A method for determining emission tomography weight matrix for free projection in three-dimensional space

By establishing a three-dimensional space free projection emission tomography model and bilinear interpolation principle to determine the weight matrix, the three-dimensional imaging problem of combustion field due to the impact of lens imaging effect and camera position error in the prior art is solved, and efficient and accurate combustion field three-dimensional reconstruction.

CN114187175BActive Publication Date: 2024-11-29SHAANXI SANHANG OPTOMETRY ELECTRONIC TECHNOLOGY CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202111354183.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-15
Publication Date
2024-11-29
Estimated Expiration
2041-11-15

AI Technical Summary

Technical Problem

In the existing emission tomography technology, the parallel projection model caused by the two-dimensional Radon transformation model is not applicable. Combined with the CCD camera lens imaging effect and camera position error, it seriously affects the reconstruction accuracy and efficiency of the three-dimensional imaging of the combustion field.

Method used

Establish a three-dimensional free projection emission tomography model that considers the lens imaging effect and the camera's three-dimensional spatial position, and determine the weight matrix through the principles of projection integration discretization and bilinear interpolation to improve calculation accuracy and efficiency.

Benefits of technology

High-precision and high-efficiency three-dimensional space free projection emission tomography reconstruction is realized, the projection acquisition range of the emission tomography system is expanded, and the reconstruction accuracy is improved. It is of great significance to the practicality of emission tomography technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114187175B_ABST
    Figure CN114187175B_ABST
Patent Text Reader

Abstract

The present invention is a method for determining an emission tomography weight matrix for free projection in three-dimensional space, which overcomes the problems of imperfect models, low efficiency and accuracy in the prior art. The method of the present invention is simple and has high calculation accuracy and efficiency. The present invention includes the following steps: Step 1, establishing a three-dimensional free projection emission tomography model considering the lens imaging effect and the three-dimensional spatial position of the camera; Step 2: discretization of the projection integral; Step 3: determining the weight matrix of the system based on the bilinear interpolation principle.
Need to check novelty before this filing date? Find Prior Art

Description

Technical field:

[0001] The invention belongs to the technical field of optical computational imaging and combustion field three-dimensional measurement, and relates to a combustion field emission spectrum tomography reconstruction method, and in particular to an emission tomography weight matrix determination method for three-dimensional space free projection. Background technology:

[0002] The imaging, measurement and diagnosis of combustion fields are the basis for the design of key instruments and meters in modern aviation, aerospace, missiles, weapons and energy industries. Among them, the three-dimensional imaging and display of the combustion process is one of the important means necessary for studying and controlling combustion. Emission Computerized Tomography (ECT) is a simple, efficient and non-contact three-dimensional imaging method for combustion fields. It uses a planar CCD camera or a fiber array to collect two-dimensional projection images of the emission spectrum intensity of the combustion field in multiple directions, and then combines computerized tomography (CT) theory to reconstruct the combustion field in three dimensions. Due to the simplicity of the measurement principle and device, it has become a research hotspot in combustion imaging detection technology in recent years and has broad application prospects.

[0003] However, the existing emission tomography technology is based on the mathematical foundation of two-dimensional Radon transform. For emission tomography systems that use CCD cameras for projection acquisition, its implementation process is subject to two limitations: (1) Radon transform uses a parallel projection model, which is no longer valid due to the imaging effect of the CCD camera lens; (2) Existing parallel projection emission tomography has strict requirements on the camera position. Due to the limitations of the actual test environment, it is generally impossible to provide an optical window in an ideal position, and position errors are inevitable when assembling the detector. These two constraints seriously affect the accuracy of tomography reconstruction. Therefore, it is required that two factors must be considered when modeling the emission tomography system: (1) the imaging effect of the lens; and (2) the three-dimensional spatial position of the camera.

