Study Of The Ghost Iamging Signal Processing Methods Based On The Compressive Sensing

As a new emerging image processing method, correlation imaging method also called ghost imaging, achieved the separation between the image detection and imaging process, so that it is possible to reconstruct the objects in more complex environments. But the image reconstruction operation needs to deal with a large number of sampled data, which means that the image reconstruction requires a much longer measurement time, and which limits the real-time image reconstruction of the associated imaging system, and severely limits the extensive application of correlation imaging technology. Thus, we proposed to take use of the compressive sensing technology to take place of the previous traditional signal acquisition and processing technology and applied to the correlation of the target image reconstruction.Compressive sensing technology makes up for the shortcomings of the traditional signal acquisition and processing technology that requires a large number of sampled data. Compared with the traditional Nyquist sampling signal processing technology , compressive sensing technology can reconstruct quickly and accurately the original signal quickly and accurately with more less sampled data, Thus reducing the time required for the information acquisition and compression process. It is considered as a new signal acquisition and reconstruction technique to break the Nyquist sampling theorem. This paper mainly focuses on the application of compression sensing in correlation imaging system data processing technology. The main works are as follows:(1): The basic theory of correlation imaging and compression perception is introduced. Secondly, the existing problems of the second-order correlation imaging technology are analyzed. The existing compression sensing technology is analyzed and analyzed, and the compression sensing technique is applied to the target image The reconstruction speed of the target image and the quality of the reconstructed image are improved.(2): The influence of different measurement matrices and sparse matrices on the restoration of ghost imaging target images is measured.The influence of different types of measurement matrices on the accuracy of signal restoration is analyzed by theoretical analysis and experimental simulation, and the sparse methods of different signals are compared. The effect of the sparsity effect of the signal on the restoration image quality of the sparse restoration algorithm is analyzed. Aiming at the light field realized in the experiment, the restoration effect of the measurement matrix for different signals is simulated under the condition of Gaussian distribution matrix, Toeplitz matrix and Hadamard matrix. In this paper,wavelet transform, sparse matrix transform and other methods are applied to sparse compression of signals, and different compression perceptual recovery algorithms are used to simulate the different signal sparse methods. Appropriate methods are selected to recover the experimental data.(3): Discrete cosine transform matrix (DCT) and Haar tansform matrix (Haar) are used as the prior sparse matrix for the target image, and the object images are reconstructed by applying the compressed sensing recovery algorithms which make use of the sparse matrix to sparsify the original signal to get sparse signals. Which is applied to image compression recovery algorithm. In order to reduce the influence of noise on image reconsruction quality and make full use of the sparse character of the differential derivative image of the experimental target image, the total variation regularization method was proposed to apply to the image compressive sensing and recovery. And the augmented Lagrangian Method based on the total variation regularization was introduced ans implemented as a efficient compressed sensing recovery algorithm for the reconstruction of the target images with sparse Image derivatives signal,which avoids the requirement of the method based on the sparsity matrix to know the a priori information of the sparse matrix of the target image in advance, not only improved the memory efficiency, but also greatly improved the quality of the rreconstructed image.