Millimeter wave radar and monocular camera time calibration method based on Gaussian mixture

Abstract

The embodiment of the invention discloses a millimeter wave radar and monocular camera time calibration method based on Gaussian mixture. The method comprises the following steps: presetting a millimeter wave radar and a monocular camera; sampling the data of the millimeter wave radar and the data of the monocular camera; according to a known conversion relation between monocular camera pixel coordinates and millimeter-wave radar coordinates, monocular camera data and millimeter-wave radar data obtained through sampling are converted to a monocular camera image coordinate system, and target pixel points of the millimeter-wave radar projected in a monocular camera image are obtained; tracking the target pixel points, and rejecting background images which are not hit on the target image by the millimeter wave radar by using an MOG Gaussian mixture model; target points projected by the millimeter wave radar to the target image are further screened by using the mahalanobis distance; estimating the time delay of the millimeter wave radar and the monocular camera; and optimizing the time delay estimation value to realize time calibration of the millimeter wave radar and the monocular camera.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous perception of mobile search and rescue robots. More specifically, it relates to a time calibration method for millimeter-wave radar and monocular camera based on Gaussian mixture model. Background Art

[0002] At the scene of accident disasters, there are often factors such as toxic gases, dust, thick smoke, and lack of oxygen, which pose severe challenges to personnel search and rescue, fire fighting, and emergency rescue work. Manual rescue faces problems such as insufficient information acquisition and easy personal injury. With the development of artificial intelligence technology and robot technology, it has become a top priority to use mobile robots to replace or assist rescue personnel to enter the rescue environment first. However, due to the severity and uncertainty of the search and rescue environment, the autonomous perception system of mobile robots faces severe challenges. For such working scenarios, the integrated perception of millimeter-wave radar and camera has many advantages. The millimeter-wave radar has the advantages of small size, stable detection performance, and being unaffected by factors such as smoke and dust. Combining the advantages of high resolution and rich scene information of visual sensors, the integrated perception of millimeter-wave radar and camera has become the top priority of the mobile robot search and rescue system.

[0003] Joint calibration of millimeter-wave radar and camera is the basis for both to achieve the integrated perception task, and the calibration accuracy of both directly determines the accuracy of the integrated perception. However, due to the installation differences in the spatial positions of the millimeter-wave radar and camera and the time delay problem of multi-source data reception, it is necessary to perform joint spatial calibration and time calibration on both to achieve a consistent interpretation of the perception task. Furthermore, since millimeter-wave radar data is usually relatively discrete and easily affected by factors such as noise and irrelevant background, higher requirements are put forward for the joint calibration method of millimeter-wave radar and monocular camera. The research on the spatial calibration technology of millimeter-wave radar and monocular camera has been relatively mature, and the present invention focuses on solving the time calibration problem between the two. Summary of the Invention

[0004] The purpose of the present invention is to provide a time calibration method for millimeter-wave radar and monocular camera based on Gaussian mixture model to solve at least one of the problems existing in the prior art.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] The first aspect of the present invention provides a time calibration method for millimeter-wave radar and monocular camera based on Gaussian mixture model, including:

[0007] Pre-set the millimeter-wave radar and the monocular camera;

[0008] Sample the data of the millimeter-wave radar and the data of the monocular camera;

[0009] Convert the sampled monocular camera data and millimeter-wave radar data into the monocular camera image coordinate system using the known external parameter conversion relationship between the monocular camera and the millimeter-wave radar, and obtain the target pixel points of the millimeter-wave radar projected on the monocular camera image;

[0010] Track the target pixel points, and use the MOG Gaussian mixture model to remove the background image on the millimeter-wave radar non-hit target image;

[0011] Use the Mahalanobis distance to screen the target points projected by the millimeter-wave radar onto the target image, and obtain the pixel position state representations of the two sensors respectively.

[0012] Estimate the time delay between the millimeter-wave radar and the monocular camera;

[0013] Optimize the time delay estimation value to achieve time calibration between the millimeter-wave radar and the monocular camera.

