The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Binary image decomposition and compression using mathematical morphology for object recognition
Author:	Gogusetti, Venu Gopal
View Online:	njit-etd1992-075 (vii, 96 [10] pages ~ 2.2 MB pdf)
Department:	Department of Computer and Information Science
Degree:	Master of Science
Program:	Computer Science
Document Type:	Thesis
Advisory Committee:	Shih, Frank Y. (Committee chair)
Date:	1992-01
Keywords:	Pattern recognition -- Data processing Morphology Data compression (Computer science) Image processing -- Digital techniques
Availability:	Unrestricted
Abstract:	Image segmentation has been studied for several years. There are several segmentation techniques which are fast and effective but all of them are lacking the property of invariency to shift, rotation and sizing. In this study a new process is introduced which overcomes the shift, size and rotation variance and the compressed data can be used for object recognition. Euclidean distance measurement is used in the compression process which is rotation invarient but is expensive in terms of time. Eucledean distance transformation is calculated using optimal double two scan algorithm with gray scale morphology, a new method developed by Dr. Shih and Mr. Wu. Mathematical morphology provides an effective tool for image analysis. Many parallel image computers support basic morphological operations. A new image segmentation technique is presented, which is more usefull in object recognition as it is invarient to shif, orientation and is proportional to sized images.