Comparisons on Similar Signals

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

NSC95-2115-M019-002

Translated Name/計畫中文名

訊號相似度的比較

Project Coordinator/計畫主持人

Mong-Shu Lee

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Computer Science and Engineering

Website

https://www.grb.gov.tw/search/planDetail?id=1276942

Year

2006

Start date/計畫起

01-08-2006

Expected Completion/計畫迄

31-07-2007

Bugetid/研究經費

395千元

ResearchField/研究領域

數學

在此計畫中我們提出一個計算一維及二維失真訊號相似度的方法，而這個作法的靈感來自於最近發展出來的空間域結構相似度演算法的成功，其主要構想是我們利用這些失真訊號所引導出複數小波係數值的改變上，較有一致性而不是無規則可循。因為這個性質，在此計畫中我們得以推廣空間域結構相似度的量測方法到頻率域，我們目前分別以基因表現時間序列及小的失真數字影像做為一維及二維訊號的測試資料，先前的實作結果顯示我們所用的方法對於訊號在小尺度的位移、縮小放大、甚至旋轉上具有強韌性，即與傳統方法比較而言，我們的方法可以提高相似度的辨認。 In this project we proposed an approach to evaluate the similarity of one and two-dimensional distorted signals. This work is inspired by the success of the spatial domain structure similarity index algorithm. The key idea is to make use of the fact that these signal distortions lead to consistent magnitude changes of dual tree wavelet coefficients. We attempt to extend the current structure similarity measure to the wavelet transform domain. We have tested our algorithm on gene expression time series data and small distorted digit images for one and two-dimensional signals, respectively. Preliminary experiments demonstrate that the proposed structure similarity measure provides robustness to translation, spatial scaling and rotation when these geometric distortions are small.

Keyword(s)

小波
訊號相似度
基因表現時間序列
Wavelets
Signal similarity
Gene expression time series

DSpace CRIS

Comparisons on Similar Signals

Details

Description