[0014] Optionally, the preset millimeter-wave radar and monocular camera include: setting the monocular camera and the millimeter-wave radar to have a co-visual relationship, setting a moving target in the overlapping field of view of the monocular camera and the millimeter-wave radar, and the target continuously moves to multiple positions when the millimeter-wave radar and the monocular camera have an overlapping field of view;

[0015] Optionally, the sampling of the millimeter-wave radar data and the monocular camera data includes sampling the monocular camera and millimeter-wave data at the same timestamp starting point, where m millimeter-wave radar coordinate points and n monocular camera pixel coordinate points are obtained in each sampling period.

[0016] Optionally, the estimation method includes selecting the state value of the first sensor as the interpolation starting point; recording the pixel position state of the first sensor at each measurement time, using the time delay of the two sensors as the current estimation value, interpolating the state of the second sensor, and obtaining the position state of the second sensor at the sampling moment, thereby estimating the time delay between the millimeter-wave radar and the monocular camera.

[0017] Optionally, the conversion relationship between the monocular camera pixel coordinates and the millimeter-wave radar coordinates is

[0018]

[0019] where u and v are the coordinate values of the millimeter-wave radar target point converted into the monocular camera image coordinate system; λ is the scale coefficient; is the internal parameter of the monocular camera, and the internal parameter of the monocular camera is a known parameter; is the external parameter between the millimeter-wave radar and the monocular camera, and the external parameter between the millimeter-wave radar and the monocular camera is a known parameter; X W 、YW and Z W are the coordinate values of the millimeter-wave radar target points in the world coordinate system.

[0020] Optionally, the calculation formula for using the MOG (Mixture of Gaussian) model to eliminate the points that the millimeter-wave radar does not hit the target is

[0021]

[0022] where x(t) is the pixel coordinate value of the point projected by the millimeter-wave radar onto the monocular camera image at time t; TG is the foreground image; NG is the background image; p(TG) is the probability that the millimeter-wave radar point hits the foreground image; p(x(t)|TG) is the probability that the point hits both the foreground image and the target image of the target; p(NG) is the probability that the millimeter-wave radar point hits the background image; p(x(t)|NG) is the probability that the millimeter-wave radar point hits the background image but does not hit the target of the target.

[0023] Optionally, the calculation formula for further screening the target points projected by the millimeter-wave radar onto the target image of the target using the Mahalanobis distance is

[0024]

[0025] where is the Mahalanobis distance between the target point projected by the millimeter-wave radar onto the target image of the target and the target image in the monocular camera; and are the mean and variance of the i-th MOG mixture Gaussian model on the target image in the monocular image, respectively; (*) T represents the transpose;

[0026] If then the screening of the target points of the millimeter-wave radar and the monocular camera is completed.

[0027] Optionally, the estimation of the time delay between the millimeter-wave radar and the monocular camera includes the following steps:

[0028] Select the sensor with the slower frame rate of the millimeter-wave radar and the monocular camera as the anchor point. In each step of the iterative optimization, use the time delay t d of the millimeter-wave radar and the monocular camera

[0029] Let t τ be the current estimated moment, t k+1 ≤t τ <t k+1 , x(t τ ) follows a Gaussian distribution, and the pixel position state of the anchor sensor is For the pixel position state of the second sensor The formula for interpolation is

[0030]

[0031] f(t τ ) = ∧(t τ , t k ) - g(t τ ) ∧(t k+1 , t k );

[0032]

[0033] Where ∧(t k , s) is the state transition matrix; ∧(t τ , t k ) is the state transition matrix from time t τ to time t k ; ∧(t k+1 , t k ) is the state transition matrix from time t k+1 to time t k ; is the system matrix; Q C is the covariance matrix of the pixel position state; Q k,τ is the covariance matrix of the state from the k-th pixel position point to the current τ pixel position point; is the inverse of the covariance matrix of the state from the k-th pixel position point to the (k + 1)-th pixel position point; ∧(t k+1 , t τ ) T is the transpose of the state transition matrix from time t k+1 to time t τ ; f(t τ ) and g(t τ ) are the intermediate functions of interpolation respectively; and follow a Gaussian process with respect to the continuous time t, k is the k-th pixel position point, τ is the current pixel position point, and s is the integration independent variable; Δt = t k - t τ ;

[0034] Let the measurement time of the anchor sensor be Where is the total measurement time of the anchor sensor, then the time delay estimate value is calculated by the formula

[0035]

[0036] Where, is the total measurement time of the aiming sensor; represents obtaining the minimum time delay estimation value The minimum time delay value required.

