Image reconstruction method based on self-adaptive directional diffusion thought and electronic equipment

The image reconstruction method based on the adaptive directional diffusion concept solves the problems of image quality and accuracy caused by incomplete data, and achieves high-quality image reconstruction under incomplete conditions.

CN119048633APending Publication Date: 2024-11-29CAPITAL NORMAL UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411539295.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-31
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Existing technologies suffer from poor image quality and low accuracy during image reconstruction due to incomplete data, which affects the judgment of image structural information.

Method used

An image reconstruction method based on adaptive directional diffusion is adopted. By acquiring the dataset and reconstruction geometric parameters, and combining the mathematical model of adaptive directional diffusion and the edge-preserving operator, candidate images are iteratively processed to improve image quality and accuracy.

Benefits of technology

It significantly improves the image quality and accuracy of incomplete image data or scan datasets, especially under conditions of CT scan data with limited or sparse angles, and can effectively remove artifacts while preserving image details.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048633A_ABST
    Figure CN119048633A_ABST
Patent Text Reader

Abstract

The invention provides an image reconstruction method based on a self-adaptive directional diffusion idea and electronic equipment, and is suitable for the fields of digital image processing, CT and magnetic resonance image reconstruction and the like. The method comprises the following steps: updating a current image # imgabs1 # to obtain an image # imgabs2 # by using image data or scanning data # imgabs0 # based on a data fidelity constraint condition, prior information of a to-be-reconstructed object and gradient information of the current image according to input parameters through three preset mathematical models based on an adaptive directional diffusion idea, and judging whether # imgabs3 # is less than or equal to epsilon or whether the upper limit of the number of iterations # imgabs4 # is reached; if not, continuing to iterate and update the image, otherwise, outputting the image; and for a plurality of output images obtained by the plurality of groups of input parameters, a final output image is determined through comparison. Particularly, for incomplete image data or incomplete scanning data sets, the image quality and accuracy can be remarkably improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical fields of digital image processing, CT and magnetic resonance image reconstruction, and in particular to an image reconstruction method and electronic equipment based on the idea of ​​adaptive directional diffusion. Background Art

[0002] With the rapid development and improvement of image processing, CT technology, and magnetic resonance technology, they are widely used in many fields such as industrial non-destructive testing and medical diagnosis. At present, relevant technologies propose that in order to obtain high-quality reconstructed images, complete data needs to be obtained. However, due to the influence of factors such as object structural characteristics, environment and equipment limitations when actually collecting data, the collected data is often incomplete (for example, only CT scan data of limited angles can be collected, data with relatively large noise, etc.). Therefore, when reconstructing CT, magnetic resonance and other images through traditional algorithms, very serious artifacts and blurs will be generated in the reconstructed image, resulting in poor quality of the reconstructed image and low reconstruction accuracy, which in turn affects the judgment of the image structure information. In the field of digital images, due to incomplete image data, the quality of the reconstructed image will also be affected, resulting in poor quality of the reconstructed image and low reconstruction accuracy, which in turn affects the judgment of the image structure information. Summary of the invention

[0003] In view of this, an object of the present invention is to provide an image reconstruction method and electronic device based on the concept of adaptive directional diffusion, which can significantly improve image quality and accuracy for incomplete image data or incomplete scan data sets.

[0004] In a first aspect, an embodiment of the present invention provides an image reconstruction method based on the concept of adaptive directional diffusion, the steps in the method include: Get the dataset , reconstruct the geometric parameter set , initialize the candidate image , determine the upper limit parameter of the number of iterations , Iteration termination threshold parameter , dynamic and static information parameters describing adaptive directional diffusion and the number of diffusion directions and mathematical model category information parameters; Iteratively process candidate images :Based on the candidate images and the prior information of the object to be reconstructed, based on the dynamic and static information parameters describing the adaptive directional diffusion and the number of diffusion directions And the mathematical model category information parameters, determine diffusion directions and the corresponding Diffusion coefficient , determine the edge preserving operator , combined with data fidelity constraints to determine the mathematical model of adaptive directional diffusion, and the candidate images Update to get :

[0005] in, , P is a known data set, is the corresponding system matrix; is the i-th diffusion direction The corresponding directional diffusion operator is, Indicates the diffusion direction The diffusion coefficient, , Represents the number of diffusion directions; since multiplying the target optimization function by a non-zero multiple does not change the solution of the optimization problem, there is no loss of generality Fixed to ; is an edge-preserving operator used to identify image edges, so that The edge of The edges remain consistent; Determine whether the difference between two adjacent iterative images is less than the iteration termination threshold ,Right now , or whether the upper limit of the number of iterations is reached If it is not satisfied, the iterative process continues; if it is satisfied, the image is output.

[0006] In one embodiment, the image reconstruction method based on the adaptive directional diffusion concept includes the following steps: , the mathematical model category information parameter describing the adaptive directional diffusion is the adaptive static directional diffusion linear differential equation model. At this time, the scheme is the preset scheme A, and the main contents include: In order to distinguish different solutions, it is useful to remember that the for , according to the prior information of the object to be reconstructed and the current image And the preset diffusion direction quantity information ,Sure static diffusion direction and The corresponding diffusion coefficient and edge preservation operator , combined with the data fidelity constraints, the image is modeled by the following preset adaptive static directional diffusion linear differential equation model. After updating, the image is obtained ,Right now:

[0007]

[0008] in, ,Pick ; is the image area, is the integral differential element; For along the direction The directional derivative of yes , whose components represent the direction cosines with respect to each coordinate axis, .

[0009] In one embodiment, the image reconstruction method based on the adaptive directional diffusion concept includes the following steps: , the mathematical model describing the adaptive directional diffusion category information parameter is the adaptive static directional diffusion Optimize the model. At this time, the solution is the preset solution B, which mainly includes: In order to distinguish different solutions, it is useful to remember that the for , according to the prior information of the object to be reconstructed and the current image And the number of preset diffusion directions ,Sure static diffusion direction, The corresponding diffusion coefficients and the edge preserving operator , combined with data fidelity constraints, through the following preset adaptive static directional diffusion Optimize the model for images After updating, the image is obtained ,Right now:

[0010]

[0011] in, ,Pick , , Represents image coordinates, which is a two-dimensional coordinate variable Or three-dimensional coordinate variables , is the image area, is the discrete gradient of u, is a unit vector whose components represent the direction cosines with respect to the coordinate axes. ; For each pixel, if the image is three-dimensional, the value of the image discrete gradient at any pixel is , if the image is two-dimensional, the value of the image discrete gradient at any pixel point is = , , , Along axis, Common discrete differential operators for the a-axis and z-axis, such as forward difference operator, backward difference operator and central difference operator; is a counting operator, Represented as an image Along the direction of norm, which means in a given direction Satisfaction The number of pixels; In one embodiment, the image reconstruction method based on the adaptive directional diffusion concept includes dynamic and static parameter information being dynamic, the number of diffusion directions being , the mathematical model category information parameter describing the adaptive directional diffusion is the adaptive dynamic directional diffusion differential equation model. At this time, the scheme is the preset scheme C, and the main contents include: In order to distinguish different solutions, it is useful to remember that the for , according to the prior information of the object to be reconstructed and the current image And the preset diffusion direction quantity information ,Sure Dynamic diffusion direction, The corresponding diffusion coefficients and the edge preserving operator , combined with the data fidelity constraints, the image is modeled by the following preset adaptive dynamic directional diffusion differential equations. After updating, the image is obtained ,Right now: First, based on the data fidelity term, the scanned geometric parameter set Related image reconstruction operators , the current image After updating, ,Right now:

[0012]

[0013] in, ,Pick Make The corresponding first-order variation is 2 , is the dynamic diffusion direction (with respect to the coordinate variable function); H is the Hesse matrix of u. When the image is two-dimensional, H= , when the image is three-dimensional, H= , through the above equation, to achieve along Adaptive diffusion.

[0014] In one embodiment, the image reconstruction method based on the adaptive directional diffusion concept includes: Update to get Adaptive directional diffusion constrained optimization model:

[0015] in, , transformed into an unconstrained optimization mathematical model, and obtained:

[0016]

[0017] in, is the data fidelity term, Q is the operator on the norm, is the weight coefficient of the data fidelity item; is the i-th diffusion direction The corresponding directional diffusion operator; is an edge-preserving operator used to identify image edges, so that The edge of Keep the edges consistent.

[0018] In one embodiment, the preset scheme A, the preset scheme B and the preset scheme C include: determining the dynamic and static information parameters of the adaptive directional diffusion, the related prior information of the object to be reconstructed, the current image gradient information and the preset diffusion direction quantity information. static or dynamic diffusion direction .

[0019] In one embodiment, the preset scheme A, the preset scheme B and the preset scheme C include: determining the dynamic and static information, the current image gradient information and the preset diffusion direction quantity information, The diffusion coefficient corresponding to the target diffusion direction .

[0020] In one embodiment, preset solution A, preset solution B and preset solution C include: is an edge-preserving operator used to identify image edges, so that The edge of The edge of the image is kept consistent; the edge preservation operator is determined according to the prior information of the object to be reconstructed or according to the difference between the adjacent pixel values ​​of each point in the current image. . In one embodiment, the preset scheme A, the preset scheme B and the preset scheme C further include adding a general regularization term R(u) during the iterative update process, for example, a TV regularization term, Regularization term, Regularization term; if the regularization term R(u) is not added, it is equivalent to R(u)=0.

