Image Super-resolution Reconstruction Based On Compressed Sensing

With the rapid development of digital image processing technology, the requirement of image quality becomes higher and higher. By improving the hardware device to improve the image resolution by a number of restrictions on the cost and technology is difficult to achieve, through the software to improve the image resolution to improve the image quality is an important method. So, the image super resolution reconstruction technology has been rapidly developed, and become the research focus of image processing technology. Image super resolution reconstruction technique is a process of reconstructing a high resolution image through one or more sequences of low resolution images. Image super resolution reconstruction technology can be used in medical image, video surveillance, military remote sensing and high definition video and other fields.In this paper, the development and the principle of image super resolution reconstruction technology are briefly introduced, which shows that the image super resolution reconstruction is a solution to the inverse problem. The solving process is the process of solving the underdetermined system of equations and the final result is to obtain the optimal solution according to the limited conditions. Image super-resolution reconstruction algorithm can be divided into three categories which are based on the interpolation of image super-resolution reconstruction, based on the reconstruction of super-resolution reconstruction and learning based super-resolution reconstruction. At present, the research of image super resolution reconstruction based on learning is a hot topic. Reconstruction based image super resolution reconstruction algorithm is used for the reconstruction of low resolution image sequences, which requires more restrictions on the sequence images, which makes it difficult to obtain the sequence images. Learning based image super resolution reconstruction for a low resolution image, the operation is simple and intuitive. Based on the learning of the image super resolution reconstruction algorithm is based on the assumption that the similar low resolution characteristics and similar high resolution characteristics are corresponding between different images. The corresponding high-resolution and low resolution images are stored in the image sample library, and the low resolution image to be reconstructed is input, and the high resolution images can be output by learning the reconstructed image. Super resolution reconstruction based on sparse domain is an important research field in the based on learning super resolution reconstructionThe compressed sensing theory is fundamentally different from the traditional Nyquis sampling theorem. The compressed sensing theory for the original signal sampling rate is far lower than the sampling rate of Nyquist's theorem. Compressed sensing theory relies on the prior knowledge of the observed signal and the image, and achieves the aim of recovering the signal and image by using the sparse expansion of the appropriate basis. Compressed sensing theory mainly includes signal sparse decomposition, the signal observation matrix and the signal reconstructionImage super-resolution reconstruction based on compressed sensing technology is the first to design a high resolution image training database. Then, the image of the high resolution image training database is degraded, and the corresponding low resolution image database is obtained. The images in the high and low resolution image database are divided into blocks, and the high frequency information of the image blocks is extracted. After the KSVD algorithm is used to train the dictionary, the sparse representation coefficients of the low resolution image blocks in the low resolution dictionary are obtained. High resolution image blocks are reconstructed by using high resolution dictionary and sparse representation coefficients. A high-resolution image is finally reconstructed from the high resolution image blocks.Image super resolution reconstruction algorithm based on compressed sensing will be combined with the theory of super resolution reconstruction algorithm based on sparse domain and compressed sensing theory, and it has improved the super-resolution reconstruction algorithm proposed by Yang. It improves the quality of the reconstructed image and the running time of the algorithm.