Multi-instrument tracking method for optical surgical navigation system and electronic equipment

Abstract

The invention relates to the technical field of surgical navigation, in particular to a multi-instrument tracking method and electronic equipment for an optical surgical navigation system, and the method comprises the following steps: S1, collecting and processing an image; s2, identifying a mark point; s3, performing three-dimensional reconstruction on the mark points; s4, performing instrument identification; s5, calculating a pose; s6, tracking the mark points in real time; s7, data association; according to the method, by combining triangulation and epipolar constraint conditions, the robustness of mark point reconstruction can be improved, and the precision of the system is improved; on the basis of the geometric feature matrix, multi-instrument identification and numbering in a complex scene can be realized; and through a data association algorithm, the problems of newly adding, losing or shielding of the target can be effectively solved, the dynamic adaptability of the surgical navigation system is improved, and the method can be widely applied to navigation support of complex surgical scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of surgical navigation, and in particular to a multi-instrument tracking method and electronic equipment for an optical surgical navigation system. Background Art

[0002] Surgical navigation is a widely used assistive technology in modern medicine. By real-time positioning and tracking of surgical instruments, it provides surgeons with precise navigation information within the surgical area, significantly improving surgical precision and safety. Surgical navigation systems typically rely on optical tracking technology, which uses a camera to capture markers on surgical instrument fixtures and calculate the instrument's spatial position and posture. This technology has been widely used in fields such as orthopedics and neurosurgery, becoming an indispensable tool in complex surgeries.

[0003] However, existing optical surgical navigation systems still face many challenges and shortcomings in difficult surgeries such as percutaneous puncture and orthopedic surgery. On the one hand, traditional navigation systems mainly track single instruments. When multiple instruments are present in the surgical scene, it is difficult to effectively distinguish and locate each instrument. On the other hand, complex surgical environments are subject to occlusion, rapid movement of instruments, and the addition or removal of dynamic targets. These changes often lead to a decrease in the stability and robustness of system tracking. In addition, existing methods have difficulty responding quickly to dynamic changes in multi-instrument scenarios, and their overall accuracy and applicability are not yet sufficient to meet the needs of complex surgical scenarios. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-instrument tracking method and electronic equipment for an optical surgical navigation system that can realize multi-instrument identification and solve the problems of dynamic tracking instability and insufficient adaptability to complex surgical scenes.

[0005] Based on the above purpose, the present invention adopts the following technical solutions:

[0006] A multi-instrument tracking method for an optical surgical navigation system comprises the following steps:

[0007] S1. Image acquisition and processing: Use a near-infrared binocular vision camera to acquire continuous frame images of the surgical scene and perform binarization processing on the images;

[0008] S2. Marker point recognition: The two-dimensional coordinates of each marker point on the surgical instrument fixture are extracted using the binary image centroid calculation method;

[0009] S3, 3D reconstruction of marker points: Based on the intrinsic and extrinsic parameter matrices calibrated by the binocular camera, the correspondence between the marker points in the left and right images is determined by the epipolar line constrained stereo matching algorithm, and the 3D spatial coordinates of the marker points are calculated using the triangulation formula;

[0010] S4. Instrument identification: Identify the instrument fixture based on the spatial distribution characteristics of the marking points on different surgical instrument fixtures;

[0011] S5. Posture calculation: Utilize the spatial geometric distribution information of the markers on the identified surgical instrument fixtures and combine it with the singular value decomposition method to complete the posture calculation of each surgical instrument;

[0012] S6. Real-time tracking of marker points: Construct a Kalman filter for each marker point on the surgical instrument fixture, initialize the state variables of the marker point, and perform real-time prediction and correction through the state transition model and observation matrix;

[0013] S7, Data Association: Use the Hungarian algorithm to match the detected markers with the tracked targets.

