An efficient calibration method for multi-spectral imaging

This paper presents an efficient calibration method for multi-spectral imaging. It is based on Digital Micro-Mirror Device and Hadamard Transform Spectral Imaging (DMD-HTSI), but can also be used for other DMD-based multispectral imaging methods. The key of this method is the location matching between the DMD coding pattern and the CCD image. The image matching determines the physical location of the spectral pixels in the DMD plane for a given CCD pixel. The encoding matrix can be derived based on linear dispersion, and the recalibration is not required when the coding pattern changes. Moreover, the grayscale encoding matrix is used to describe the interlaced pixels instead of the conventional Hadamard 0/1 matrix. The experimental results show that the reconstructed spectrum matches that measured by a commercial spectrometer very well. Additionally, the spatial quality of reconstructed multi-spectral images using the grayscale matrix is much better than that of the conventional Hadamard 0/1 matrix.