Denoising Of TASI Data Based On Compressed Sensing Method Study

This paper is supported by the project of "research and application demonstration of thermal infrared hyperspectral mineralization alteration mineral extraction". Denoising of hyperspectral data has been a hot issue in recent years.Commonly used remote sensing image denoising methods often use noise and information in the frequency domain of the difference distribution as a theoretical basis for denoising,But in fact, noise and information are distributed in the whole frequency domain, and it is difficult to realize the complete separation of information and noise.In this paper, we mainly deal with the denoising of hyperspectral TASI data.Compressed sensing technology is introduced in the denoising method.The details of the image and the effective information can be expressed in a suitable over complete dictionary(approximate).Noise due to its randomness in the dictionary does not have the sparsity of the method of denoising.Therefore, an adaptive dictionary learning method for TASI data is studied.The main work of this paper is as follows:1)The characteristics of hyperspectral TASI data are analyzed systematically,The noise sources and distribution of TASI data are studied,It is concluded that the Gauss white noise is used as the noise of the discussion.The additive noise model is used as the noise model.The noise distribution of the 32 bands of TASI data is evaluated systematically.It is concluded that the TASI data noise is mainly concentrated in the following three bands(30, 31, and 32 bands).2)Studying the theory of compressive sensing,The common image denoising method and the compression sensing denoising method are used to simulate the single band TASI data with the addition of Gauss white noise.The denoising results of the two methods are compared and analyzed.From the subjective visual and quantitative evaluation,Compressed sensing denoising method is better than the common image denoising method.3)For TASI data,An adaptive image denoising method based on compressive sensing theory is proposed.The dictionary can use the image itself information to carry out the dictionary training,Overcome the single structure of the fixed dictionary,Using adaptive Dictionary(Trained) to TASI data denoising,To compare the denoising results of the fixed Dictionary(DCT Dictionary) and the global dictionary(Global Dictionary),Adaptive dictionary denoising method is better than the other two kinds of dictionary denoising method.