سال انتشار: ۱۳۸۱

محل انتشار: دومین کنفرانس ماشین بینایی و پردازش تصویر

تعداد صفحات: ۷

نویسنده(ها):

Mahdi Amiri – Sharif University of Technology, Computer Eng. Department,Digital Multimedia Research Laboratory (DML) P.O. Box 11365-9517, Tehran, Iran
Hamid Reza Rabiee – Sharif University of Technology, Computer Eng. Department,Digital Multimedia Research Laboratory (DML)
Farid Behazin – Defense Systems Institute, Aerospace Industries Organization, Tehran, Iran

چکیده:

In this paper we present the building blocks of a general multiresolution algorithm for object detection. The proposed algorithm is based on the decomposition of images using undecimated wavelet packet trees. It exploits adaptation both in the selection of best bases and in selection of asuitable threshold for the chosen features. The entropy measure has been used for best basis selection, and features have been chosen from the wavelet coefficients. Moreover, a detection system based on the proposed algorithm is under development