[0037] The second aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. The characteristic is that when the processor executes the program, it implements the method provided by the first aspect of the present invention.

[0038] The third aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored. The characteristic is that when the program is executed by a processor, it implements the method provided by the first aspect of the present invention.

[0039] The beneficial effects of the present invention are as follows:

[0040] (1) When calibrating the time of the millimeter-wave radar and the monocular camera, the present invention can eliminate the influence of irrelevant backgrounds and discrete point interference points of the millimeter-wave radar, improving the time calibration accuracy. This method is also applicable to the spatial synchronization calibration of the millimeter-wave radar and the monocular camera.

[0041] (2) The present invention uses the MOG method for continuous state estimation, does not rely on accurate initial values, and users can query the data state of the sensor at any time of interest.

[0042] (3) The present invention uses the Mahalanobis distance to evaluate the relationship between the projection points of the millimeter-wave radar on the target image and the pixel distribution on the target image, solving the influence of different measurement scales of the two sensors on time calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The following further details the specific embodiments of the present invention with reference to the accompanying drawings.

[0044] Figure 1 Shows the flowchart of the time calibration method for millimeter-wave radar and monocular camera based on Gaussian mixture in the embodiments of the present invention.

[0045] Figure 2 Shows the schematic diagram of the data sampling device of the time calibration method for millimeter-wave radar and monocular camera based on Gaussian mixture in the embodiments of the present invention.

[0046] Figure 3 Shows the time calibration target device diagram used in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] To more clearly illustrate the present invention, the following combines embodiments and appendices Figures 1-3A further description of the present invention is provided. Similar components in the drawings are denoted by the same reference numerals. Those skilled in the art should understand that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the protection scope of the present invention.

[0048] The present invention uses a moving target for time calibration of a millimeter-wave radar and a monocular camera, which only requires that the state of the target is moving and continuous, and is independent of the type, number, and arrangement position of the target. Therefore, the present invention can realize automatic time calibration of the millimeter-wave radar and the monocular camera.

[0049] As Figure 1 shown, the present invention provides a time calibration method for a millimeter-wave radar and a monocular camera based on a mixture of Gaussians, and its implementation steps are as follows:

[0050] Pre-set the millimeter-wave radar and the monocular camera;

[0051] Sample the data of the millimeter-wave radar and the data of the monocular camera;

[0052] Use the known external parameter conversion relationship between the monocular camera and the millimeter-wave radar to convert the sampled monocular camera data and millimeter-wave radar data into the monocular camera image coordinate system, and obtain the target pixel points projected by the millimeter-wave radar in the monocular camera image;

[0053] Track the target pixel points, and use the MOG mixture of Gaussians model to remove the background image where the millimeter-wave radar does not hit the target image;

[0054] Use the Mahalanobis distance to screen the target points projected by the millimeter-wave radar onto the target image, and obtain the pixel position state representations of the two sensors respectively.

[0055] Estimate the time delay between the millimeter-wave radar and the monocular camera;

[0056] Use the LM (Levenberg-Marquarelt) algorithm to optimize the time delay estimation value to achieve time calibration of the millimeter-wave radar and the monocular camera.

[0057] In a possible implementation manner, in step S1, the millimeter-wave radar and the monocular camera are pre-set. First, it is necessary to ensure that the monocular camera and the millimeter-wave radar have sufficient co-visibility. The target can move continuously to multiple positions while ensuring sufficient overlapping fields of view of the millimeter-wave radar and the monocular camera.

[0058] Set a moving target within the overlapping field of view of the two. In the same way of the present invention, multiple moving targets recognizable by the millimeter-wave radar and the monocular camera in the natural scene can also be used for time calibration, and the automatic time calibration of the millimeter-wave radar and the monocular camera can be achieved through this setting.

[0059] In a possible implementation manner, the data sampling device of the time calibration method for the millimeter-wave radar and the monocular camera based on the Gaussian mixture is as Figure 2 shown, including the robot 1, the millimeter-wave radar 2, the monocular camera 3, the corner reflector target 4, the field of view angle range 5 of the monocular camera, the field of view angle range 6 of the millimeter-wave radar, and the range 7 where the target can move.