[0004] Among the tomographic reconstruction algorithms, the algebraic iteration algorithm is the most commonly used reconstruction method in tomographic technology. It discretizes the tomographic projection process, uses a weight matrix to represent the contribution of the physical function of different discrete grids in the measured object to the projection information, and converts the projection process into a series of linear equations. The reconstruction process is to solve the linear equations using different forms of algebraic iteration methods. Among them, the calculation accuracy of the weight matrix seriously affects the accuracy and quality of tomographic reconstruction. Different projection models determine different weight matrices, so it is of great significance to study the weight matrix determination method considering the above two factors. Although a small number of studies have used ray tracing and lens imaging theory to study this problem, the computational efficiency and accuracy are relatively low. Summary of the invention:

[0005] The purpose of the present invention is to provide a method for determining an emission tomography weight matrix for free projection in three-dimensional space, which overcomes the problems of imperfect models, low efficiency and accuracy in the prior art. The method of the present invention is simple and has high calculation accuracy and efficiency.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for determining an emission tomography weight matrix for free projection in three-dimensional space, characterized in that it comprises the following steps:

[0008] Step 1: Establish a three-dimensional free projection emission tomography model that takes into account the lens imaging effect and the three-dimensional spatial position of the camera;

[0009] Step 2: Discretization of projection integral;

[0010] Step 3: Based on the bilinear interpolation principle, determine the weight matrix of the system.

[0011] Step one includes the following steps:

[0012] S1: The lens imaging effect is manifested as the coordinates of the object point in the camera coordinate system (x, y, z) and the image point coordinates in the camera imaging surface coordinate system (x', y') satisfying the relationship

[0013]

[0014] Where f is the focal length of the camera lens;

[0015] S2: Any three-dimensional spatial position of the camera is expressed as a world coordinate system (x w ,y w ,z w ) and the camera coordinate system (x, y, z) satisfy the relationship

[0016]

[0017] Where R is the rotation matrix and T is the translation vector;

[0018] S3: The final tomographic projection model is

[0019]

[0020] Where f(x, y, z) is the luminous intensity value of the object point in the camera coordinate system, z min and z max The upper and lower boundaries where each projection ray intersects the reconstruction area in the camera coordinate system.

[0021] Step 2 includes the following steps:

[0022] S1: The three-dimensional object to be measured f(x w ,y w ,z w ) is evenly divided into M×N×P discrete grids with a grid size of Δg×Δg×Δg. Assume that the value of each grid is f i (x w ,y w ,z w ) is a constant, where i = 1, 2, ..., M × N × P;

[0023] S2: Determine the upper limit z of the projection integral based on the camera coordinate values ​​of the 8 vertices in the reconstruction area max and the lower limit z min ;

[0024] S3: Divide the integral interval into Q micro-elements evenly. Assume that the physical function of each micro-element interval is a constant. The projection model is: where (x',y') and (x m ,y m ,z m )Satisfaction relationship

[0025]

[0026] According to the conversion relationship between the world coordinate system and the camera coordinate system, the position of the integral differential element in the world coordinate system (x mw ,y mw ,z mw ).

[0027] Step three includes the following steps:

[0028] S1: Determine the minimum adjacent discrete grid number of each microelement;

[0029] S2: Determine the distance between the microelement and the center of the smallest adjacent discrete grid;

[0030] S3: According to the bilinear interpolation principle, the physical function value of the microelement is determined by the physical functions of the eight adjacent discrete grid center points, and the weight factor of each discrete grid to the microelement is obtained;

[0031] S4: Perform integration operations on all the infinitesimal elements to obtain the weight matrix of each discrete grid pair projection.

[0032] Compared with the prior art, the present invention has the following advantages and effects:

[0033] 1. The method of the present invention obtains a tomographic model that takes into account the lens imaging effect and the three-dimensional spatial position of the camera, providing a theoretical basis for the emission tomographic reconstruction of free projection in three-dimensional space; on the basis of the model, the weight matrix is ​​determined based on the bilinear interpolation principle, which has a simple principle and high calculation accuracy and efficiency.

[0034] 2. The present invention can expand the projection acquisition range of the emission tomography system and improve the reconstruction accuracy, which is of great significance for the practical application of emission tomography technology. Description of the drawings:

[0035] Figure 1 It is a schematic diagram of the coordinate system definition of a multi-camera emission tomography system;

[0036] Figure 2 It is a schematic diagram of the camera projection imaging model;

[0037] Figure 3 It is the projection imaging result of the cube at different rotation angles;

[0038] Figure 4 It is a hollow sphere and its reconstruction result;

[0039] Figure 5 is the synthetic Shepp-Logan model and its reconstruction result. Specific implementation method:

[0040] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] The present invention is a method for determining an emission tomography weight matrix for free projection in three-dimensional space. Firstly, a three-dimensional free projection emission tomography model is established on the basis of considering the lens imaging effect and the three-dimensional spatial position of the camera; then, the projection integral is discretized and represented based on the projection model; finally, the weight matrix of the tomography system is determined by using the bilinear interpolation principle.

