherical mesh representation used to modeling the partial view and from the view range data itself. The used characteristics are global, which means that they can not represent the views univocally. However, their staged application allows the initial object database to be reduced to selecting just one candidate in the final stage with a high success rate. Yet, the most powerful search reduction is achieved in the first stage where the new Weighted Cone Curvature (WCC) parameter is processed. The work is devoted to describe the overall method making especial emphasis in the WCC feature and its application to partial views recognition. Results with real objects range data are also presented in the paper.
KeywordsWave Front Machine Intelligence Range Data Partial Model Iterative Close Point
- 4.Campbell, R.J., Flynn, P.J.: Eigenshapes for 3D object recognition in range data. In: Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, Fort Collins, Colorado, June 1999, vol. 2, pp. 2505–2510 (1999)Google Scholar
- 7.Skocaj, D., Leonardis, A.: Robust recognition and pose determination of 3-D objects using range image in eigenspace approach. In: Proc. of 3DIM 2001, pp. 171–178 (2001)Google Scholar