Nonconvex compressed sensing image reconstruction method based on local similarity and local selection

Abstract

The invention discloses a nonconvex compressed sensing image reconstruction method based on local similarity and local selection. The method comprises the following steps: 1) carrying out observation and reception after an image is partitioned; 2)utilizing a local growth method to carry out clustering on observation vectors of all the image blocks; 3) carrying out population initialization on the image block corresponding to each kind of observation vector according to the scheme that the polyatom direction and monatom direction coexist; 4) utilizing an improved genetic algorithm to carry out crossing, variation and selection operation based on a local selection mechanism on the populations obtained in the step 3), reconstituting corresponding image blocks and obtaining optimal atom combinations; 5) utilizing a clone selection optimization algorithm to study the optimal atom combinations on the aspects of dimension and displacement; and 6) piecing the image blocks obtained in the step 5) together in sequence to obtain a complete reconstructed image, and outputting the complete reconstructed image. The reconstructed image is good in visual effect and high in peak signal to noise ratio, and can be used for nonconvex compressed sensing reconstruction of image signals under the condition of low sampling rate.

Description

technical field

[0001] The invention belongs to the technical field of image processing, and further relates to image reconstruction, in particular to a non-convex compressive sensing image reconstruction method based on local similarity and local selection. Background technique

[0002] In the field of image reconstruction technology, a new data acquisition theory—compressed sensing theory is a major change in the field of information processing in recent years. The theory states that the signal can be sampled at low speed and with a small number of samples, and can be accurately reconstructed, which greatly reduces the device storage limit and computational complexity. Compressed sensing has become a research hotspot in the academic circles, and has been continuously applied in the fields of compressed imaging systems and biosensing. Compressed sensing technology mainly involves the following three aspects: signal sparse representation, observation matrix design and signa...

Examples

Embodiment 1

[0064] Embodiment 1, with reference to figure 1 A detailed description.

[0065] The invention is a non-convex compressive sensing image reconstruction method based on local similarity and local selection. The method can perform low-speed sampling and a small amount of sampling on the image signal, and then reconstruct the image accurately, which greatly reduces the storage limit and calculation of the device. The specific implementation steps are as follows:

[0066] Step (1), observe and receive the original image after segmentation.

[0067] Input the original image and divide it into 16*16 non-overlapping blocks, use the random Gaussian observation matrix Φ to observe each block to obtain the measurement vector y, the sending end sends the observation matrix Φ and the measurement vector y of each block, and the receiving end receives it;

[0068] In the present embodiment, the image of 512 * 512 is divided into image blocks of 16 * 16 to obtain 1024 image blocks; all im...

Embodiment 2

[0075] Embodiment 2, in conjunction with attached figure 1 -6 Description.

[0076] On the basis of Example 1, the step (2) uses the local similarity of the standard deviation of the observed vectors to cluster the observed vectors of all image blocks using a local growth method, specifically including the following steps:

[0077] 2.1) Calculate the standard deviation of each observation vector.

[0078] 2.2) Set a clustering mark for all image blocks, initially all marks are 0, where mark 0 means not clustered, and mark 1 means included in a certain category.

[0079] 2.3) Starting from the first image block, perform the following operations on each image block in turn: if the image block clustering mark is 1, no operation is performed; if the mark is 0, perform the Mth i clustering of classes.

[0080] Carry out the Mth i The specific steps of class clustering are as follows:

[0081] In the first step, the current image block i is used as the seed image block, and the...

technical field

[0001] The invention belongs to the technical field of image processing, and further relates to image reconstruction, in particular to a non-convex compressive sensing image reconstruction method based on local similarity and local selection. Background technique

[0002] In the field of image reconstruction technology, a new data acquisition theory—compressed sensing theory is a major change in the field of information processing in recent years. The theory states that the signal can be sampled at low speed and with a small number of samples, and can be accurately reconstructed, which greatly reduces the device storage limit and computational complexity. Compressed sensing has become a research hotspot in the academic circles, and has been continuously applied in the fields of compressed imaging systems and biosensing. Compressed sensing technology mainly involves the following three aspects: signal sparse representation, observation matrix design and signa...

Examples

Embodiment 1

[0064] Embodiment 1, with reference to figure 1 A detailed description.

[0065] The invention is a non-convex compressive sensing image reconstruction method based on local similarity and local selection. The method can perform low-speed sampling and a small amount of sampling on the image signal, and then reconstruct the image accurately, which greatly reduces the storage limit and calculation of the device. The specific implementation steps are as follows:

[0066] Step (1), observe and receive the original image after segmentation.

[0067] Input the original image and divide it into 16*16 non-overlapping blocks, use the random Gaussian observation matrix Φ to observe each block to obtain the measurement vector y, the sending end sends the observation matrix Φ and the measurement vector y of each block, and the receiving end receives it;

[0068] In the present embodiment, the image of 512 * 512 is divided into image blocks of 16 * 16 to obtain 1024 image blocks; all im...

Embodiment 2

[0075] Embodiment 2, in conjunction with attached figure 1 -6 Description.

[0076] On the basis of Example 1, the step (2) uses the local similarity of the standard deviation of the observed vectors to cluster the observed vectors of all image blocks using a local growth method, specifically including the following steps:

[0077] 2.1) Calculate the standard deviation of each observation vector.

[0078] 2.2) Set a clustering mark for all image blocks, initially all marks are 0, where mark 0 means not clustered, and mark 1 means included in a certain category.

[0079] 2.3) Starting from the first image block, perform the following operations on each image block in turn: if the image block clustering mark is 1, no operation is performed; if the mark is 0, perform the Mth i clustering of classes.

[0080] Carry out the Mth i The specific steps of class clustering are as follows:

[0081] In the first step, the current image block i is used as the seed image block, and the...

Images