CN110827215A - ERT image reconstruction artifact removing method based on fuzzy clustering - Google Patents

Abstract

The present invention relates to an ERT image reconstruction artifact removal method based on fuzzy clustering, comprising the following steps: 1) using at least two image reconstruction algorithms to respectively calculate the grayscale vector of the measured field; 2) 1) medium grayscale The vectors are combined to form a gray matrix as the feature vector in the clustering algorithm; 3) The fuzzy clustering algorithm is applied to the gray matrix, and the gray matrix is clustered; 4) According to the statistical characteristics of the clusters Find out the clusters representing "target", "background" and "artifact"respectively; 5) classify the cluster representing the artifact into the cluster representing the background.

Description

An Artifact Removal Method for ERT Image Reconstruction Based on Fuzzy Clustering

technical field

The invention belongs to the field of electrical tomography image reconstruction algorithms, and in particular relates to an ERT image reconstruction artifact removal method based on fuzzy clustering.

Background technique

Electrical tomography (ET) is a technology for reconstructing the distribution of electrical properties inside the object field under test. The hardware equipment used in this technology is relatively simple in structure and low in cost. sexual characteristics. The technology has broad application prospects in the fields of medical clinical monitoring and industrial measurement.

In electrical resistance tomography (ERT), current excitation is applied to the measured object field with array electrodes, boundary voltage data is measured, and then the conductivity distribution in the measured field is reconstructed. Taking a 16-electrode resistance tomography sensor as an example, the adjacent current excitation and adjacent voltage measurement methods are used. At each excitation, with the excitation electrodes removed, a total of 13 measurements were made. For example, take the No. 1 and No. 2 electrodes as excitation electrodes, and measure the voltage signals on the 3, 4, 4, 5, 5, 6, ..., 15, 16 electrode pairs. There are 16 excitations in total, that is, 13×16=208 measurements. In order to take into account both the temporal resolution and the spatial resolution, the measured sensitive field is generally divided into 812 pixels. The reconstruction problem of resistance tomography image is to solve the gray value of these 812 pixels.

Due to the "soft field" effect of ERT and the ill-posed nature of the inverse problem, the spatial resolution of the reconstructed image is poor. The traditional linear back-projection algorithm (LBP) is to back-project the voltage data according to the relationship between the boundary voltage value and the projection area of the measured field. However, due to the limitation of hardware equipment, the back-projection data is far less than the grayscale that needs to be solved. If the degree value is high, the reconstructed image has large artifacts, the boundary between the target and the background is blurred, and the spatial resolution is poor.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art, and to propose a method for removing artifacts in ERT image reconstruction based on fuzzy clustering, which can effectively reduce the impact of artifacts on image resolution. All pixels of the reconstructed image can be divided into 3 categories: object, background and artifact. Among them, artifacts greatly affect the resolution of imaging, and effective processing of artifacts can improve the quality of reconstructed images to a certain extent. Considering that the result obtained by the above algorithm is the gray value, that is, the numerical data, therefore, the gray value can be divided by the clustering algorithm, and then the gray value can be further processed according to the division result, so as to obtain the ideal value. Imaging effect.

The technical scheme of the present invention is as follows:

A method for removing artifacts from ERT image reconstruction based on fuzzy clustering, comprising the following steps:

1) Utilize at least two kinds of image reconstruction algorithms to calculate the gray level vector of the measured field respectively;

2) Combine the grayscale vectors in 1) to form a grayscale matrix as the feature vector in the clustering algorithm;

3) Apply the fuzzy clustering algorithm to the grayscale matrix, and perform cluster analysis on the grayscale matrix;

4) Find out the clusters representing "target", "background" and "artifact" respectively according to the statistical characteristics exhibited by the clusters;

5) The clusters representing artifacts are classified into clusters representing the background.

The present invention has the following advantages due to taking the above technical solutions:

1. Compared with the traditional electrical tomography algorithm, this algorithm integrates several classical imaging algorithms to complement each other's advantages. On this basis, the gray value is further clustered and analyzed to reduce the false value. The influence of the shadow can make the boundary between the target and the background clearer and improve the spatial resolution of imaging.

