Date of Submission

9-22-1994

Date of Award

9-22-1995

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Subject Name

Computer Science

Department

Machine Intelligence Unit (MIU-Kolkata)

Supervisor

Pal, Sankar Kumar (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

Recognition of objects in an image, according to Suetens et al. [1), relers to the task of finding and labeling parts of a two-dimensional image of a scene that correspond to the real objects in the scene. Object recognition is necessary in a variety of domains like robot navigation, aerial imagery analysis, industrial inspection and so on. Normally, different strategies for object recognition (1-(5] involve establishing some model for each object, i.e., some general description of each object, and then labeling different parts of the scene according to the knowledge about the models.Object models can have two-dimensional (2D) or three-climensional (3D) descrip- tions. 2D descriptions are generated from viewer centered representations where each view is represented using shape features derived from graylevel or binary im- ages of the prototype object models. On the other hand, 3D descriptions require view point independent volumetric representations those permit computations at an arbitrary viewpoint. Generation of 3D descriptions from the captured 2D im- ages is a computationally difficult problem (6] (one approach is to generate 2 D sketch from the 2D image [7]), whereas the 2D descriptions of the objects can be constructed more easily. Construction of 2D descriptions is straight forward for flat objects like hammer, spanner etc., and such descriptions are often used in in- spection problems of flat industrial objects. Moreover, in several other tasks like character recognition, analysis of remotely sensed imagery etc, 3D information is neither necessary nor available, and information processing is to be performed only on the basis of 2D descriptions. The present thesis concerus with the tasks related only to 2D descriptions (2D object recognition). 2D object recognition involves mainly two stages, first, extraction of features from the captured image and the associated preprocessing tasks, and second, interpretation of the extracted feature set. These two stages are briefly explained below. A : Feature Extraction and Related Preprocessing In the image grabbing process, the light coming from the scene is projected onto a plane (image plane), and the image plane content is digitized into a two-dimensional array. Each location in the array specifies a position in the image plane, and the location contains an integer value specifying the intensity of the image at that position, i.e., the amount of light received from tlhe scene after projection (at that location). In the case of a color image, the color information is also stored in each location of the array. To recognize the objects present in the image of the scene, the image is preprocessed in sev- eral stages, and consequently some characterizing features are derived from tlhe scene after projection (at that location). In the case of a color image, the color information is also stored in each location of the array. To recognize the objects present in the image of the scene, the image is preprocessed in sev- eral stages, and consequently some characterizing features are derived from the image. These features are used for further interpretation task. Different stages involved in the feature extraction and allied preprocessing tasks are as follows :

Comments

ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28842887

Control Number

ISILib-TH292

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

DOI

http://dspace.isical.ac.in:8080/jspui/handle/10263/2146

Download

Included in

Mathematics Commons

Share

COinS