[0060] In a possible implementation manner, the time calibration target device adopted in the embodiment of the present invention is a radar corner reflector. Since the millimeter-wave radar data is relatively discrete and is easily affected by factors such as noise to generate irrelevant measurement points, in order to more effectively verify the embodiment of the present invention, the present invention provides a low-cost time calibration target for the millimeter-wave radar and the monocular camera, using the radar corner reflector as the calibration target. The lengths of the three outer sides a of the radar corner reflector are equal, the lengths of the three inner sides c are equal, and the three inner sides must be perpendicular to each other in pairs; b is the mounting base for fixing the radar corner reflector on the triangular bracket.

[0061] In one embodiment, as Figure 3 shown, the length of side a of the radar corner reflector is 424.2 mm, the length of side c is 300 mm, and the corner reflector is fixed on the triangular bracket. The lengths of side a and side c are not limited, as long as the millimeter-wave radar and the monocular camera can accurately and effectively detect the corner reflector.

[0062] In a possible implementation manner, in step S2, the monocular camera data and the millimeter-wave data are sampled at the same timestamp starting point. At the same time, the present invention does not require the monocular camera and the millimeter-wave radar to have the same sampling frequency. In each sampling period of the monocular camera data and the millimeter-wave data, m millimeter-wave radar coordinate points and n monocular camera pixel coordinate points are obtained. Preferably, in this embodiment, the millimeter-wave radar publishing frequency is 10 Hz, the monocular camera publishing frequency is 30 Hz, and a low data sampling frequency is selected as the starting point. Therefore, the sampling period is 0.1 s, and at least 50 frames of millimeter-wave radar and monocular camera data are collected.

[0063] In a possible implementation, in step S3, the millimeter-wave radar and the monocular camera data are spatially synchronized; the m millimeter-wave radar coordinate points sampled in step S2 are used to convert both the monocular camera and the millimeter-wave radar data to the monocular camera image coordinate system by using the known external parameter conversion relationship between the monocular camera and the millimeter-wave radar, and the target pixel points projected by the millimeter-wave radar in the monocular camera image are obtained. The specific calculation method is as follows: The conversion relationship between the monocular camera pixel coordinates and the millimeter-wave radar coordinates is

[0064]

[0065] where u and v are the coordinate values of the millimeter-wave radar target point converted to the monocular camera image coordinate system; λ is the scale coefficient; is the internal parameter of the monocular camera, and the internal parameter of the monocular camera is a known parameter; is the external parameter between the millimeter-wave radar and the monocular camera, and the external parameter between the millimeter-wave radar and the monocular camera is a known parameter; X W 、Y W and Z W are the coordinate values of the millimeter-wave radar target point in the world coordinate system.

[0066] In a possible implementation, in step S4, the background image where the millimeter-wave radar does not hit the target image of the target is removed, and the steps are as follows:

[0067] Let x(t) represent the pixel value of the point projected by the millimeter-wave radar onto the monocular camera image at time t, TG represent the foreground image, and NG represent the background image, then:

[0068] If the probability of correctly hitting the target is:

[0069]

[0070] Then it means that the millimeter-wave point hits the image of the target, and the millimeter-wave projection point x’(t) is selected as the alternative point for the time calibration of the millimeter-wave radar and the monocular camera this time. Thus, the background removal outside the target of the millimeter-wave radar and the monocular camera is completed, and the preliminary screening of the data points required by the present invention is realized.

[0071] In a possible implementation, the further screening calculation method for the target points projected by the millimeter-wave radar onto the target image in step S5 is:

[0072]

[0073] where: represents the Mahalanobis distance between the target point projected by the millimeter-wave radar onto the target image and the target image in the monocular camera; and respectively represent the mean and variance of the i MOG Gaussian mixture models on the target image in the monocular image. In the present invention, i = 3 is taken.

[0074] If it means that the millimeter-wave radar point hits the monocular camera image plane correctly. Thus, the further screening of the target points of the millimeter-wave radar and the monocular camera is completed.

