Quantitative fuzzy measures for threshold selection

Discrimination of Image Qualities for some applications in Computer Vision is very much important. The process of evaluation of image quality is very difficult as it possesses noise which is fuzzy in nature. Threshold selection for object extraction, for a bimodal or multi modal histogram is still a...

Full description

Saved in:
Bibliographic Details
Published inPattern recognition letters Vol. 21; no. 1; pp. 1 - 7
Main Authors Ramar, K., Arumugam, S., Sivanandam, S.N., Ganesan, L., Manimegalai, D.
Format Journal Article
LanguageEnglish
Published Elsevier B.V 01.01.2000
Subjects
Fuzzy measures
Fuzzy neuro approach, comparison of performance
Image enhancement
Threshold, object extraction
Threshold, object extraction
Fuzzy neuro approach, comparison of performance
Fuzzy measures
Image enhancement
Online AccessGet full text

Cover

More Information
Summary:Discrimination of Image Qualities for some applications in Computer Vision is very much important. The process of evaluation of image quality is very difficult as it possesses noise which is fuzzy in nature. Threshold selection for object extraction, for a bimodal or multi modal histogram is still a difficult problem. It is proposed in this paper that based on the four fuzzy measures, namely, linear and quadratic index of fuzziness, logarithmic and exponential entropy measures, how the best threshold will be identified and used. A comparative study of four such fuzzy measures for real life images has been carried out and promising results are obtained. Selection of the best threshold is tried out using a neural network (BPN) as it helps for fine tuning.
ISSN:0167-8655
1872-7344
DOI:10.1016/S0167-8655(99)00120-8