Analysis method of geometric error transfer model based on visual simulation system

Abstract

The present invention relates to a method for analyzing a geometric error transfer model of a simulation system. Based on the conversion process from a world coordinate system to a screen coordinate system, a physical transfer process model is constructed and the accuracy of the geometric transfer model is verified through experiments. Influencing factors of geometric errors are proposed, and the main error factors are screened through sensitivity analysis. Based on the geometric transfer model of a visual simulation system, the error transfer function is constructed by deriving the geometric transfer model through the error transfer function construction rule. The experiment is designed, and the deviation between the error transfer function result and the actual error result is calculated with the physical model as the actual system. The present invention ensures that before the simulation system simulates an unknown environment, the geometric error transfer function is used to analyze its accuracy, and the analysis of the influence of the error factors is more intuitive, which is conducive to the error optimization control of the system.

Description

Technical Field

[0001] The invention relates to the technical field of performance analysis of a simulation system, in particular to a method for analyzing a geometric error transfer model of a simulation system. Background Art

[0002] Visual simulation technology has been applied in more and more fields by more and more countries. Visual simulation imaging technology has been rapidly developed. In order to test optoelectronic equipment, there are generally mathematical simulation, field experiments, traditional indoor experiments, semi-physical simulation and other methods. Among them, semi-physical simulation combines actual objects with virtual digital scenes and introduces them into the test loop, which is more accurate than digital simulation. Therefore, in recent years, semi-physical simulation testing of systems in indoor environments has gradually become a hot research topic in various countries. However, the success of such semi-physical simulation system application testing depends largely on the simulation fidelity of the infrared visual simulation system, that is, the degree to which the infrared visual simulation system reproduces the complex real natural environment. In the field of military simulation and application, the fidelity of the infrared visual simulation system for target and background simulation will directly affect the accuracy of the performance identification of complex optoelectronic equipment. If the infrared visual simulation system is to be truly applied to the test and identification of optoelectronic equipment, it is necessary to ensure the high fidelity of the simulation system.

[0003] The verification of the geometric fidelity of infrared visual simulation systems has always been the focus and development direction of research in this field. At present, there are relatively mature verification schemes and theories for simulation fidelity and confidence abroad, and a large number of experimental verifications have been done, accumulating rich experimental data and results. There are also many units in China that have developed infrared visual simulation systems, and there are also many relatively mature results in static infrared image generation and dynamic infrared scene integration. The development technology for infrared target simulators has also made great progress, and the performance has become relatively mature. For example, the Rochester Institute of Technology has conducted a large number of experiments on digital simulation scenes and developed a scene simulation model DIRSIG (DIRSIG Digital Imaging and Remote Sensing Image Generation), which is a comprehensive image simulation and generation model based on physics-driven, used to generate multi-spectral, hyperspectral, and single-band images in the 0.3-20μm band. It can also be used to test image system design or simulate image generation of image evaluation algorithms. Based on the DIRSIG scene simulation model, Scott Brown and Jim Salacain verified the accuracy of the geometric transfer model, and compared the simulated image with the real image in terms of geometric projection and optical distortion to obtain a high fidelity result. Xidian University in China has also achieved a series of research results in this regard. Chen Yechao and others studied the fidelity analysis of infrared vision simulation systems. The geometric fidelity analysis used a geometric physical transfer model to estimate the position parameters through the model, and calculated the target area pixel point change and position change indicators respectively, and combined the two indicators to evaluate and analyze the geometric fidelity.

[0004] From the current status of technological development, the geometric fidelity analysis methods of visual simulation systems can be divided into the following categories:

[0005] (1) Analyze the geometric fidelity by comparing the simulated image with the real image. However, this analysis method does not analyze the mechanism of error generation, making it difficult to optimize the system through evaluation indicators.

[0006] (2) By establishing a geometric physical transfer model, the transfer process of the simulation system is simulated, and the output of the geometric transfer model is used as the simulation result to compare with the real image. This method analyzes the mechanism of error generation to a certain extent, but because there are many influencing variables, it is impossible to directly express the impact of a single error.

