Image Super Resolution Reconstruction Based On Sparse Representation

Image super-resolution reconstruction is an important image processing technique,which can improve the resolution of images by overcoming the inherent resolution limitation of the image-forming system,in the precondition of not changing the image-forming system.Image super-resolution reconstruction algorithms can be classified into single image super-resolution reconstruction and multiple images super-resolution reconstruction,according to the different low resolution images.This paper focus on single image super-resolution reconstruction algorithm based on sparse representation and multiple images super-resolution reconstruction based on regularization.The multiple images super-resolution reconstruction algorithms based on regularization usually build a single prior image model,it is difficult to keep the image edge information while suppressing the noise.In order to solve this problem,this paper propose a super-resolution algorithm based on adaptive regions of images,which using multiple low-resolution images.This algorithm classified the image into smooth blocks and non-smooth blocks based on gradient information,and then choose the prior model of image according to the different image regions,so that the reconstruction result has clear edge.Simulation results show that,compared with traditional methods,the PSNR of this paper's result increased by 0.07 to 0.26 dB,and RMSE decreased by 0.15 on average.To solve the high complexity problem of dictionary building and the low efficiency of image super-resolution algorithm based on sparse representation.This paper proposed a single image super-resolution algorithm based on sparse adaptive dictionary,which can choose dictionaries by the regional characteristics of the image.This algorithm calculated the structure information of images and analysed the regional information using it firstly,then choose the dictionary with different sizes.The simulation results show that this method can reduce the complexity of the algorithm on the basis of guaranteeing image quality compared with other methods.To solve the complexity problem of sparse reconstruction process,according to human visual characteristic,this paper proposed a super-resolution algorithm for color images based on ridge regression.This paper select the image blocks that human was interested in as the training samples using the local variance,then build the high resolution dictionary and the low resolution dictionary.Last calculate the projection matrix to reduce the complexity of sparse reconstruction.The simulation results show that the reconstruction images have good visual effect,and not only solve the fuzzy problem of the high resolution image,but also reduce the complexity of the algorithm.The algorithms proposed in this paper can be applied to image processing and other fields,and can also guide the hardware implementation of algorithms,which has great theoretical importance and practical value.