Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/60482
Title: Image recognition of occluded objects based on improved curve moment invariants
Authors: Lichun, K.
Bin, L.K. 
Jin, Y.
Keywords: Curve moment invariants
Image recogniton
Object recognition
Occluded object
Issue Date: Jun-2009
Source: Lichun, K.,Bin, L.K.,Jin, Y. (2009-06). Image recognition of occluded objects based on improved curve moment invariants. Journal of Digital Information Management 7 (3) : 152-158. ScholarBank@NUS Repository.
Abstract: This paper presents some fundamental knowledge of image recognition of partially occluded objects, and brings forward an improved set of curve moment invariants to recognize occluded objects in images. Occluded objects in an image are not rich in information, while the improved curve moment invariants can represent those objects in a unique way. Therefore, the occluded objects can be efficiently recognized through their shape representation by the set of curve moment invariants as their features. A number of model objects are predefined in the experiments to build the prior knowledge database for the sake of matching and recognizing unknown objects in future input images. Image pre-processing is employed for both model objects and unknown input objects to obtain their boundary and partition it into subparts. Corner points along the boundary are detected after image pre-processing so that the objects can be described by the curve moment invariants of each subpart. Finally, the matching of curve moment invariants between model object and unknown object is carried on to recognize the unknown object in the input scene image. The results of experiments show that this method is robust and efficient to recognize partially occluded objects in 2D images; furthermore, the invariance to the affine transformation of objects in images are also proved in the paper.
Source Title: Journal of Digital Information Management
URI: http://scholarbank.nus.edu.sg/handle/10635/60482
ISSN: 09727272
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

Page view(s)

30
checked on Dec 8, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.