[0007] (3) Through the combination of different evaluation indicators, a more objective and comprehensive evaluation indicator is formed. For example, indicator 1: the offset between pixels, indicator 2: the sum of the number of pixels in the image. The combination of the two indicators will result in a more comprehensive evaluation system.

[0008] Although the current geometric error analysis of visual simulation systems has achieved some results, the results of the research are still insufficient in analyzing the influence of various error parameters, and cannot accurately describe the influence of error parameters under different circumstances. How to establish a set of analysis methods that can not only analyze the system errors, but also point out which parameters have a great influence on the errors and need to be improved and controlled. Such related methods have few results in the field of environmental simulation. Summary of the invention

[0009] In view of the shortcomings of the prior art, the present invention provides a geometric error analysis method based on an environmental simulation system. The analysis method first analyzes the system and constructs its geometric physical transfer model to establish a set of error influencing parameters. An error transfer model is established based on the geometric physical transfer model to analyze the specific impact of the influencing parameters on the error. The method provides an analytical basis for system optimization.

[0010] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0011] The geometric error transfer model analysis method based on the visual simulation system includes the following steps:

[0012] Construct a geometric physical transfer model of the visual simulation system: take the corner points in the urban buildings as the target points, measure the world coordinate system of the target points through the measuring equipment, convert the target points from the world coordinate system to the screen coordinate system, and output the coordinates of the target points in the screen coordinate system;

[0013] Select multiple geometric transfer error factors in the geometric physical transfer model of the visual simulation system, and select multiple geometric transfer error factors whose sensitivity meets the requirements through sensitivity analysis;

[0014] According to the geometric transfer error factors and the construction rules of the error transfer function, the error transfer function is constructed for the geometric physical transfer model of the visual simulation system;

[0015] The reliability of the geometric-physical transfer model is verified by the error transfer function.

[0016] The measuring equipment includes a total station and a handheld distance meter.

[0017] The conversion of the target point from the world coordinate system to the screen coordinate system is specifically as follows:

[0018] The target points in the city buildings are transformed from the world coordinate system to the camera coordinate system through the quaternion rotation matrix, from the camera coordinate system to the projection coordinate system, from the projection coordinate system to the normalized device coordinate system, and finally to the screen coordinate system.

[0019] The geometric transfer error factors include: camera field of view angle error, world coordinate system error, and camera attitude error.

[0020] The construction rule of the error transfer function is:

[0021] If the error factors are independent of each other, that is, a change in one error will not cause a change in another error, then:

[0022]

[0023] Among them, x out Output coordinate function for the x-axis of the screen coordinate system. θ is the error factor, Δθ is the error range, is the error of the x-axis of the synthesized screen coordinate system;

[0024] If there is a transfer relationship between the error factors, first calculate the error transfer function of the previous layer, that is, the error that is generated first in the physical transfer relationship, and then use it as the input error of the next layer, that is, the error that is generated later in the physical transfer relationship, and calculate the total error transfer function. The format is as follows:

[0025]

[0026] in, The error of the previous layer during the transmission process, is the final output error, x final is the final output function.

[0027] The verification of whether the geometric physical transfer model is reliable by using the error transfer function is specifically as follows:

[0028] The parameters of the building target points collected by the measuring equipment in real time are input into the geometric and physical transfer model as reference values ​​to obtain the reference coordinates of the building in the screen coordinate system. The error is added to the reference value and input into the geometric and physical transfer model to obtain the actual coordinates in the screen coordinate system after the error is added. The reference coordinates are subtracted from the actual coordinates to obtain the actual error. The error and the reference value are input into the error transfer function to obtain the estimated error. The deviation between the estimated error and the actual error is compared. If the deviation is less than the threshold, the geometric and physical transfer model is reliable.

[0029] The parameters include: coordinate values ​​in the world coordinate system of the camera and coordinate values ​​in the world coordinate system of the target point.

