ZAMAN-FREKANS DÜZLEMİNDE SİNYAL BİLEŞENİ ÇIKARIMI İÇİN YENİ BİR YÖNTEM

Özet

have non-overlapping compact support in the time-frequency plane, is developed. Proposed technique is observed to be successful even under high noise levels. The method is composed of three main steps: 1)detection of signal components, 2)estimation of instantaneous frequencies of the detected components 3) filtering in the time-frequency plane. By construction, it is an iterative algorithm which detects and extracts one component at a time. Time-frequency distributions are utilized for signal component detection and instantaneous frequency estimation. Principle curve projections, which is very robust to noise, is used for instantaneous frequency estimation. Filtering in the time-frequency plane is accomplished by frequency warping. The performance of the proposed algorithm is analyzed on synthetic data sets for different noise levels

Abstract

A new time-frequency signal analysis technique for detection and extraction of signal components, which have non-overlapping compact support in the time-frequency plane, is developed. Proposed technique is observed to be successful even under high noise levels. The method is composed of three main steps: 1)detection of signal components, 2)estimation of instantaneous frequencies of the detected components 3) filtering in the time-frequency plane. By construction, it is an iterative algorithm which detects and extracts one component at a time. Time-frequency distributions are utilized for signal component detection and instantaneous frequency estimation. Principle curve projections, which is very robust to noise, is used for instantaneous frequency estimation. Filtering in the time-frequency plane is accomplished by frequency warping. The performance of the proposed algorithm is analyzed on synthetic data sets for different noise levels

Anahtar kelimeler: Sinyal bileşeni, Zaman-frekans dağılımı, Anlık frekans, Ana eğri izdüşümleri