Study Of Remote Sensing Images Denoising Methods Based On Quantum Superposition Principle

Remote sensing images are disturbed by different kind of noises in process of receiving,coding and transmission,which caused bad influence on the further use of remote sensing images.Compared with other images,remote sensing images have richer edge and contour features,so there is a higher demand for remote sensing image denoising.Quantum mechanics is considered to be one of the two basic pillars of modern physics and it is also one of the most important achievements of physics in the 20 th century,it explains the microscopic phenomena that classical theory cannot explain.With the development of quantum mechanics,quantum signal processing theory has been proposed based on the mathematical framework and it be applied in image processing.The main tasks include:(1)Three kinds of noise types,namely additive noise,multiplicative noise and quantization noise are introduced.Four noise models are also introduced by Probability density function(PDF).The classical image denoising methods in frequency domain and space domain are introduced separately.Three classical wavelet denoising algorithms are discussed,including wavelet decomposition and reconstruction denoising,nonlinear wavelet transform threshold denoising and wavelet transform modulus maxima method denoising.(2)A remote sensing image denoising algorithm which combined quantum superposition principle with the dual-tree complex wavelet transform be proposed.First it carried out the dual-tree complex wavelet transform of the remote sensing image containing noise and established a superposition model of the signal and noise.Then according to the correlation of the signals in the image and it be considered the good directional analysis ability of the dual-tree complex wavelet transform,this paper used muti-quantum system measures contraction principle and quantum superposition principle to adaptive shrinking to fixed templates to estimate signal variance.The final comparison experiments show that the proposed algorithm can obtain higher peak signal-to-noise ratio and lower mean variance than the comparison algorithm,the algorithm in this paper is proved to be effective.(3)The remote sensing image processing software was designed based on all algorithm in this paper.Matlab and C++ hybrid programming be used in this software and Matlab engine and DLL dynamic link library in two ways to execute the algorithm,the algorithm result was displayed in the GUI interface of Qt.The designed software can be used in applications.