[0075] In a possible implementation manner, the step of estimating the time delay between the millimeter-wave radar and the monocular camera in step S6 includes:

[0076] Step S61: Select the sensor with a slower frame rate of the millimeter-wave radar and the monocular camera as the anchor point. In each step of iterative optimization, use the time delay t d of the millimeter-wave radar and the monocular camera

[0077] as the current estimate; τ Let t k+1 ≤t τ <t k+1 , x(t τ ) follows a Gaussian distribution, and the pixel position state of the anchor sensor is The formula for interpolating the pixel position state of the second sensor is:

[0078]

[0079] f(t τ ) = ∧(t τ , t k ) - g(t τ )∧(t k+1 , t k );

[0080]

[0081] where ∧(t k , s) is the state transition matrix; ∧(t τ , t k ) is the state transition matrix from time t τ to time t k ;

[0082] ∧(t k+1 , t k ) is the state transition matrix from time t k+1 to time t k ; is the system matrix; Q C is the covariance matrix of the pixel position state; Q k,τis the covariance matrix of the state from the k-th pixel position to the current τ pixel position; is the inverse of the covariance matrix of the state from the k-th pixel position to the (k + 1)-th pixel position; ∧(t k+1 ,t τ ) T is the transpose of the state transition matrix from time t k+1 to time t τ ; f(t τ ) and g(t τ ) are intermediate functions for interpolation respectively; and follow a Gaussian process with respect to continuous time t, k is the k-th pixel position, τ is the current pixel position, and s is the integration independent variable; Δt = t k - t τ ;

[0083] Step S63: Estimate the time delay through a minimization criterion, and the calculation method is as follows

[0084]

[0085] where, is the total measurement time of the aiming sensor; represents obtaining the minimum time delay estimate value The required minimum time delay value.

[0086] Step S7, time calibration optimization of the millimeter-wave radar and the monocular camera; Use the LM (Levenberg-Marquardt) algorithm to minimize and optimize the time delay estimation values of the millimeter-wave radar and the monocular camera, thereby realizing the time calibration of the millimeter-wave radar and the monocular camera.

[0087] When calibrating the time of the millimeter-wave radar and the monocular camera, the present invention can eliminate the influence of irrelevant backgrounds and discrete interference points of the millimeter-wave radar, improve the time calibration accuracy, and this method is also applicable to the spatial synchronization calibration of the millimeter-wave radar and the monocular camera. The present invention uses the MOG method for continuous state estimation, does not require relying on accurate initial values, and users can query the data state of the sensor at any time of interest. The present invention uses the Mahalanobis distance to evaluate the relationship between the projection points of the millimeter-wave radar on the target image and the pixel distribution on the target image, and solves the influence of different measurement scales of the two sensors on time calibration.

[0088] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "upper" and "lower" is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present invention. Unless otherwise clearly specified and defined, the terms "installed", "connected" and "coupled" shall be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0089] It should also be noted that in the description of the present invention, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0090] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is impossible to list all the implementation manners here. Any obvious changes or variations derived from the technical solutions of the present invention still fall within the protection scope of the present invention.

Claims

1. A millimeter wave radar and monocular camera time calibration method based on mixed Gaussian, characterized in that: include: Pre-set millimeter wave radar and monocular camera; Sampling the data of millimeter wave radar and monocular camera; The sampled monocular camera data and millimeter-wave radar data are converted to the monocular camera image coordinate system using the known monocular camera pixel coordinate and millimeter-wave radar coordinate conversion relationship to obtain the target pixel point projected by the millimeter-wave radar on the monocular camera image. Track the target pixel points and use the MOG mixed Gaussian model to remove the background image on the target image that is not hit by the millimeter-wave radar; The target points projected onto the target image by the millimeter-wave radar are further screened using the Mahalanobis distance; Estimate the time delay between millimeter wave radar and monocular camera; The time delay estimate is optimized to achieve time calibration of the millimeter wave radar and the monocular camera.

2. The method according to claim 1, characterized in that The pre-setting of the millimeter-wave radar and the monocular camera includes: setting the monocular camera and the millimeter-wave radar to have a common viewing relationship, setting a moving target within the overlapping field of view of the monocular camera and the millimeter-wave radar, and the target continuously moving to multiple positions when the millimeter-wave radar and the monocular camera have an overlapping field of view.

