Accession Number : ADA138421

Title :   Image Filtering with Boolean and Statistical Operators.

Descriptive Note : Master's thesis,

Corporate Author : AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING

Personal Author(s) : Wells,R D

PDF Url : ADA138421

Report Date : Dec 1983

Pagination or Media Count : 230

Abstract : Edge extraction is an image processing technique for defining the edge information in an image. This effort researches different edging processes as applied to preprocessing for two pattern recognition processes. The first one is a cross-correlation method to find a target given that the target has a known size, orientation, and aspect. Correlation is performed in the spatial frequency domain with two-dimensional fast Fourier transforms of the searched edge image and a hand drawn edge template to correct for translation only. The second pattern recognition process researched also uses edging as one step of a purely spatial domain algorithm. The approach locates targets in infrared images that can be described as hot clusters. A cluster recognition algorithm by Hamadani is implemented and altered for testing of local thresholding and thresholding rules. The algorithm is shown to be effective on real infrared images. (Author)

Descriptors :   *Image processing, *Boolean algebra, *Statistical processes, *Pattern recognition, Edges, Extraction, Images, Cross correlation, Target detection, Spatial distribution, Frequency, Fast Fourier transforms, Two dimensional, Templates, Algorithms, Infrared images, Clustering, Threshold effects, Air Force research, Theses

Subject Categories : Statistics and Probability
      Optical Detection and Detectors
      Infrared Detection and Detectors
      Optics

Distribution Statement : APPROVED FOR PUBLIC RELEASE