[0021] In one embodiment, when multiple groups of data are input according to the data input by the user, an image is output according to each group of data, and a method for determining a final output image for the images outputted for each group of data includes: The final output image is determined by the prior information of the target image or by calculating the reference-free image index. In one embodiment, the method includes: when performing fine imaging, inputting at least three sets of parameters, wherein the parameters are selected so that preset scheme A, preset scheme B, and preset scheme C can each be executed at least once, and determining a final output image based on output images after execution of each set of parameters and based on prior information of a target image or by calculating a reference-free image index; When performing finer imaging, multiple sets of parameters may be input, and the parameters are selected so that two of the preset schemes A, B, and C are executed. The final output image is determined based on the output images after the execution of each set of parameters and based on the prior information of the target image or by calculating the index of the no-reference image. When performing rapid imaging, a set of parameters is input, and any preset scheme is selected from preset scheme A, preset scheme B, and preset scheme C to execute to determine the final output image.

[0022] In a second aspect, an embodiment of the present invention further provides an electronic device, including a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement any one of the methods provided in the first aspect.

[0023] In a third aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect. The embodiment of the present invention brings the following beneficial effects: An embodiment of the present invention provides an image reconstruction method and electronic device based on the concept of adaptive directional diffusion. The method uses image pixel information or a scan data set and an image reconstruction operator, and adds a regularization term known in the industry to process a candidate image. Then, according to preset prior information and gradient information of the image and preset diffusion direction quantity information, a target diffusion direction set and its corresponding diffusion coefficient set are determined. The image is updated through three preset mathematical models of adaptive directional diffusion concepts to improve the accuracy of the reconstructed image. In particular, for incomplete image data or incomplete scan data sets (for example, limited angle / sparse angle / low-dose CT scan data), the present invention can significantly improve image quality and accuracy.

[0024] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0026] Figure 1 A schematic diagram of the overall process of an image reconstruction method based on the idea of ​​adaptive directional diffusion provided by an embodiment of the present invention; Figure 2 A schematic diagram of a specific process of an image reconstruction method based on preset scheme A provided in an embodiment of the present invention; Figure 3 A schematic diagram of a specific flow chart of an image reconstruction method based on preset scheme B provided in an embodiment of the present invention; Figure 4 A schematic diagram of a specific process of an image reconstruction method based on preset scheme C provided in an embodiment of the present invention; Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] See also Figure 1 The flowchart of an image reconstruction method based on the idea of ​​adaptive directional diffusion is shown in FIG. 1 . The method mainly includes three alternative schemes: preset scheme A, preset scheme B and preset scheme C. Specifically: Step 1: Get the dataset , reconstruct the geometric parameter set , initialize the candidate image , determine the upper limit parameter of the number of iterations , Iteration termination threshold parameter , dynamic and static information parameters describing adaptive directional diffusion and the number of diffusion directions and mathematical model category information parameters; Step 2: Iteratively process candidate images :Based on the candidate images and the prior information of the object to be reconstructed, based on the dynamic and static information parameters describing the adaptive directional diffusion and the number of diffusion directions And the mathematical model category information parameters, determine diffusion directions and the corresponding Diffusion coefficient , and the edge preserving operator , combined with the data fidelity constraints, the following unconstrained optimization model of adaptive directional diffusion is determined, and the candidate images Update to get :

[0029] and

[0030] in, is the data fidelity term, Q is the operator on the norm, is the weight coefficient of the data fidelity item; is the i-th diffusion direction The corresponding directional diffusion operator; Represents the diffusion coefficient along each diffusion direction. Since multiplying the target optimization function by a non-zero multiple does not change the solution of the optimization problem, there is no loss of generality. Fixed to , , Indicates the number of diffusion directions; is an edge-preserving operator used to identify image edges, so that The edge of Keep the edges consistent.

[0031] Step 3: Determine whether the difference between two adjacent iterative images is less than the iteration termination threshold. ,Right now , or whether the upper limit of the number of iterations is reached If it is not satisfied, continue to iterate and process the candidate image , if satisfied, then output the image.

[0032] In one embodiment, in step 2, based on the candidate image and the prior information of the object to be reconstructed, based on the dynamic and static information parameters describing the adaptive directional diffusion and the number of diffusion directions As well as the mathematical model category information parameters, the method for determining the specific implementation plan is as follows: If the input dynamic and static information parameter of adaptive directional diffusion is static, and the mathematical model category information parameter is an adaptive static directional diffusion linear differential equation model, then the preset scheme A is executed; If the input dynamic and static information parameter of adaptive directional diffusion is static, and the mathematical model category information parameter is adaptive static directional diffusion Optimize the model and execute the preset plan B; If the input dynamic and static information parameters of the adaptive directional diffusion are dynamic, and the mathematical model category information parameters are the adaptive dynamic directional diffusion differential equation model, the preset scheme C is executed.

