Spectral sparse recovery form a single RGB image

Authors:
  • GWGuangyuan Wu
    Guangyuan Wu
  • YXYifan Xiong
    Yifan Xiong
  • XLXiaozhou Li
    Xiaozhou Li

Abstract
An efficient procedure for spectral reflectance recovery using a new compressive sensing-based spectral sparse recovery method is proposed by using the optimized sparse basis functions to recover spectral reflectance from a single RGB image. Unlike the conventional spectral recovery methods, the sparse basis functions of the proposed method are an optimal solution for obtaining the information needed to consider the characteristics of the RGB response values. The novelty of the proposed method is to obtain the optimized sparse basis functions considering the influence of the spectral reflectance dataset and the sample’s colorimetric characteristic, which meets the minimum of the spectral recovery error. Spectral and colorimetric accuracy of the proposed method is studied to verify the effectiveness compared to existing methods. As the spectral recovery results show, the proposed method is extremely effective in recovering spectral reflectance.

This publication has 23 references indexed in Scilit: