Spectral reflectance estimation based on adaptive selection of training samples from multichannel responses
Utilizing multispectral imaging, the surface spectral reflectance of non-fluorescent object can be estimated from the corresponding multichannel responses. The estimation accuracy is significantly impacted by the training samples used to calculate the transform matrix from multichannel responses to reflectance. In this study, the training samples were adaptively selected according to the correlation of the response vectors between test sample and training samples. The adaptive spectral estimation results were compared to those of the conventional approach by different channel numbers. It is verified that the adaptive spectral estimation outperforms the conventional one in both spectral and colorimetric accuracy especially when fewer channels are adopted.
multispectral imaging spectral reflectance estimation correlation of response vectors adaptive training samples
Peng Xu Haisong Xu
State Key Laboratory of Modern Optical Instrumentation, College of Optical Science and Engineering, Zhejiang University, Hangzhou 310027, China
国际会议
The 3rd Conference of Asia Color Association(ACA2016 China)第三届亚洲颜色学术会议
江苏 常熟
英文
97-100
2016-05-21（万方平台首次上网日期，不代表论文的发表时间）