[0014] Preferably, the specific process of device identification in step S4 includes:

[0015] S41, calculate center point: calculate the center point of all marked points;

[0016] S42, marking point numbering: numbering the marking points according to the distance from the marking point to the center point;

[0017] S43, calculating the distance between the marking points: calculating the Euclidean distance between the marking points, and constructing the Euclidean distance feature vector of the marking points;

[0018] S44. Calculate the angle feature between the marker points: every three marker points can form a triangle, and calculate and store the vector angles in all triangles;

[0019] S45, constructing the topological features of the instrument fixture: constructing the topological features of the instrument fixture based on the distances between the marking points obtained in step S43 and the angle features between the marking points obtained in step S44;

[0020] S46, select model: Based on the distance and angle characteristics between the marking points, match the current instrument with the standard instrument model, and select the standard model with the smallest error as the final matching result. Preferably, the specific process of calculating the center point in step S41 includes:

[0021] Calculate the center points of all markers on the surgical instrument fixture:

[0022]

[0023] Among them, P i =(x i ,y i ,z i ) is the three-dimensional coordinate of the i-th marker point, P c is the center point of the marker;

[0024] Preferably, the specific process of marking point numbers in step S42 includes:

[0025] Calculate each marker point P i To the center point P c The Euclidean distance d i :

[0026] d i =||P i -P c ||

[0027] And according to d i The points are numbered from small to large values;

[0028] Preferably, the specific process of calculating the distance between the marking points in step S43 includes:

[0029] Compute the Euclidean distance between markers:

[0030] d ij =‖P i -P j ‖

[0031] Further construct the Euclidean distance feature vector of the marked points:

[0032] d=[d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 ]

[0033] Among them, d is the Euclidean distance feature vector of the marker point, d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 is the Euclidean distance between the four markers.

[0034] Preferably, the specific process of calculating the angle feature between the marking points in step S44 includes:

[0035] The vector angle formed by the three points is defined as:

[0036]

[0037] For four marker points, four triangles can be formed, each containing three angles, for a total of 12 angles. Since the third angle of each triangle can be calculated from the first two angles, and there are redundant angles in the four triangles, only six independent angles need to be stored. For the four marker points on the fixture, all four triangles composed of three marker points are selected. Two independent internal angles are calculated for each triangle, and the third internal angle is ignored. By eliminating redundant angles between different triangles, a feature vector containing six global independent angles is finally generated:

[0038] θ=[θ 123 ,θ 124 ,θ 134 ,θ 234 ,θ 314 ,θ 412 ]

[0039] Among them, θ is the angle feature vector of the calculated mark point, θ 123 ,θ 124 ,θ 134 ,θ 234 ,θ 314 ,θ 412 There are six independent stored angles.

[0040] Preferably, the specific process of selecting a model in step S46 includes:

[0041] Calculate the error between the distance and angle features of the currently detected fixture markers and the standard model:

[0042]

[0043] Among them, d new and θ new is the distance and angle characteristics of the currently detected fixture mark point; d k and θ k is the corresponding feature of the kth standard model; by calculating the error err(k) between the currently detected fixture and different standard models, the model with the smallest error is selected as the matching result.

[0044] Preferably, the specific process of real-time tracking of the marker point in step S6 includes:

[0045] A Kalman filter is constructed for each marker point on the surgical instrument, and the state variables of the marker point are initialized, including dynamic parameters such as position, velocity, acceleration, etc. The state vector x k Initial value of:

[0046] x k =[xyzv x v y v za x a y a z ] T

[0047] Among them, (x, y, z) is the three-dimensional position of the marker point, (v x ,v y ,v z ) is the velocity component, (a x ,a y ,a z ) is the acceleration component;

[0048] According to the state vector of the previous moment, the state of the marked point at the current moment is predicted through the state transition model:

[0049]

[0050] in, is the predicted current state, x k-1 is the state at the previous moment, A is the state transfer matrix;

[0051] The state transfer matrix A is set as follows:

[0052]

[0053] Among them, Δt is the time step, which represents the time interval from the previous moment to the current moment;

[0054] The prediction formula of the prior estimated covariance matrix at each moment is:

[0055] P k|k-1 =A·P k-1 ·A T +Q

[0056] Among them, P k-1 is the covariance matrix of the posterior state estimate at the previous moment, which represents the uncertainty of the system state estimate at the previous step; Q is the process noise matrix, which represents the uncertainty of the system prediction;

[0057] Use the actual position of the marker detected by the binocular camera as the observation value z k , calculate the residual:

[0058]

[0059] Where H is the observation matrix;

[0060] The measurement matrix H is set as follows:

[0061]

[0062] Calculate the Kalman gain:

[0063] K k =P k|k-1 ·H T (H.P k|k-1 ·H T +R) -1

[0064] Where R is the measurement noise covariance matrix;

[0065] According to the residual y k and the Kalman gain K k Revised forecast value:

[0066]

[0067] Update the covariance matrix:

[0068] P k =(IK k ·H)·P k|k-1

[0069] Among them, P k is the posterior estimated covariance matrix; the updated state is used as the input for the next moment, and the prediction and correction steps are executed cyclically to achieve real-time tracking of the markers.

