Open AccessData Expansion Method Based on Wavelet Transform
Author(s) -
Xiaochun Li,
Li Wen,
Zhe Liu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1631/1/012128
Subject(s) - magnification , artificial intelligence , computer science , wavelet transform , pattern recognition (psychology) , wavelet , computer vision , image (mathematics)
For small sample data, data expansion is an effective means to improve the accuracy of target recognition. To image magnification to enhance image resolution as a starting point, the author puts forward two kinds of image magnification method based on wavelet transform. Qualitative analysis on the experimental data and quantitative analysis using average gradient, cross entropy as performance index, show that the proposed image magnification method has obvious advantages in terms of image definition, maintaining image gray scale information. Finally, the author proposes the method of image expansion based on the image magnification data of the four methods, which greatly enriches the diversification of the training set on resolution, clarity, contrast, integrity and so on, and theoretically analyzes the feasibility of improving the recognition accuracy of the data with few samples and small targets.