Compressive sensing based image processing method and device

Abstract

The invention provides a compressive sensing based image processing method and device. The compressive sensing based image processing method comprises the steps of S1, performing wavelet transform and Gaussian measurement matrix processing on received signals of a two-dimensional image to obtain measurement vectors of all columns of the two-dimensional image and a sensing matrix [Theta]; S2, reconstructing an original signal by using a sparsity adaptive compressive sampling matching pursuit algorithm based on the measurement vector y of each column of the two-dimensional image and the sensing matrix [Theta], wherein regularization processing is performed on atom selection, variable step processing is performed on the iteration process, and a sparse approximation signal x^ of the original signal is acquired; S3, reconstructing the original two-dimensional image based on the sparse approximation signals of all columns of the two-dimensional image. Compared with the prior art, the compressive sensing based image processing method is higher in efficiency and further reduces the number of iterations based on the variable step processing, thereby being capable of acquiring the most approximate sparsity signal, and solving problems of long time consumption of signal reconstruction and inaccurate sparsity estimation.

Description

technical field [0001] The present invention relates to the field of image processing, and more specifically, to an image processing method and device based on compressed sensing. Background technique [0002] In the process of image transmission, usually the image needs to be sampled before transmission, and then transmitted. The receiver needs to reconstruct the original image based on the sampled data. [0003] The traditional Nyquist theory shows that the original signal can only be accurately reconstructed when the sampling frequency is at least twice the signal bandwidth. The disadvantage of this method is that the sampling rate and complexity are too high, and a lot of redundant information will be discarded after sampling, resulting in a waste of resources. Aiming at the above shortcomings, a new signal processing theory, namely compressed sensing theory, was proposed in 2006. For many practical signals, such as image signals, when represented by some appropriate ...

AI Technical Summary

Problems solved by technology

The above reconstruction algorithms have their own shortcomings, or the reconstruction accuracy will be affected when the dilution estimation is inaccurate, or the reconstruction will take a long time to affect the user experience, etc., and ultimately affect the reconstruction quality

Images