[0033] For preset plan A: for ,from Start iterating over the image , according to the prior information and image of the object to be reconstructed And the preset diffusion direction quantity information ,Sure static diffusion direction and The corresponding diffusion coefficient and edge indicator function are combined with the data fidelity term, and the image is modeled by the following preset adaptive static directional diffusion linear differential equation. After updating, the image is obtained ,Right now: ; ; in, is the image area, is the integral differential, is the data fidelity term, Q is the operator on the norm, and this preset scheme is as follows: For example, is the weight coefficient of the fidelity term, is an edge-preserving operator used to identify image edges, so that The edge of The edges are consistent. This preset solution is as follows: = For example, is the edge indicator function; For along the direction The directional derivative of yes , whose components represent the direction cosines with respect to each coordinate axis, .

[0034] That is, see Figure 2 A specific flow chart of an image reconstruction method based on an adaptive static directional diffusion differential equation model is shown, and the mathematical model involved in the preset scheme A is The alternating solution method is adopted, namely: First, based on the data fidelity term, the scanned geometric parameter set Related image reconstruction operators Set the current image After updating, ,Right now: ; Secondly, according to the prior information of the object to be reconstructed and the current image And the preset diffusion direction quantity information ,Sure static diffusion direction and The corresponding diffusion coefficient and edge indicator function are used to adjust the current image through the preset adaptive static directional diffusion linear differential equation model. After updating, the image is obtained ,Right now: ; Among them, the solution methods for the above functional minimization problem include at least the following three methods: 1. The Euler-Lagrange equation for the above functional minimization problem is derived as follows: ; in, For along the direction The second-order directional derivative of . Using the second-order central difference format to discretize it, if it is a two-dimensional case, we get a Point difference format, solve the The linear equations corresponding to the point difference format are obtained ; If it is a three-dimensional case, a 27-point difference format is obtained to solve the The linear equations corresponding to the point difference format are obtained

[0035] 2. The Euler-Lagrange equation for the above functional minimization problem is derived as follows: ; in, For along the direction By artificially introducing time t, the above Euler-Lagrange equation is transformed into a parabolic equation:

[0036] The parabolic equation is solved by time-step iteration using the fully explicit format, and we get ; 3. Minimizing the above functional ; Decompose into A sub-question, if you remember ,but The first sub-problem The sub-questions are: , ... , Above The first sub-problem The Euler-Lagrange equation corresponding to the subproblem is: ; Among them, the coordinate rotation of the above equation is performed so that a certain coordinate axis (with axis as an example) and direction coincide with each other, thus transforming the second-order directional derivative in the equation into The second-order derivative in the direction can be discretized by the second-order central difference of the one-dimensional case, so the corresponding linear equation system is a tridiagonal matrix linear equation system, so the pursuit method can be used for rapid solution. After the solution is completed, the coordinate axis can be restored to its original initial position, and the coordinates in the current coordinate system can be solved to obtain .

[0037] Further, for the preset plan B: for ,from Start iterating over the image , , according to the prior information and image of the object to be reconstructed And the preset diffusion direction quantity information ,Sure static diffusion direction and The corresponding diffusion coefficients are combined with the data fidelity term, and the following adaptive static directional diffusion is preset Optimize the model for images After updating, the image is obtained ,Right now: ; ; in, is the data fidelity term, Q is the operator on the norm, and this preset scheme is as follows: For example, is the weight coefficient of the fidelity term, is an edge-preserving operator used to identify image edges, so that The edges are consistent. This preset solution is as follows: = For example; , Represents the image coordinates, which is a two-dimensional coordinate variable Or three-dimensional coordinate variables , is the image area, is the discrete gradient of u, is a unit vector whose components represent the direction cosines with respect to the coordinate axes. ; For each pixel, if the image is three-dimensional, the value of the image discrete gradient at any pixel is , if the image is two-dimensional, the value of the image discrete gradient at any pixel point is = , , , Along axis, Common discrete differential operators for the a-axis and z-axis, such as forward difference operator, backward difference operator and central difference operator; is a counting operator, Represented as an image Along the direction of norm, which means in a given direction Satisfaction The number of pixels; That is, see Figure 3 The adaptive static directional diffusion Schematic diagram of the specific process of the image reconstruction method of the optimization model, and the mathematical model involved in the preset scheme B , an alternating solution method can be used. If it is a two-dimensional case, the process is as follows: First, based on the data fidelity term, the scanned geometric parameter set Related image reconstruction operators Set the current image After updating, ,Right now: ; Secondly, the solution is as follows in the direction The optimization problem is:

[0038] Then decompose the above functional minimization problem into A sub-question, if you remember ,remember for ,but The first sub-problem The sub-questions are:

[0039] Introducing auxiliary variables and , corresponding to and , while introducing parameters , then the optimization problem becomes:

[0040] in, , and Auxiliary variables and In the element The corresponding and .when When it is large enough, the above optimization problem is equivalent to the original optimization problem without introducing auxiliary variables. At this time, the model can be transformed into alternating minimization , and These two sub-questions: Subproblem 1: Calculate u

[0041] To solve this problem, we first perform a fast Fourier transform (FFT), and then diagonalize the derivative operator to speed up the solution:

[0042] in, is the FFT operator, for conjugation of; Sub-problem 2: Calculation v

[0043] for The pixels at each pixel and are estimated separately, so the above formula can be transformed into an optimization problem for each pixel:

[0044] Every pixel In the above formula,

[0045] By solving, we can get: when , =(0,0); in other cases, =( , ).

