| 摘要 |
A software solution that increases the Signal to Noise Ratios (SNRs) of hyperspectral systems for a fixed spectral resolution is proposed. This technique solves the problem tha t sensors with higher spectral resolution suffer from poor SNR. The proposed technology includes a n averaging process and a deconvolution process. For any given hyperspectral signature with high spectral resolution, a filter with wider band than the hyperspectral sensors is applied to get the convoluted signal to increase SNR. The trade-off is the reduction of the spectral resolution. A proper deconvolution process will then be applied to the averaged signal to reconstruct the original signature. The final signal can has nearly restored high spectral resolution while maintaining th e high SNR of the averaged signal. This technique could be implemented in the existing hyperspectral systems so that the user could achieve adequate SNR at any resolution independent of optical design. Based on this idea, new hyperspectral sensors could be designed with wide and overlapping bands to have both high spectral resolution and high SNR.
|