[0070] Preferably, the specific process of target correction in step S7 includes:

[0071] S71, constructing a cost matrix: the elements of the cost matrix are based on the Euclidean distance cost matrix between the detected landmarks and the predicted positions;

[0072] S72, optimal allocation: optimally allocate the cost matrix obtained in step S71, with the goal of minimizing the total matching cost, and output the matching result;

[0073] S73. Correct the target: classify the target according to the matching results: for newly added targets, initialize the filter; for unmatched targets, update their status; for unmatched targets, use historical predictions for correction; consecutive unmatched targets are judged as lost and removed.

[0074] An electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements any step of any of the above-mentioned multi-instrument tracking methods for an optical surgical navigation system.

[0075] The beneficial effects of the present invention are:

[0076] The present invention combines triangulation with epipolar constraints, which can effectively improve the robustness of the marker point reconstruction process, thereby effectively improving the accuracy of surgical instrument positioning and increasing the stability and precision during surgical navigation.

[0077] The present invention is based on a geometric feature matrix and can simultaneously identify multiple different surgical instrument features, thereby being suitable for multi-surgical instrument identification and numbering in complex scenarios and capable of being used in more complex surgical scenarios.

[0078] The present invention uses the Hungarian algorithm and Kalman filter to perform real-time prediction and correction of marker points on surgical instrument fixtures, which can effectively deal with the problems of added, lost or occluded targets, and greatly improve the dynamic adaptability of the navigation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 Schematic diagram of the process of the present invention;

[0080] Figure 2 Schematic diagram of the 3D reconstruction principle of the present invention;

[0081] Figure 3 Schematic diagram of the method for numbering marking points of a surgical instrument fixture according to the present invention;

[0082] Figure 4 Schematic diagram of the topological features of the surgical instrument fixture of the present invention;

[0083] Figure 5 This is a flow chart of the Hungarian algorithm of the present invention;

[0084] Figure 6 Schematic diagram of the marker point matching and classification mechanism of the present invention. DETAILED DESCRIPTION

[0085] The following is a further explanation of the present invention with reference to specific embodiments. Figure 1 As shown, this embodiment is a multi-instrument tracking method for an optical surgical navigation system, comprising the following steps:

[0086] S1. Image acquisition and processing: Use a near-infrared binocular vision camera to acquire continuous frame images of the surgical scene and perform binarization on the images.

[0087] S2. Marker point recognition: The two-dimensional coordinates of each marker point on the surgical instrument fixture are extracted using the binary image centroid calculation method.

[0088] S3, 3D reconstruction of marker points: Based on the intrinsic and extrinsic parameter matrices calibrated by the binocular camera, the corresponding relationship between the marker points in the left and right images is determined by the epipolar line constrained stereo matching algorithm, and the 3D spatial coordinates of the marker points are calculated using the triangulation formula; Figure 2As shown in the figure, the left camera and the right camera project the spatial points onto their respective image planes, and infer the three-dimensional position based on the intersection of the two spatial lines, thereby achieving accurate three-dimensional reconstruction of the marker points.

[0089] S4. Instrument Identification: Instrument fixture identification is performed based on the spatial distribution characteristics of the marking points on different surgical instrument fixtures. The specific process includes:

[0090] S41. Calculate the center point: Calculate the center point of all marked points on the surgical instrument fixture:

[0091]

[0092] Among them, P i =(x i ,y i ,z i ) is the three-dimensional coordinate of the i-th marker point, P c is the center point of the marker;

[0093] S42, Marking point numbering: Number the marking points according to the distance from the marking point to the center point:

[0094] like Figure 3 As shown, calculate each marker point P i To the center point P c The Euclidean distance d i :

[0095] d i =||P i -P c ||

[0096] And according to d i The points are numbered from small to large values;

[0097] S43. Calculate the distance between the marker points: Calculate the Euclidean distance between the marker points and construct the Euclidean distance feature vector of the marker points:

[0098] First, calculate the Euclidean distance between the markers:

[0099] d ij =‖P i -P j ‖

[0100] Further construct the Euclidean distance feature vector of the marked points:

[0101] d=[d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 ]

[0102] Among them, d is the Euclidean distance feature vector of the marker point, d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 is the Euclidean distance between the four markers.