[0030] The present invention has the following beneficial effects and advantages:

[0031] 1. The method of the present invention uses the total station as the origin to establish a coordinate system, and calculates the camera's yaw angle and pitch angle by measuring the coordinates of the camera and the camera's field of view center point, thereby ensuring the reliability of the measurement parameters.

[0032] 2. The present invention adopts the error transfer function to analyze the geometric fidelity of the system, and the analysis of the influence of each error parameter is more intuitive, which is conducive to the optimization of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is the overall flow chart of the present invention;

[0034] Figure 2 It is a schematic diagram of geometric coordinate transfer of the present invention;

[0035] Figure 3 It is a schematic diagram of error transmission of the present invention;

[0036] Figure 4 Parameter measurement orientation diagram for the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0038] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0039] A method for analyzing a geometric error transfer model of a visual simulation system, characterized in that it comprises the following steps:

[0040] Step 1: Based on the conversion process from the world coordinate system to the screen coordinate system of urban buildings, a geometric physical transfer model of the visual simulation system is constructed, and the accuracy of the geometric physical transfer model is verified through experiments;

[0041] Step 2: Propose the influencing factors of geometric transfer error and screen the main error factors through sensitivity analysis;

[0042] Step 3: Based on the geometric and physical transfer model of the visual simulation system, the error transfer function is constructed by deriving the geometric and physical transfer model through the error transfer function construction rule;

[0043] Step 4: Design an experiment, taking the geometric physics transfer model as the actual system, and calculate the deviation between the error transfer function result and the actual error result.

[0044] The conversion from the world coordinate system to the screen coordinate system includes the following processes:

[0045] First, convert from the world coordinate system to the camera coordinate system, from the camera coordinate system to the projection coordinate system, from the projection coordinate system to the normalized device coordinate system, and finally to the screen coordinate system.

[0046] The experimental verification of the accuracy of the geometric physical transfer model includes the following process:

[0047] The experiment is designed to measure the world coordinate system and camera posture: take the total station as the origin, establish the coordinate axis with its posture, and measure the camera coordinates and the coordinates of the center point of the camera field of view. The yaw angle and pitch angle of the camera can be calculated from the camera coordinates and the coordinates of the center point of the camera field of view, and the roll angle of the camera can be measured from the camera placement posture. The remaining model parameters can be provided by the instrument parameters themselves.

[0048] The actual parameters of the camera when photographing urban buildings are measured through experimental instruments, and the parameters are substituted into the geometric and physical transfer model to calculate the estimated coordinate results. The results are compared with the corresponding coordinates of the urban buildings in the actual photographed images, and the deviation is calculated. If the deviation is within the allowable range, it proves that the geometric and physical transfer model is reliable.

[0049] The error transfer function construction rule is:

[0050] If the error parameters are independent of each other, then:

[0051]

[0052] Among them, x out is the x-axis output function, θ is the error parameter, Δθ is the error range, is the error after synthesis.

[0053] If there is a transfer relationship between errors, first calculate the error transfer function of the previous layer, and then use it as the input error of the next layer to calculate the total error transfer function. The format is as follows:

[0054]

[0055] Among them, x out is the x-axis output function, The error of the previous layer during the transmission process, is the final output error, x final is the final output function.

[0056] The deviation process between the calculated error transfer function result and the actual error result is as follows:

[0057] Set a set of input parameters as reference values, substitute them into the geometric physics transfer model as reference coordinates. Add errors to the reference values ​​to obtain input parameters with errors, input them into the geometric physics transfer model to obtain the actual coordinates after adding errors, subtract the reference coordinates before adding errors from the actual coordinates after adding errors to obtain the actual errors. Input the errors and reference values ​​into the error transfer model to obtain the estimated errors, and compare the deviations between the estimated errors and the actual errors.

[0058] like Figure 1As shown, a geometric error transfer model analysis method for a visual simulation system of the present invention comprises the following steps:

[0059] (1): Based on the transformation process from the world coordinate system to the screen coordinate system, a geometric-physical transfer model of the visual simulation system is constructed, and the accuracy of the geometric-physical transfer model is verified through experiments;

[0060] (2): Propose the influencing factors of geometric transfer error and screen the main error factors through sensitivity analysis;

[0061] (3): Based on the geometric and physical transfer model of the visual simulation system, the error transfer function is constructed by deriving the geometric and physical transfer model through the error transfer function construction rules;

[0062] (4): Design an experiment, take the geometric physical transfer model as the actual system, and calculate the deviation between the error transfer function result and the actual error result.

