高光谱图像的光谱域小波去噪应用研究

Imaging spectroradiometers are highly susceptible to noise. Accurately quantitative processing with more high quality is obligatory before any derivative analysis, especially for precise agriculture application. Aiming at the Pushbroom Imaging Spectrometer (PIS) developed by us, a wavelet-based threshold denoising method was developed for its hyperspectral imagery data. And its capacity was evaluated through comparing with other popular denoising methods in pixel scale and in region scale. Furthermore, the method was validated in chlorophyll concentration retrieval application based on its red-edge extraction. The result revealed that the determination coefficient R2 of chlorophyll concentration retrieval model was improved from 0.586 to 0.811. It showed that the proposed denoising method allowed efficient denoising while maintaining image quality, and presented significant advantages over conventional denoising methods.