3. The method according to claim 1, characterized in that The sampling of the millimeter wave radar data and the monocular camera data includes sampling the monocular camera and millimeter wave data at the same timestamp starting point, wherein m millimeter wave radar coordinate points and n monocular camera pixel coordinate points are obtained in each sampling period.

4. The method according to claim 1, characterized in that: The estimating of the time delay of the millimeter-wave radar and the monocular camera includes selecting the state value of the first sensor as the interpolation starting point; recording the pixel position state of the first sensor at each measurement time, using the time delay of the two sensors as the current estimated value, interpolating the state of the second sensor, and obtaining the position state of the second sensor at the sampling moment, thereby estimating the time delay of the millimeter-wave radar and the monocular camera.

5. The method according to claim 1, characterized in that The conversion relationship between the monocular camera pixel coordinates and the millimeter wave radar coordinates is: Among them, u and v are the coordinate values ​​of the millimeter wave radar target point converted to the monocular camera image coordinate system; λ is the scale factor; is the internal parameter of the monocular camera, which is a known parameter; is the external parameter between the millimeter-wave radar and the monocular camera, and the external parameter between the millimeter-wave radar and the monocular camera is a known parameter; X W , Y W and Z W is the coordinate value of the millimeter wave radar target point in the world coordinate system.

6. The method according to claim 5, characterized in that The calculation formula for using the MOG mixed Gaussian model to eliminate points that the millimeter-wave radar does not hit the target is: Wherein, x(t) is the pixel coordinate value of the point projected by the millimeter-wave radar onto the monocular camera image at time t; TG is the foreground image; NG is the background image; p(TG) is the probability that the millimeter-wave radar point hits the foreground image; p(x(t)|TG) is the probability that the point hits both the foreground image and the target image; p(NG) ​​is the probability that the millimeter-wave radar point hits the background image; p(x(t)|NG) is the probability that the millimeter-wave radar point hits the background image but not the target.

7. The method according to claim 6, characterized in that The calculation formula for further screening the target points projected onto the target image by the millimeter wave radar using the Mahalanobis distance is: in, is the Mahalanobis distance between the target point projected by the millimeter-wave radar on the target image and the target image in the monocular camera; and are the mean and variance of the i MOG mixed Gaussian models on the target image in the monocular image; (*) T represents transpose; like This completes the screening of target points for millimeter-wave radar and monocular camera.

8. The method according to claim 7, characterized in that The estimating of the time delay of the millimeter wave radar and the monocular camera comprises the following steps: The millimeter-wave radar and monocular camera sensors with slower frame rates are selected as anchor points. In each step of iterative optimization, the time delay t between the millimeter-wave radar and the monocular camera is used. d As a current estimate; Assume t τ is the current estimated time, t k+1 ≤t τ <t k+1 ,x(t τ ) obeys Gaussian distribution, and the pixel position state of the anchor sensor is The pixel position status of the second sensor The formula for interpolation is f(t τ )=∧(t τ ,t k )-g(t τ )∧(t k+1 ,t k ); Among them, ∧(t k ,s) is the state transfer matrix; ∧(t τ ,t k ) is t τ Time to t k The state transfer matrix at the moment; ∧(t k+1 ,t k ) is t k+1 Time to t k The state transfer matrix at the moment; is the system matrix; Q C is the covariance matrix of the pixel position state; Q k,τ is the covariance matrix from the kth pixel position to the current τ pixel position state; is the inverse of the covariance matrix of the states from the kth pixel position to the k+1th pixel position; ∧(t k+1 ,t τ ) T t k+1 Time to t τ The transpose of the state transfer matrix at the moment; f(t τ ) and g(t τ ) are the intermediate functions of interpolation respectively; and It obeys the Gaussian process with respect to continuous time t, k is the kth pixel position, τ is the current pixel position, and s is the integral independent variable; Δt=t k -t τ ; Assume the measurement time of the anchor sensor is τ=1,2,…N,where is the total measurement time of the anchor sensor, then the estimated time delay is The calculation formula is in, The total measurement time of the aiming sensor; Indicates the minimum time delay estimate The minimum time delay value required.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images