2. The clustering algorithm adopts the fuzzy clustering algorithm instead of the traditional hard division algorithm, which makes the target edge smoother and closer to the actual edge.

Description of drawings

1 is a model diagram of an original object field under test in an embodiment of the present invention;

Fig. 2 is the result of image reconstruction using LBP algorithm to the simulation model shown in Fig. 1;

FIG. 3 is the result of image reconstruction using the algorithm of the present invention for the simulation model shown in FIG. 1 .

Detailed ways

The purpose of the present invention is to overcome the problem of low spatial resolution of the existing image reconstruction algorithms, and propose a method for removing artifacts from ERT image reconstruction based on fuzzy clustering. The object of fuzzy clustering is a set of data, and the data is divided by extracting the feature vector of the data, so that the intra-class difference is small and the inter-class difference is large. By clustering the grayscale matrix, the influence of artifacts on image resolution is reduced.

A method for removing artifacts from ERT image reconstruction based on fuzzy clustering of the present invention will be described in detail below with reference to the accompanying drawings and embodiments. Specific steps are as follows:

1) Figure 1 shows the flow pattern with a discrete target medium. For the measured object field shown in Figure 1, if the regularization (TK) algorithm and the LBP algorithm are used to calculate the grayscale vector, the grayscale value of the reconstructed image can be It is shown in formula (1).

_GT = {g _t1 ,g _t2 ,...,g _t812 }, _{GL ={g l1 ,} g _l2 ,...,g _l812 } (1)

Among them, _GT represents the grayscale vector obtained by the TK algorithm, and _GL represents the grayscale vector obtained by the LBP algorithm.

The image reconstructed by the LBP algorithm is shown in Figure 2. The area with a small circular area at the top represents the target, the area with the largest area at the bottom is the background area, and the gray-white area in between is the artifact area.

2) Combine the grayscale vectors in 1) to form a grayscale matrix, which is used as a feature vector in the clustering, that is, combine the grayscale vectors G _T and G _L to form a gray matrix, and calculate the gray mean vector As shown in formula (2):

3) Use fuzzy clustering algorithm to perform cluster analysis on the gray matrix in 2). The fuzzy clustering algorithm requires a given number of clusters. From the meaning of the inverse problem itself, the number of clusters can be set to 3, that is, the data can be divided into three categories: target G _o , background G _b and artifact G _t , namely

4) Calculate the statistical characteristics of each cluster. The three clusters in step 3) have different statistical characteristics: the one with the largest mean is the target, on the contrary, the one with the smallest mean is the background, and the one with the largest variance represents the artifact. According to the above characteristics, the meaning of each cluster in the clustering result can be clarified.

中像素点的灰度值赋值为中像素点的平均灰度值，记为

即最终的灰度向量为：5) Classify the cluster representing the artifact into the cluster representing the background, namely

The gray value of the middle pixel is assigned as The average gray value of the middle pixel, denoted as

That is, the final grayscale vector is:

进行图像反演，如图3所示。图3中伪影区域被有效地减小，更加接近原始物场分布。6) Using formula (4) in

Perform image inversion, as shown in Figure 3. The artifact area in Fig. 3 is effectively reduced, closer to the original object field distribution.

Claims (1)

1. a kind of ERT image reconstruction artifact removal method based on fuzzy clustering, comprises the following steps: 1) Use at least two image reconstruction algorithms to calculate the grayscale vector of the measured field respectively. 2) Combine the grayscale vectors in 1) to form a grayscale matrix as the feature vector in the clustering algorithm; 3) Apply the fuzzy clustering algorithm to the grayscale matrix, and perform cluster analysis on the grayscale matrix; 4) Find out the clusters representing "target", "background" and "artifact" respectively according to the statistical characteristics exhibited by the clusters; 5) The clusters representing artifacts are classified into clusters representing the background.

Images