Unformatted text preview:

Object recognitionInvariantsModel PropertiesShape representationModels used for shape representationContd.Model Based RecognitionGraph SearchingPowerPoint PresentationRecognitionConclusion1Object recognition •What is object recognition?•Why is it needed? 1.To move around safely avoiding objects 2.To pick and place objects 3.Inspect objects•Two major issues: 1.representation of models 2.matching of models to images2Invariants•We can use invariants to recognize simple objects: 1.They provide a simple means of comparison 2.They can provide orientation and position information. 3.Too simplistic for some applications 4.Does not give unique means of identification.3Model Properties•Desirable model properties 1.direct pairing of model features and observed data features 2.direct estimation of object position and orientation 3.prediction of the appearance of the object from any position4Shape representation•Common shape representation methods are characterized on different points: 1.Input representation form 2.Object reconstruction ability 3.Incomplete shape recognition ability 4.Local global description characters 5.Robustness of description.• Models used for shape representation are wireframe models,set- theoretic models and boundary representation models.5Models used for shape representation•Wireframe Models 1.Store edges and vertices in a list. 2.Polyhedral representation. 3.Edge based matching. 4.Problems with ambiguous and impossible representations. 5.Problems due to no solid matter of the object represented. Solid Models are better.6Contd.•Set Theoretic Modelling 1.Model is composed of primitive shapes assembled with set operators. 2.Set operators used are union(U),difference(-) and intersection.7Model Based Recognition•Recognition is matching problem between scene and model description.•There are three methods which are broadly used : 1.Tree-search methods 2.Relaxed labeling methods 3.Graph Search methods(we look at this method in detail).8Graph Searching•Object model and scene features are represented in relational graph structure.•A graph consists of 1.A set of nodes connected by links(also called edges or arcs). 2.Each node represents an object feature. 3.Node can be labeled with each of the feature’s properties. 4.Links of the graph represent relationships between features a.distance between centroids of features b.Adjacency of the features.910Recognition•A matter of matching two graphs-graph of object model to the graph of scene containing object.•Matching methods must take into account occlusion and overlapping objects.•A graph derived from a solid model of the mug would contain the bottom which is missing from the view shown (and scene model).•So the problem is that of finding a subgraph of the complete graph derived from the solid model.•This is a large search space problem -- Use Constraints.•Thus Graph theory is big topic in its own right.11Conclusion•The generation of sufficiently rich and descriptive representations from images,segmentation and grouping to identify those features that belong together, and the matching of these to object models are difficult research problems under active


View Full Document

UMBC CMSC 671 - Object recognition

Download Object recognition
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Object recognition and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Object recognition 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?