[0046] In the three-dimensional case, the process is as follows: First, based on the data fidelity term, the scanned geometric parameter set Related image reconstruction operators Set the current image After updating, ,Right now: ; Secondly, the solution is as follows in the direction The optimization problem is: .

[0047] Then decompose the above functional minimization problem into A sub-question, if you remember , for .but The first sub-problem The sub-questions are:

[0048] Introducing auxiliary variables , and w, corresponding to , and , while introducing parameters , then the optimization problem becomes:

[0049] in, , , and Auxiliary variables , and In the element The corresponding , and .when When it is large enough, the above optimization problem is equivalent to the original optimization problem without introducing auxiliary variables. At this time, the model can be transformed into alternating minimization , , w and These two sub-questions: Subproblem 1: Calculate u

[0050] To solve this problem, we first perform a fast Fourier transform (FFT), and then diagonalize the derivative operator to speed up the solution:

[0051] in, is the FFT operator, for conjugation of; Sub-problem 2: Calculation v and w

[0052] for The pixels at each pixel , and are estimated separately, so the above formula can be transformed into an optimization problem for each pixel:

[0053] Every pixel In the above formula, we have:

[0054] By solving, we can get: when hour, =(0,0,0); in other cases, =( , , ).

[0055] Further, for the preset solution C: for ,from Start iterating over the image ; Based on the prior information and image of the object to be reconstructed And the edge preserving operator , combined with the data fidelity term, the image is conditioned according to the following preset adaptive dynamic directional diffusion differential equation model After updating, the image is obtained ,Right now:

[0056]

[0057] in, ,Pick Make The corresponding first-order variation is 2 , is the dynamic diffusion direction (with respect to the coordinate variable function); H is the Hesse matrix of u. When the image is two-dimensional, H= When the image is three-dimensional, H = . is an edge-preserving operator used to identify image edges, so that The edge of The edges are consistent. This preset solution is as follows: = For example, is the edge indicator function; through the above equation, to achieve Adaptive diffusion.

[0058] That is, see Figure 4 A specific flow chart of an image reconstruction method based on an adaptive dynamic directional diffusion differential equation model is shown, and the mathematical model involved in the preset scheme C is The alternating solution method is adopted, namely: First, based on the data fidelity term, the scanned geometric parameter set Related image reconstruction operators Set the current image After updating, ,Right now: ; Secondly, according to the prior information of the object to be reconstructed and the current image Gradient information and the number of preset diffusion directions ,Sure Dynamic diffusion direction and The corresponding diffusion coefficient and edge indicator function are used to adjust the current image through the preset adaptive dynamic directional diffusion differential equation model. After updating, the image is obtained ,Right now:

[0059] Among them, the above functional minimization problem can be solved by solving the following Euler-Lagrange equation: , That is, solving the above adaptive dynamic directional diffusion differential equation gives .

[0060] Through the above equation, we can realize Adaptive diffusion. The dynamic diffusion directions in the preset scheme C For the coordinate variables For example, in the two-dimensional case, if we take two dynamic diffusion directions ( ), one method is to take is the unit gradient direction of u ,Pick is the unit tangent direction of u ,at this time: ; .

[0061] That is, see Figure 4 A specific flow chart of an image reconstruction method based on an adaptive dynamic directional diffusion differential equation model is shown, and the differential equation involved in the preset scheme C is There are two solution methods: 1. Direct explicit format iterative solution , The above equation can be solved by using Newton's iteration method after differential discretization; it can also be solved by introducing artificial variables , the following time-dependent differential equation is obtained, and the time-stepping technique is used to explicitly solve it. The differential equation corresponding to the explicit solution method of the time-stepping technique is:

[0062] in, is the boundary of the image is a two-dimensional or three-dimensional coordinate variable.

[0063] 2. Alternate display format iterative solution: First, the image reconstruction operator is used to solve ,Right now: ; The adaptive dynamic directional diffusion differential equation model described in claim 4: , remember , the above equation can be decomposed into The sub-equations are: , the solution u of this equation is ,remember for .

[0064] It can be solved by using Newton's iteration method after differential discretization; it can also be solved by introducing artificial variables , the following time-dependent differential equation is obtained, and the time-stepping technique is used to explicitly solve it. The differential equation corresponding to the explicit solution method of the time-stepping technique is:

[0065] in, is the boundary of the image, is a two-dimensional or three-dimensional coordinate variable.

