سال انتشار: ۱۳۹۰

محل انتشار: هفتمین کنفرانس ماشین بینایی و پردازش تصویر

تعداد صفحات: ۶

نویسنده(ها):

Mehdi Sajjadi – Digital Signal Processing Research Lab,Department of Electrical and Computer Engineering,Isfahan University of Technology, Isfahan, 84156-83111, Iran
Rassoul Amirfattahi – Digital Signal Processing Research Lab,Department of Electrical and Computer Engineering,Isfahan University of Technology, Isfahan, 84156-83111, Iran
Mohammad Reza Ahmadzadeh – Digital Signal Processing Research Lab,Department of Electrical and Computer Engineering,Isfahan University of Technology, Isfahan, 84156-83111, Iran

چکیده:

Brain CT images are extensively used in detectionof Cerebrovascular Accidents (CVA). But CT scan is not reallya sensitive test for diagnosis of ischemic stroke. Most of thetimes, early CT images barely show or even don’t show anynoticeable sign of ischemia. But it is possible to put emphasison subtle signs of ischemia by means of image contrastenhancement. Contrast enhancement using multiscaletransforms is one of the most popular approaches in recentyears. Nonsubsampled Contourlet Transform (NSCT) is adirectional multiscale transform that is able to take geometricalstructures into account. In this paper we are about to customizethis transform and improve its performance in contrastenhancement of brain CT images. To this end, we propose a setof simpler and more efficient filters for NonsubsampledPyramid (NSP) and introduce an improved method formodification of directional subbands’ coefficients.