Spectral reconstruction with dictionary learning based on spatially modulated photonic crystal

Conventional spectral analysis systems with complex optical components, moving parts, and long path lengths are usually costly and bulky. As the demand of reduced size, cost, and power consumption is growing for different applications of spectral analysis, spectral reconstructive methods attracts more and more attention. However the key element of the spectral-to-spatial mapping device usually needs careful engineering and precisely fabricated. Here we propose a reconstructive system based on a spatially modulated photonic crystal made of UV-curable polymer working as the mapping device, with the advantages of low-cost, and easy-to-fabricate, and a spectral reconstruction algorithm based on Orthogonal Matching Pursuit (OMP) and dictionary learning for spectral reconstruction. Experimental results demonstrate the capability of the spectral reconstruction of the system in visible range.