Visual recognition is a vital problem in the area of computer vision. From robotics to information retrieval, the applications demand the ability to identify and localize categories, places, and objects. The goal is to recognize objects in an image by matching with known objects stored in a database. This paper proposes an algorithm to address the problem of recognition of partially occluded objects using 2D imaging. Shape-based object recognition has been used to match the shape model created from the image with the model templates. Model templates with different orientation of the object to be recognized have been used in the matching process. Pyramid level matching and unique image scaling have been used to recognize the occluded objects. The proposed algorithm has been tested using a real time setup and the results were presented.
Volume 11 | 02-Special Issue
Pages: 44-50