Paper Info
Title
Fake modern Chinese painting identification based on spectral–spatial feature fusion on hyperspectral image
Abstract
Chinese painting is famous and valuable for special painting materials, skills and final art effects used, yet this has resulted in many fake paintings being produced. Those fake paintings were normally made by using modern high resolution scanning and printing technology, thus it is very hard to identify the fake ones by human vision. To address this challenging problem, in this paper, a hyperspectral image based features fusion method is proposed. Firstly, we scan Chinese paintings using a visual band hyperspectral camera with the spectral frequency ranging from 400 to 900 nm. Then, the spectral and spatial features are extracted respectively by using the principal component analysis and a convolution neural network. Finally, we fuse these two kinds of features and input the feature set into a support vector machines for classification. All samples of real and fake paintings are obtained from local Chinese painting organization. The experimental result shows the effectiveness of the proposed method with an accuracy achieved of 84.6 %, which is significantly higher than other approaches where only spectral or spatial feature is used.
Year
DOI
Venue
2016
https://doi.org/10.1007/s11045-016-0429-9
Multidimensional Systems and Signal Processing
Keywords
Field
DocType
Hyperspectral image,Chinese paintings,Fake identification,Spectral–spatial feature fusion
Computer vision,Feature fusion,Convolutional neural network,Support vector machine,Painting,Hyperspectral imaging,Ranging,Artificial intelligence,Fuse (electrical),Mathematics,Principal component analysis
Journal
Volume
Issue
ISSN
27
4
0923-6082
Citations 
PageRank 
References 
1
0.38
7
Authors
5
Name
Order
Citations
PageRank
Zheng Wang1848.26
Dongying Lu210.38
Dong Zhang310.38
Meijun Sun47411.77
Y. Zhou516337.69