A New Hyperspectral Images Filtering Algorithm Based on Non-linear Wavelet Analyze
一种基于非线性增益小波滤波的高光谱影像去噪技术研究

A non-linear strength wavelet filtering algorithm is proposed to remove noise from the hyper spectral images. First, the spectral curve is decomposed into high frequency coefficient and low frequency coefficient with wavelet transformation. Deviation is calculated with low frequency parts to determine the threshold of noise and detail feature. A non-linear filtering algorithm is proposed to deal with the noise and detail feature of high frequency coefficient with the deviation threshold of low frequency coefficient. Comparing with the result of mean filter smoothing and least squares smoothing, the non-linear strength wavelet filtering algorithm (nLWF) can obtain better effect in noise remove and detail feature conversation.