[0063] The conversion process from the world coordinate system to the screen coordinate system is as follows:

[0064] Convert the target points in the urban buildings from the world coordinate system (with the total station as the origin, the z-axis as the altitude direction, and the x-axis and y-axis as the set value) to the camera coordinate system (with the camera as the origin, the camera lens facing outward as the y-axis, the horizontal and vertical y-axis as the x-axis, and the altitude direction as the z-axis), from the camera coordinate system (projection conversion based on the y-axis) to the projection coordinate system (two-dimensional coordinate system, remove the y-axis in the camera coordinate system, and convert the camera coordinate system z-axis to the projection coordinate system y-axis), from the projection coordinate system to the normalized device coordinate system (the orientation remains unchanged, the coordinate values ​​are scaled and translated), and finally converted to the screen coordinate system (after normalization, x and y are both 1, and enlarged to the same size as the screen pixels).

[0065] The process of designing the experiment to measure the model parameters is as follows: Figure 4 As shown: Take the total station as the origin, establish the coordinate axis with its posture, and measure the camera coordinates and the coordinates of the center point of the camera field of view. The camera's yaw angle and pitch angle can be calculated from the camera coordinates and the coordinates of the center point of the camera field of view, and the camera's roll angle can be measured from the camera's placement posture. The four target point positions, the equipment directly measures the center point parameter indicators as shown in Table 1:

[0066] Table 1 Measurement data

[0067] parameter value Camera position (m) (0.03,5.257,-0.01) Center point pitch angle, yaw angle (°) (-9.7697，-67.0531) Center point slope distance (m) 224.5

[0068] The camera position is the coordinate value of the camera in the world coordinate system, the center point pitch angle and yaw angle are the pitch angle and yaw angle between the total station as the origin and the center point in the world coordinate system, and the center point slant distance is the straight-line distance between the total station and the center point.

[0069] The measured data include: the center point pitch angle θ1 and yaw angle measured with the total station as the origin Slope distance r, camera coordinates (x1, y1, z1). From the pitch angle θ1 and the slant distance r, the formula:

[0070] l=r×sin(-θ1)z=-r×cos(θ1)

[0071] Get the z-axis value of the target point, the horizontal distance between the target point and the total station is the horizontal straight-line distance l. Then, the horizontal distance l and the yaw angle By formula:

[0072]

[0073] Get the x, y position of the target point. The horizontal distance between the camera and the target object can be obtained from the coordinates (x, y, z) and (x1, y1, z1) of the target point and the camera. And the height difference h = z-z1, through the formula:

[0074]

[0075] Get the camera's pitch angle θ and yaw angle The roll angle can be directly obtained from experimental conditions.

[0076] The experiment finally measured the yaw angle, pitch angle, and roll angle of the camera in the world coordinate system as shown in Table 2:

[0077] Table 2 Camera orientation data

[0078] parameter value Camera pitch angle, yaw angle, roll angle (-9.8556°，-68.3154°，0°)

[0079] Through Figure 2 The geometric coordinate transfer process shown in the figure builds the physical model as follows:

[0080]

[0081]

[0082] Where (d, a, b, c) is the camera's quaternion orientation, (x c ,y c ,z c ) is the coordinate of the camera in the world coordinate system, (x world ,y world ,z world ) is the coordinate of the target point in the world coordinate system, (α, β) is the horizontal and vertical field of view of the camera, and (w, h) is the horizontal and vertical pixel value of the screen coordinate system.

[0083] The reliability verification results of the physical transfer model are shown in Table 3:

[0084] Table 3 Reliability verification results of physical transfer model

[0085] Location Predicted value True value Error value Center Point (321,240) (320,240) 1

[0086] The resulting value is the coordinate value in the screen coordinate system.

