Learning Strategies to Select Point Cloud Descriptors for 3D Object Classification: A Proposal
Fulltekst
RELATERTE DOKUMENTER
Figure 1: A watertight manifold surface triangulation re- constructed by our eigencrust algorithm; a photograph of the source object; the point cloud input to the algorithm, with
We present an approach to object detection and recognition in a digital image using a classification method that is based on the application of a set of features that include
In contrast to shape retrieval approaches based on global descriptors where object similarity can be determined in a straight forward way by computing the distance between
We move from an objective reading of the object (point cloud) to a reading enhanced by the cognitive input from an operator (architectural semantisation work).. This
The individual physics engine has to provide appropriate shape types to derive an equivalent physical object based on the data of a 3D object in X3DOM.. Beyond that,
It is based on a combination of a single view 3D shape descriptor and a temporal descriptor (initially used to accumulate appearance descriptors of an object over time) that
In this work, a partial 3D object retrieval method is pro- posed, which starts from a variation of fast PFH (FPFH) that is adaptive to the mean point distances of a point cloud
The result of the learning process is an autonomously learned strategy of selection of descriptors with the property that the successive application of these descriptors to a 3D