Development of a Connectionist Model for Image Skeletonization.

Date of Submission

December 1994

Date of Award

Winter 12-12-1995

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science

Department

Machine Intelligence Unit (MIU-Kolkata)

Supervisor

Pal, Nikhil Ranjan (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

Image skeletonization, also popularly known as thiming, is a process in which a given image is transformed to a single pixel thick image preserving important im- age features such as skeletal points, image symmetry, connectivity etc. It helps in reducing the storage space requirement for storing imput data by removing unnee- essary details and is also useful in extraction of the important image features, It is an important preprocessing step for many image analysis tasks such as finger print identification, optical character rccognition, remot.ely sensed imagery, automated in- spection of PCBS ctc.In literature, there exists several thiuning algorithus which can be broadly classitied into two categories: scquential and parallel algorithms. A parallel algorithm uses only the result obtained from the previous iteration for deleting(or reducing graylevel of) a pixel in the current iteration [1, 3, 4, 5, 7, 11]. On the other hand, a sequent ial algorithm makes use of both the result obtained from the previous iteration as well as the results obtained so far in the current iteration to process the candidate pixel [7, 13, 14]. The sequential algorithms have advantage that they are faster to implement. on the usual general purpose sequential computers than parallel algorithms |17].In case of two-tone images, there are only two graylevels one for background pixels and other for object pixels. Using a set of rules, it is required to decide whether a pixel is to be deleted or not. If the pixel satisfies the deletion criteria then its graylevel is changed to background graylevel. On the other hand, in case of graytone images, the whole deletion procedure becomcs rather more complicated due to the fact that the object as well as background both may have a range of graylevels instead of single graylevels. One approach for thinning a graytone image, is to first convert it into a two-tone image by suitable thresholding and then apply any two-tone thinning algorithm to get. the desired result. A problem associated with this method is that the information other than the object outline is lost forever and further processing in the graytone domain is not possible. Also, the method is very seusitive to noise. All these shortcomings could possibly be eliminated to an extent if thinning is perfor in the graytone domain itself. An additional advantage of using a graytone thim " algorithm is t hat the resulting image is enbanced both in contrast of the object and

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:28843133

Control Number

ISI-DISS-1994-18

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/10263/6193

This document is currently not available here.

Share

COinS