[0066] In one embodiment, in preset scheme A, preset scheme B, and preset scheme C, the target diffusion direction can be determined by the related prior information of the object to be reconstructed, the current image gradient information, the preset diffusion direction quantity information, and the dynamic and static information parameters describing the adaptive directional diffusion. , specifically including the following: 1. If the prior information of the object to be reconstructed is known and the dynamic and static information parameters describing the adaptive directional diffusion are static, the reconstructed object can be obtained directly based on the prior information. The main characteristic directions are used to manually determine the target diffusion direction set; Second, if there is no prior information about the object to be reconstructed and the dynamic and static information parameters describing the adaptive directional diffusion are static, The diffusion angle is set as a hyperparameter and manually adjusted The angle of each diffusion direction is determined, thereby manually determining the target diffusion direction set; 3. When quickly determining the diffusion direction set, if the dynamic and static information parameters describing the adaptive directional diffusion are static, the preset diffusion direction quantity information can be used to determine the diffusion direction set. Divide the 180° angle equally, and we get directions to roughly determine the target diffusion direction set; 4. If the dynamic and static information parameters describing the adaptive directional diffusion are static, the diffusion direction set can be determined more precisely according to the gradient information of the current image and the preset diffusion direction quantity information. According to the gradient direction corresponding to each pixel in the current image, the unit vector obtained by normalizing the modulus of the gradient vector is called the gradient direction. The gradient moduli of the same gradient direction are accumulated and sorted, and the top ranked ones are selected. The gradient direction is used as the diffusion direction, so as to more precisely determine the target diffusion direction set; 5. If there is no prior information about the object to be reconstructed and the dynamic and static information parameters describing the adaptive directional diffusion are dynamic, we can take is the gradient of u or its function, For A non-zero vector or function thereof on the perpendicular tangent plane, ; 6. If there is prior information about the object to be reconstructed and the dynamic and static information parameters describing the adaptive directional diffusion are dynamic, they can be set manually. Dynamic diffusion direction.

[0067] In one embodiment, in preset scheme A, preset scheme B, and preset scheme C, the diffusion coefficient corresponding to each target diffusion direction can be determined by the current image gradient information, the preset diffusion direction quantity information, and the dynamic and static information parameters describing the adaptive directional diffusion. , specifically including the following: 1. diffusion direction or The diffusion direction function corresponds to The diffusion coefficient is set as a hyperparameter and manually adjusted The value of a parameter, thereby artificially determining the diffusion coefficient set; 2. When quickly determining the diffusion coefficient set, directly set The diffusion coefficients are equal, that is, , thereby roughly determining the diffusion coefficient set; 3. If the dynamic and static information parameters describing the adaptive directional diffusion are static, the diffusion coefficient set can be determined more precisely according to the gradient information of the current image and the preset diffusion direction quantity information. According to the gradient direction corresponding to each pixel in the current image, the gradient modulus lengths of the same gradient direction are accumulated and sorted, and the top ranked ones are selected. diffusion directions, divide the accumulated gradient modulus corresponding to the target diffusion direction by all The sum of the accumulated gradient modulus lengths corresponding to the diffusion directions is obtained, and the obtained ratio is determined as the diffusion coefficient of the target diffusion direction, so as to determine the diffusion coefficient set more precisely.

[0068] In one embodiment, in preset scheme A, preset scheme B and preset scheme C, the edge indication function can be determined according to the prior information of the object to be reconstructed or according to the difference between the adjacent pixel values ​​of each point in the current image. , where the edge indicator function Used to indicate the edge points and non-edge points of an image, including at least the following three methods: 1. If the prior information of the object to be reconstructed is known, the edge points can be directly determined based on the prior information. The value at the edge point is , and the value at non-edge points is ,in and is the parameter; 2. If there is no prior information about the object to be reconstructed, It can be defined as a piecewise function related to the adjacent pixel differences of each pixel of the current image, with the current image in the preset scheme A For example, if it is a two-dimensional case, let The coordinates of any pixel point can be calculated based on the sum of the absolute values ​​of the differences between the four adjacent pixels (upper, lower, left, and right) at each pixel. Or the sum of the absolute values ​​of the differences between eight adjacent pixels (upper, lower, left, right, upper left, lower left, upper right, lower right) Determine the edge point; when or When it is greater than the preset threshold, is the edge point, The value is , otherwise the value at the non-edge point is ,in as a parameter.

[0069] In the three-dimensional case, it is assumed that the sum of the absolute values ​​of the six adjacent aberrations at each pixel can be Or the sum of the absolute values ​​of the differences of 26 adjacent pixels Determine the edge point, when or When it is greater than the preset threshold, is the edge point, The value is , otherwise the value at the non-edge point is ,in and as a parameter.

[0070] For each pixel point, the 26 adjacent pixel points in the 26 directions are respectively the upper, lower, left, right, front and back pixel points adjacent to the edge of the pixel point (6); the upper left pixel point adjacent to both the left pixel point and the upper pixel point also includes the lower left, front left, back left, upper right, lower right, front right, back right, upper front, upper back, lower front and lower back pixel points (12); the upper left front pixel point adjacent to the left pixel point, the upper pixel point and the front pixel point also includes the upper left back, lower left front, lower left back, upper right front, upper right back, lower right front and lower right back pixel points (8).