[0042] The present invention specifically comprises the following steps:

[0043] Step 1: Establish a 3D free-projection emission tomography model that takes into account the lens imaging effect and the 3D spatial position of the camera

[0044] S1: The lens imaging effect is manifested as the coordinates of the object point in the camera coordinate system (x, y, z) and the image point coordinates in the camera imaging surface coordinate system (x', y') satisfying the relationship

[0045]

[0046] Where f is the focal length of the camera lens.

[0047] S2: Any three-dimensional spatial position of the camera is expressed as a world coordinate system (x w ,y w ,z w ) and the camera coordinate system (x, y, z) satisfy the relationship

[0048]

[0049] Where R is the rotation matrix and T is the translation vector.

[0050] S3: The final tomographic projection model is

[0051]

[0052] Where f(x, y, z) is the luminous intensity value of the object point in the camera coordinate system, z min and z max The upper and lower boundaries where each projection ray intersects the reconstruction area in the camera coordinate system.

[0053] Step 2: Projection integral discretization:

[0054] S1: The three-dimensional object to be measured f(x w ,y w ,z w ) is evenly divided into M×N×P discrete grids with a grid size of Δg×Δg×Δg. Assume that the value of each grid is f i (x w ,y w ,z w ) is a constant, where i = 1, 2, …, M × N × P.

[0055] S2: Determine the upper limit z of the projection integral based on the camera coordinate values ​​of the 8 vertices in the reconstruction area max and the lower limit z min ;

[0056] S3: Divide the integral interval into Q micro-elements evenly. Assume that the physical function of each micro-element interval is a constant. The projection model is: where (x',y') and (x m ,y m ,z m )Satisfaction relationship

[0057]

[0058] According to the conversion relationship between the world coordinate system and the camera coordinate system, the position of the integral differential element in the world coordinate system (x mw ,y mw ,zmw ).

[0059] Step 3: Determine the system weight matrix based on the bilinear interpolation principle

[0060] S1: Determine the minimum adjacent discrete grid number of each microelement;

[0061] S2: Determine the distance between the microelement and the center of the smallest adjacent discrete grid;

[0062] S3: According to the bilinear interpolation principle, the physical function value of the microelement is determined by the physical functions of the eight adjacent discrete grid center points, and the weight factor of each discrete grid to the microelement is obtained;

[0063] S4: Perform integration operations on all the infinitesimal elements to obtain the weight matrix of each discrete grid pair projection.

[0064] Example:

[0065] The present invention comprises the following steps:

[0066] 1. Obtain a three-dimensional free-projection emission tomography model, including:

[0067] See also Figure 1 The coordinate system definition of the emission tomography system with multiple cameras in the system has a fixed world coordinate system (x w ,y w ,z w ), the luminous intensity of the combustion field under test is expressed as f(x w ,y w ,z w ). A camera at any position in the world coordinate system is represented by the camera coordinate system (x, y, z), where the (x, y) plane is parallel to the CCD target surface, the x-axis is the direction of the long side of the CCD target surface, the y-axis is the direction of the short side of the CCD target surface, and the origin of the coordinates is the intersection of the camera optical axis and the lens. The coordinate system (x', y') of each camera imaging surface is the CCD target surface, the x' axis is the opposite direction of the x-axis, the y' axis is the opposite direction of the y-axis, and the origin of the coordinates is the intersection of the lens optical axis and the CCD target surface.

[0068] In three-dimensional space, from the world coordinate system (x w ,y w ,z w ) to any camera coordinate system (x, y, z) can be obtained by using the Euler angles (nutation angle ψ, precession angle θ and rotation angle φ) and three translation quantities (T x ,T y ,T z ) is determined, and its conversion relationship can be expressed as

[0069]

[0070] Among them, the rotation matrix

[0071]

[0072] Translation Vector

[0073]

[0074] For simplicity, we can express formula (1) as

[0075]

[0076] In the actual emission tomography system, the parameters r1, r2, ..., r9 in the rotation matrix and the parameters T in the translation vector corresponding to each camera x ,T y ,T z It can be accurately determined by camera calibration.

