Compressive spectral imaging system for soil classification with three-dimensional convolutional neural network

• DOI: 10.1364/oe.27.023029
• Publication Date: 2019-07-29T16:13:41Z
• Authors: Yue Yu; Tingfa Xu; Ziyi Shen; Yuhan Zhang; Xi Wang
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 0401 agriculture, forestry, and fisheries; 04 agricultural and veterinary sciences; 02 engineering and technology

Compressive spectral imaging systems have promising applications in the field of object classification. However, for soil classification problem, conventional methods addressing this specific task often fail to produce satisfying results due tradeoff between invariance and discrepancy each soil. In paper, we explore a liquid crystal tunable filters (LCTF)-based system propose three-dimensional convolutional neural network (3D-CNN) We first obtain set compressive measurements via low spatial resolution detector, hyperspectral images are reconstructed with improved as well domains by sensing (CS) method. Furthermore, different from previous spectral-based restricted extract features type independently, on account potential property individual solid, our method proposes apply principal component analysis(PCA) achieve dimensionality reduction domain. Then, differential perception model flexible feature extraction, finally introduce 3D-CNN framework solve multi-soil problem. Experimental demonstrate that algorithm not only is able accelerate ability discriminability but also performs against methods.