Imaging Science and Photochemistry ›› 2017, Vol. 35 ›› Issue (1): 88-96.DOI: 10.7517/j.issn.1674-0475.2017.01.088

Previous Articles    

Study on Spectral Reconstruction Method Based on Optimized Selected Samples

LONG Yanqun1, WANG Huiqin1,2, WANG Ke1,2, WANG Zhan3, ZHAO Suwen1   

  1. 1. School of Information and Control Engineering, Xi'an University of Architecture and Technology, Xi'an 710055, Shaanxi, P. R. China;
    2. School of Management, Xi'an University of Architecture and Technology, Xi'an 710055, Shaanxi, P. R. China;
    3. Shanxi Provincial Institute of Cultural Relics Protection, Xi'an 710075, Shaanxi, P. R. China
  • Received:2016-07-07 Online:2017-01-15 Published:2017-01-15

Abstract:

Aiming at the defect of existing sample selection methods for spectral reflectance reconstruction, a new method based on the kernel fuzzy C-means clustering optimization sample selection is presented.This method considers the extensiveness of spectral reflectance space and the similarity of chroma space, meeting the accuracy of spectral reconstruction. Firstly, C samples are selected as the initial clustered core in the spectral reflectance space. And the original spectrais turned into(a*,b*) color space. At the same time, 2D chrominance space is mapped to the 3D feature space with kernel function, making characteristics linearly separable,to achieve a better result.The experiments show that this method selects training samples for spectral reflectance reconstruction can further improve the accuracy of spectral reconstruction, and it is better than the existing methods in the chroma and accuracy assessment effects.

Key words: spectroscopy, kernel fuzzy clustering, sample selection, spectral reconstruction