[0077] Since the imaging lens with a suitable depth of field can be selected according to the size of the combustion field to be measured, the influence of the depth of field does not need to be considered in the camera imaging model, so it can be equivalent to a pinhole camera model, such as Figure 2 shown.

[0078] The straight line formed by any image point P' on the imaging plane and the origin O of the camera coordinate system is called a projection ray. The projection ray intersects with the measured area, and all object points P m (m=1,2,……,n) will be imaged at the same image point P'. Assuming the focal length of the camera lens is f, the positions of the object point and the image point satisfy the triangle similarity principle, that is:

[0079]

[0080] The luminous intensity information collected by this image point is

[0081]

[0082] where z min and z max are the upper and lower boundaries where each projection ray intersects with the reconstruction area, and f(x, y, z) is the luminous intensity value of all object points on the projection ray.

[0083] 2. Discretize the projection integral, including:

[0084] The three-dimensional object f(x w ,y w ,z w ) is evenly divided into M×N×P discrete grids with a grid size of Δg×Δg×Δg. Assume that the value of each grid is f i (x w,y w ,z w ) is a constant, where i = 1, 2, ..., M × N × P. If there are X cameras in the tomography system, and the number of pixels of each camera is Y, then the tomography system has a total of XY projection rays. If the intersection intercept of the jth projection ray and the i-th grid is recorded as ω ij , obviously there are only a few i corresponding to ω ij is non-zero. Thus, the pixel intensity I corresponding to the jth projection ray j It can be expressed discretely as:

[0085]

[0086] where ω ij is called the projection weight factor of the i-th grid to the j-th ray, which represents the contribution of the i-th grid to the j-th projection. ij It constitutes the weight matrix of the tomography system.

[0087] The world coordinate system of the 8 vertices in the reconstruction area is (x wc ,y wc ,z wc ) p , where p = 1, 2, ..., 8. Substituting the coordinate values ​​of these 8 points into formula (1), we can get the positions of the 8 vertices in the camera coordinate system (x c ,y c ,z c ) p . Compare 8 z c The value of the integral interval [z min ,z max ]. Discretize the camera imaging projection model expressed in formula (4) and convert the integral interval [z min ,z max ] is evenly divided into Q microelement intervals, that is, the number of integration steps is Q, and the step length is

[0088]

[0089] Assume that the material function of each infinitesimal interval is a constant, so

[0090]

[0091] where (x',y') and (x m ,y m ,z m )Satisfaction relationship

[0092]

[0093] According to formula (1), the position of the integral differential element in the world coordinate system can be determined as:

[0094]

[0095] Therefore, we can get

[0096]

[0097] 3. Calculate the weight matrix of the tomography system using the bilinear interpolation principle, including:

[0098] According to the bilinear interpolation principle, the value of the physical function at any position in the discrete three-dimensional space can be determined by the physical function value of its adjacent discrete grid center point. The specific process is:

[0099] (1) Determine the minimum adjacent discrete grid number of each microelement

[0100]

[0101] (2) Determine the distance between the microelement and the center of the smallest adjacent discrete grid, respectively: Δx m , Δy m and Δz m .

[0102] (3) The physical function value of the microelement is determined by the physical function of the eight adjacent discrete grid center points.

[0103]

[0104] According to the coefficient determined by the distance between adjacent discrete grids, the weight factor of each discrete grid to the infinitesimal element can be obtained.

[0105] (4) The projection pixel value of each ray can be written as follows

[0106]

[0107] By integrating all the infinitesimal elements according to this formula, the weight matrix of each discrete grid pair projection can be obtained.

[0108] Experimental example:

[0109] In order to intuitively reflect the accuracy of the three-dimensional camera imaging model and the weight matrix calculation method based on bilinear difference, the imaging projection of a cube in three-dimensional space is first calculated. The cube is divided into 50×50×50 discrete grids with a grid size of 0.4mm, and the value of each grid is a constant 1. The number of pixels of the simulated camera is 138×138, the pixel size is 3.75μm, and the focal length of the lens is 12mm. When the translation vector is [0mm, 0mm, 800mm], the projections at different rotation angles calculated according to the method proposed in the invention are as follows Figure 3 The calculated projection results are completely consistent with the theoretical visual projection results, proving the correctness of the above method.

