Creation of innovative signal and image processing method by realization of variable filter band discrete wavelet transform

2016 Fiscal Year Final Research Report

• Project/Area Number: 26420387
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Measurement engineering
• Research Institution: Toyohashi University of Technology
• Principal Investigator: Zhang Zhong 豊橋技術科学大学, 工学(系)研究科(研究院), 教授 (50254579)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: ウェーブレット変換 / 胎児心電 / 腫瘤 / 半導体 / 欠陥検出

Outline of Final Research Achievements

Discrete wavelet transform (DWT) has excellent characteristics, such as high calculation speed, perfect reconstruction, and has been applied wide field. However, the mother wavelet has the characteristics of the octave band, and it does not necessarily become the optimum basis. Therefore, it exerts a serious influence on data compression, noise deletion and the like. Also, image processing using it has poor direction selectivity. In this research, we propose a new variable filter band discrete wavelet transform (VFB - DWT) by devising the optimal orthogonal base construction method suitable for the signal while preserving the splendor of the conventional DWT. In addition, VFB-DWT was expanded to 2D / 3D DWT to realize free direction selectivity, signal and image processing method was created. Furthermore, we applied this method to extraction of fetal electrocardiogram, defect detection of semiconductor circuits, contour extraction of 3D masses, and confirmed its effectiveness.

Free Research Field

計測工学