[0071] If there is no prior information about the object to be reconstructed, It can also be defined as a monotonically increasing function related to the adjacent pixel difference of each pixel point of the current image gradient. Taking the exponential function as an example, for example, in the preset scheme A or the preset scheme C, in the two-dimensional case, it can be set or ); In the three-dimensional case, we can assume or ,in and is a parameter.

[0072] In one embodiment, the preset scheme A, the preset scheme B and the preset scheme C further include adding a general regularization term R(u) during the iterative update process, for example, a TV regularization term, Regularization term, Regularization term; if the regularization term R(u) is not added, it is equivalent to R(u)=0.

[0073] In one embodiment, in claim 9, when multiple sets of data are input, an image is outputted according to each set of data, and a method for determining a final output image for the images outputted for each set of data includes: 1. If there is prior information, such as the TV value and CT value of the target image, it can be calculated separately These indicators are compared, and the one closest to the target image is selected to obtain the optimal output image; Second, if there is no prior information, manual screening can be performed or the image with the best calculation index can be selected as the output image based on the calculation of non-reference image indicators, such as PIQE (perception-based image quality evaluator), BRISQUE (non-reference image spatial quality evaluator), NIQE (natural image quality evaluator), etc.

[0074] In one embodiment, according to claim 10, the user may also input one or more sets of parameters according to actual conditions to determine the final output image, and the method includes: When performing rapid imaging, a set of parameters is input, that is, one of the preset schemes A, B, and C is selected and executed once, and a final image is output; When performing fine imaging, at least three sets of parameters are input, and the parameters are selected so that preset scheme A, preset scheme B, and preset scheme C can each be executed at least once, and the final output image is determined based on the output image after the execution of each set of parameters and based on the prior information of the target image or by calculating the no-reference image index; When performing more detailed imaging, multiple groups of parameters can be input, and the parameters are selected so that two of the preset schemes A, preset scheme B, and preset scheme C are executed. The final output image is determined based on the output image after the execution of each group of parameters and based on the prior information of the target image or by calculating the no-reference image index.

[0075] In summary, the present invention can adaptively adjust the diffusion direction during image processing or image reconstruction, that is, by combining the prior information of the target image and analyzing the local features and gradient information of the current image, the diffusion direction and diffusion coefficient can be dynamically adjusted to adapt to the image structure of different regions, thereby more effectively achieving image smoothing and edge retention, while removing image noise artifacts while retaining the details and texture of the image, thereby obtaining a clearer and more realistic image. The present invention can be used for imaging technologies such as image processing, CT, and magnetic resonance. The key technology of the present invention is not only applicable to situations where data is incomplete (such as CT reconstruction under extreme conditions such as limited angles, sparse angles, and low doses), but also to image processing and reconstruction under conditions where data is relatively complete, so it has strong applicability and versatility.

[0076] An embodiment of the present invention provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0077] Figure 5 A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 50, a memory 51, a bus 52 and a communication interface 53, wherein the processor 50, the communication interface 53 and the memory 51 are connected via the bus 52; the processor 50 is used to execute an executable module stored in the memory 51, such as a computer program.

[0078] The memory 51 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 53 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0079] The bus 52 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0080] Among them, the memory 51 is used to store programs, and the processor 50 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 50 or implemented by the processor 50.

[0081] The processor 50 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 50. The above processor 50 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to be executed, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 51, and the processor 50 reads the information in the memory 51 and completes the steps of the above method in combination with its hardware.

[0082] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0083] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., various media that can store program codes.

[0084] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An image reconstruction method based on the idea of ​​adaptive directional diffusion, characterized in that: The steps in the method include: Get the dataset , reconstruct the geometric parameter set , initialize the candidate image , determine the upper limit parameter of the number of iterations , Iteration termination threshold parameter , dynamic and static information parameters describing adaptive directional diffusion and the number of diffusion directions and mathematical model category information parameters; Iteratively process candidate images : Based on the candidate images and the prior information of the object to be reconstructed, based on the dynamic and static information parameters describing the adaptive directional diffusion and the number of diffusion directions And the mathematical model category information parameters, determine diffusion directions and the corresponding Diffusion coefficient And the edge preserving operator , combined with data fidelity constraints, determine the mathematical model of adaptive directional diffusion, and Update to get : in, , P is a known data set, A is the corresponding system matrix; is the i-th diffusion direction The corresponding directional diffusion operator is, Indicates the diffusion direction The diffusion coefficient , Represents the number of diffusion directions; since multiplying the target optimization function by a non-zero multiple does not change the solution of the optimization problem, there is no loss of generality Fixed to is an edge-preserving operator used to identify image edges, so that The edges remain consistent; Determine whether the difference between two adjacent iterative images is less than the iteration termination threshold ,Right now , or whether the upper limit of the number of iterations is reached If it is not satisfied, the iterative process continues; if it is satisfied, the image is output.

2. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1 is characterized in that: The optimization model of the adaptive directional diffusion includes the dynamic and static information parameters as static, the number of diffusion directions as The mathematical model describing the adaptive directional diffusion is an adaptive static directional diffusion linear differential equation model. The image updating method includes: When using preset plan A, set , according to the prior information of the object to be reconstructed and the current image And the preset diffusion direction quantity information ,Sure static diffusion direction and The corresponding diffusion coefficients and the edge preserving operator , combined with data fidelity constraints, the image is simulated by a preset adaptive static directional diffusion linear differential equation model. After updating, the image is obtained ,Right now: in, ,Pick ; is the image area, is the integral differential element; For along the direction The directional derivative of is a unit vector whose components represent the direction cosines with respect to the coordinate axes. .

3. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1 is characterized in that: The optimization model of the adaptive directional diffusion includes the dynamic and static information parameters as static, the number of diffusion directions as The mathematical model describing adaptive directional diffusion is adaptive static directional diffusion The optimization model and image update method also include: When using preset solution B, set for , according to the prior information of the object to be reconstructed and the current image And the number of preset diffusion directions ,Sure static diffusion direction and The corresponding diffusion coefficients and the edge preserving operator , combined with data fidelity constraints, through the following preset adaptive static directional diffusion Optimize the model for images After updating, the image is obtained ,Right now: in, ,Pick , , Represents the image coordinates, which is a two-dimensional coordinate variable Or three-dimensional coordinate variables , is the image area, for The discrete gradient of is a unit vector whose components represent the direction cosines with respect to the coordinate axes. ; For each pixel, if the image is three-dimensional, the value of the image discrete gradient at any pixel is , if the image is two-dimensional, the value of the image discrete gradient at any pixel point is = , Along axis, Common discrete differential operators for the a-axis and z-axis, including: forward difference operator, backward difference operator and central difference operator; is a counting operator, Represented as an image Along the direction of norm, which means in a given direction Satisfaction The number of pixels.

4. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: The optimization model of the adaptive directional diffusion includes dynamic and static parameter information as dynamic, the number of diffusion directions as , the mathematical model describing the adaptive directional diffusion is an adaptive dynamic directional diffusion differential equation model, and the image updating method further includes: When using preset solution C, set for , according to the prior information of the object to be reconstructed and the current image And the preset diffusion direction quantity information ,Sure Dynamic diffusion direction and The corresponding diffusion coefficients and the edge preserving operator , combined with the data fidelity constraints, the image is modeled by the following preset adaptive dynamic directional diffusion differential equations. After updating, the image is obtained ,Right now: in, ,Pick The corresponding first-order variation is , is the dynamic diffusion direction, H is the Hesse matrix of u, when the image is two-dimensional, H= , when the image is three-dimensional, H= , through the above model Realize along each dynamic direction Adaptive diffusion.

5. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: The diffusion directions include: By describing the dynamic and static information parameters of adaptive directional diffusion, the relevant prior information of the object to be reconstructed, the current image gradient information and the preset number of diffusion directions ,Sure static or dynamic diffusion direction .

6. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: The diffusion coefficients include: By describing the dynamic and static information of adaptive directional diffusion, the current image gradient information and the preset number of diffusion directions ,Sure Diffusion coefficient corresponding to diffusion direction .

7. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: The edge preserving operator is characterized by: is an edge-preserving operator used to identify image edges, so that The edge of The edge of the image is kept consistent; the edge preservation operator is determined according to the prior information of the object to be reconstructed or according to the difference between the adjacent pixel values ​​of each point in the current image. .

8. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that:In the image iterative update process, the optimization model of adaptive directional diffusion also includes a regularization term R(u), wherein the regularization term R(u) includes: TV regularization term, The regularization term and Regularization term, if the regularization term R(u) is not added, it is equivalent to R(u)=0.

9. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: When multiple sets of data are input, an image will be output according to each set of data. For the images output by each set of data, the method for determining the final output image includes: The final output image is determined by the prior information of the target image or by calculating the reference-free image index.

10. The image reconstruction method based on the adaptive directional diffusion concept according to claim 1, characterized in that: The method comprises: When performing fine imaging, at least three sets of parameters are input, and the parameters are selected so that preset scheme A, preset scheme B, and preset scheme C can each be executed at least once, and the final output image is determined based on the output image after the execution of each set of parameters and based on the prior information of the target image or by calculating the no-reference image index; When performing finer imaging, multiple sets of parameters may be input, and the parameters are selected so that two of the preset schemes A, B, and C are executed. The final output image is determined based on the output images after the execution of each set of parameters and based on the prior information of the target image or by calculating the index of the no-reference image. When performing rapid imaging, a set of parameters is input, and any preset scheme is selected from preset scheme A, preset scheme B, and preset scheme C to execute to determine the final output image.

11. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 10.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 10.