[0087] The physical transfer model is reliable as proved by experimental data.

[0088] Through sensitivity analysis, we can screen out the error parameters with greater influence, such as Figure 3 As shown in the figure, based on the physical model, the partial derivatives of each error parameter are taken and integrated to obtain the geometric error transfer function of the visual simulation system as follows:

[0089]

[0090]

[0091] is the camera orientation error, (Δx c ,Δy c ,Δz c ) is the world coordinate system error of the camera, (Δx world ,Δy world ,Δz world ) is the world coordinate system error of the target point, and (Δα, Δβ) is the camera field of view angle error.

[0092] In the error transmission model, the error estimation range of each error parameter is shown in Table 4:

[0093] Table 4 Error parameters and their error ranges

[0094]

[0095] The error transfer function is verified and the verification results are shown in Table 5:

[0096] Table 5 Error transfer function analysis

[0097]

[0098] The error of the geometric error transfer model is calculated through experimental verification.

Claims

1. A geometric error transfer model analysis method based on a visual simulation system, characterized in that: The following steps are involved: Construct a geometric physical transfer model of the visual simulation system: take the corner points in the urban buildings as the target points, measure the world coordinate system of the target points through the measuring equipment, convert the target points from the world coordinate system to the screen coordinate system, and output the coordinates of the target points in the screen coordinate system; Select multiple geometric transfer error factors in the geometric physical transfer model of the visual simulation system, and select multiple geometric transfer error factors whose sensitivity meets the requirements through sensitivity analysis; According to the geometric transfer error factors and the construction rules of the error transfer function, the error transfer function is constructed for the geometric physical transfer model of the visual simulation system; Verify the reliability of the geometric physics transfer model through the error transfer function; The construction rule of the error transfer function is: If the error factors are independent of each other, that is, a change in one error will not cause a change in another error, then: Among them, x out Output coordinate function for the x-axis of the screen coordinate system. θ is the error factor, Δθ is the error range, is the error of the x-axis of the synthesized screen coordinate system; If there is a transfer relationship between the error factors, first calculate the error transfer function of the previous layer, that is, the error that is generated first in the physical transfer relationship, and then use it as the input error of the next layer, that is, the error that is generated later in the physical transfer relationship, and calculate the total error transfer function. The format is as follows: in, The error of the previous layer during the transmission process, is the final output error, x final is the final output function; The verification of whether the geometric physical transfer model is reliable by using the error transfer function is specifically as follows: The parameters of the building target points collected by the measuring equipment in real time are input into the geometric and physical transfer model as reference values ​​to obtain the reference coordinates of the building in the screen coordinate system. The error is added to the reference value and input into the geometric and physical transfer model to obtain the actual coordinates in the screen coordinate system after the error is added. The reference coordinates are subtracted from the actual coordinates to obtain the actual error. The error and the reference value are input into the error transfer function to obtain the estimated error. The deviation between the estimated error and the actual error is compared. If the deviation is less than the threshold, the geometric and physical transfer model is reliable.

2. The geometric error transfer model analysis method based on the visual simulation system according to claim 1 is characterized in that: The measuring equipment includes a total station and a handheld distance meter.

3. The geometric error transfer model analysis method based on the visual simulation system according to claim 1 is characterized in that: The conversion of the target point from the world coordinate system to the screen coordinate system is specifically as follows: The target points in the city buildings are transformed from the world coordinate system to the camera coordinate system through the quaternion rotation matrix, from the camera coordinate system to the projection coordinate system, from the projection coordinate system to the normalized device coordinate system, and finally to the screen coordinate system.

4. The geometric error transfer model analysis method based on the visual simulation system according to claim 1 is characterized in that: The geometric transfer error factors include: camera field of view angle error, world coordinate system error, and camera attitude error.

5. The method for analyzing a geometric error transfer model based on a visual simulation system according to claim 1, characterized in that: The parameters include: coordinate values ​​in the world coordinate system of the camera and coordinate values ​​in the world coordinate system of the target point.

Images