[0110] Two three-dimensional simulation fields F1 and F2 are used to verify the effectiveness and accuracy of the three-dimensional free projection tomographic reconstruction algorithm. F1 is a three-dimensional hollow sphere that can reflect the integrity of the reconstruction results; F2 is a synthetic Shepp-Logan model commonly used in tomography. Both fields are divided into 50×50×50 cubic discrete grids with a grid size of 0.4mm. It is assumed that there are 12 cameras in the tomographic system to collect projections in different directions. The number of pixels of the camera is 138×138, the pixel size is 3.75μm, and the focal length of the lens is 12mm. The translation vectors are all [0mm, 0mm, 800mm], the precession angle θ and the rotation angle φ are both 5°, and the nutation angle ψ is uniformly distributed in [-90°, 90°].

[0111] First, the projection imaging of different cameras is calculated using the method proposed in the present invention, and then the tomographic reconstruction is performed using the ART algorithm. The reconstruction result of the hollow sphere F1 is shown in Figure 4 As shown in the figure. From the structural point of view, the sphere structure in the F1 reconstruction result is very complete and perfectly matches the original model structure, indicating that the integrity of the reconstruction result is very high; the root mean square error of the reconstruction result is 2.2083×10 -4 , indicating that the reconstruction accuracy is very high. The reconstruction results of the synthetic Shepp-Logan model are as follows Figure 5 As shown, the root mean square error of the reconstruction result is 7.2078×10 -5 , with very high reconstruction accuracy.

[0112] The three simulation results in this experimental example verify the correctness of the mathematical model of camera imaging projection in three-dimensional space and the accuracy of the weight matrix calculation method.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any equivalent structural changes made using the contents of the description and drawings of the present invention should be included in the patent protection scope of the invention.

Claims

1. A method for determining an emission tomography weight matrix for free projection in three-dimensional space, characterized in that: The following steps are involved: Step 1: Establish a three-dimensional free projection emission tomography model that takes into account the lens imaging effect and the three-dimensional spatial position of the camera; Step 2: Discretization of projection integral; Step 3: Based on the bilinear interpolation principle, determine the weight matrix of the system; Step three includes the following steps: S1: Determine the minimum adjacent discrete grid number of each microelement; S2: Determine the distance between the microelement and the center of the smallest adjacent discrete grid; S3: According to the bilinear interpolation principle, the physical function value of the microelement is determined by the physical functions of the eight adjacent discrete grid center points, and the weight factor of each discrete grid to the microelement is obtained; S4: Perform integration operations on all the infinitesimal elements to obtain the weight matrix of each discrete grid pair projection.

2. The method for determining an emission tomography weight matrix for free projection in three-dimensional space according to claim 1, characterized in that: Step one includes the following steps: S1: The lens imaging effect is manifested as the coordinates of the object point in the camera coordinate system (x, y, z) and the image point coordinates in the camera imaging surface coordinate system (x', y') satisfying the relationship Where f is the focal length of the camera lens; S2: Any three-dimensional spatial position of the camera is expressed as a world coordinate system (x w ,y w ,z w ) and the camera coordinate system (x, y, z) satisfy the relationship Where R is the rotation matrix and T is the translation vector; S3: The final tomographic projection model is Where f(x, y, z) is the luminous intensity value of the object point in the camera coordinate system, z min and z max The upper and lower boundaries where each projection ray intersects the reconstruction area in the camera coordinate system.

3. The method for determining an emission tomography weight matrix for three-dimensional space free projection according to claim 1, characterized in that: Step 2 includes the following steps: S1: The three-dimensional object to be measured f(x w ,y w ,z w ) is evenly divided into M×N×P discrete grids, the grid size is △g×△g×△g, assuming that the value of each grid is f i (x w ,y w ,z w ) is a constant, where i = 1, 2, ..., M × N × P; S2: Determine the upper limit z of the projection integral based on the camera coordinate values ​​of the 8 vertices in the reconstruction area max and the lower limit z min ; S3: Divide the integral interval into Q micro-elements evenly. Assume that the physical function of each micro-element interval is a constant. The projection model is: where (x',y') and (x m ,y m ,z m )Satisfaction relationship According to the conversion relationship between the world coordinate system and the camera coordinate system, the position of the integral differential element in the world coordinate system (x mw ,y mw ,z mw ).

Citation Information

Patent Citations

  • Multi-direction flame emission chromatographic system

    CN106600687A

  • Method for determining emission tomography weight matrix based on bilinear interpolation principle

    CN110400253A