[0103] S44. Calculate the angle feature between the marker points: Every three marker points can form a triangle. Calculate and store the vector angles in all triangles:

[0104] The vector angle formed by the three points is defined as:

[0105]

[0106] For four marker points, four triangles can be formed, each containing three angles, for a total of 12 angles. Since the third angle of each triangle can be calculated from the first two angles, and there are redundant angles in the four triangles, only six independent angles need to be stored. For the four marker points on the fixture, all four triangles composed of three marker points are selected. Two independent internal angles are calculated for each triangle, and the third internal angle is ignored. By eliminating redundant angles between different triangles, a feature vector containing six global independent angles is finally generated:

[0107] θ=[θ 123 ,θ 124 ,θ 134 ,θ 234 ,θ 314 ,θ 412 ]

[0108] Among them, θ is the angle feature vector of the calculated mark point, θ 123 ,θ 124 ,θ 134 ,θ 234 ,θ 314 ,θ 412 There are six independent stored angles.

[0109] S45, constructing the topological features of the instrument fixture: constructing the topological features of the instrument fixture based on the distances between the marking points obtained in step S43 and the angle features between the marking points obtained in step S44;

[0110] like Figure 4 As shown in the figure, the topological features of the surgical instrument fixture are constructed based on the distance and angle features:

[0111] G=(d,θ)

[0112] S46. Model selection: Based on the distance and angle characteristics between the marker points, the current device is matched with the standard device model, and the standard model with the smallest error is selected as the final matching result:

[0113] Calculate the error between the distance and angle features of the currently detected fixture markers and the standard model:

[0114]

[0115] Among them, d new and θ new is the distance and angle characteristics of the currently detected fixture mark point; d k and θ k is the corresponding feature of the kth standard model; by calculating the error err(k) between the currently detected fixture and different standard models, the model with the smallest error is selected as the matching result to complete the identification of the type of surgical instrument.

[0116] S5. Posture calculation: Using the three-dimensional spatial distribution information of the marker points on the identified surgical instrument fixture, the singular value decomposition method is adopted to calculate the optimal rigid transformation matrix (rotation matrix and displacement vector) between the current marker point set and the standard instrument model, thereby completing the spatial posture estimation of each surgical instrument.

[0117] S6. Real-time tracking of marker points: A Kalman filter is constructed for each marker point on the surgical instrument fixture, the state variables of the marker points are initialized, and real-time prediction and correction are performed using the state transition model and observation matrix. The specific process includes:

[0118] A Kalman filter is constructed for each marker point on the surgical instrument, and the state variables of the marker point are initialized, including dynamic parameters such as position, velocity, acceleration, etc. The state vector x k Initial value of:

[0119] x k =[xyzv x v y v z a x a y a z ] T

[0120] Among them, (x, y, z) is the three-dimensional position of the marker point, (v x ,v y ,v z ) is the velocity component, (a x ,a y ,a z ) is the acceleration component;

[0121] According to the state vector of the previous moment, the state of the marked point at the current moment is predicted through the state transition model:

[0122]

[0123] in, is the predicted current state, x k-1 is the state at the previous moment, A is the state transfer matrix;

[0124] The state transfer matrix A is set as follows:

[0125]

[0126] Among them, Δt is the time step, which represents the time interval from the previous moment to the current moment;

[0127] The prediction formula of the prior estimated covariance matrix at each moment is:

[0128] P k|k-1 =A·P k-1 ·A T +Q

[0129] Among them, P k-1 is the covariance matrix of the posterior state estimate at the previous moment, which represents the uncertainty of the system state estimate at the previous step; Q is the process noise matrix, which represents the uncertainty of the system prediction;

[0130] Use the actual position of the marker detected by the binocular camera as the observation value z k , calculate the residual:

[0131]

[0132] Where H is the observation matrix;

[0133] The measurement matrix H is set as follows:

[0134]

[0135] Calculate the Kalman gain:

[0136] K k =P k|k-1 ·H T (H.P k|k-1 ·H T +R) -1

[0137] Where R is the measurement noise covariance matrix;

[0138] According to the residual y k and the Kalman gain K k Revised forecast value:

[0139]

[0140] Update the covariance matrix:

[0141] P k =(IK k ·H)·P k|k-1

[0142] Among them, P k is the posterior estimated covariance matrix; the updated state is used as the input for the next moment, and the prediction and correction steps are executed cyclically to achieve real-time tracking of the markers.

[0143] S7, Data Association: Use the Hungarian algorithm to match the detected markers with the tracked targets, including the following steps:

[0144] S71. Construct a cost matrix: The elements of the cost matrix are based on the Euclidean distance cost matrix between the detected landmarks and the predicted positions.

[0145] S72, optimal allocation: optimally allocate the cost matrix obtained in step S71, with the goal of minimizing the total matching cost; specifically, Figure 5 As shown in the figure, an assignment process based on the Hungarian algorithm is adopted, including inputting the detected cost matrix, filling the cost matrix if it is not a square matrix, and then performing row reduction, column reduction, trial assignment, covering the zero line and updating the matrix in sequence until the optimal assignment is completed and the matching result is output.

[0146] S73, Modify the target: Figure 6 As shown, based on the matching results obtained in step S72, the targets are classified. For newly detected targets, a new Kalman filter is initialized; for successfully matched targets, their state variables are updated; for unmatched targets, corrections are made using historical prediction results. If a target has not been matched for multiple consecutive frames, it is considered lost and removed from the tracking list, thus achieving dynamic management of the target set.

[0147] The above description is merely a further explanation of the present invention in conjunction with specific embodiments. All descriptions do not limit the scope of protection of the present invention. Any changes or replacements that can be easily thought of by any technician in this field within the technical scope disclosed by the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A multi-instrument tracking method for an optical surgical navigation system, characterized in that: The following steps are involved: S1. Image acquisition and processing: Use a near-infrared binocular vision camera to acquire continuous frame images of the surgical scene and perform binarization processing on the images; S2. Marker point recognition: The two-dimensional coordinates of each marker point on the surgical instrument fixture are extracted using the binary image centroid calculation method; S3, 3D reconstruction of marker points: Based on the intrinsic and extrinsic parameter matrices calibrated by the binocular camera, the correspondence between the marker points in the left and right images is determined by the epipolar line constrained stereo matching algorithm, and the 3D spatial coordinates of the marker points are calculated using the triangulation formula; S4. Instrument identification: Identify the instrument fixture based on the spatial distribution characteristics of the marking points on different surgical instrument fixtures; S5. Posture calculation: Utilize the spatial geometric distribution information of the markers on the identified surgical instrument fixtures and combine it with the singular value decomposition method to complete the posture calculation of each surgical instrument; S6. Real-time tracking of marker points: Construct a Kalman filter for each marker point on the surgical instrument fixture, initialize the state variables of the marker point, and perform real-time prediction and correction through the state transition model and observation matrix; S7, Data Association: Use the Hungarian algorithm to match the detected markers with the tracked targets.

2. The multi-instrument tracking method for an optical surgical navigation system according to claim 1, wherein: The steps The specific process of S4 device identification includes: S41, calculate center point: calculate the center point of all marked points; S42, marking point numbering: numbering the marking points according to the distance from the marking point to the center point; S43, calculating the distance between the marking points: calculating the Euclidean distance between the marking points, and constructing the Euclidean distance feature vector of the marking points; S44. Calculate the angle feature between the marker points: every three marker points can form a triangle, and calculate and store the vector angles in all triangles; S45, constructing the topological features of the instrument fixture: constructing the topological features of the instrument fixture based on the distances between the marking points obtained in step S43 and the angle features between the marking points obtained in step S44; S46. Select model: Based on the distance and angle features between the marker points, match the current device with the standard device model, and select the standard model with the smallest error as the final matching result.

3. The multi-instrument tracking method for an optical surgical navigation system according to claim 2, wherein: The specific process of calculating the center point in step S41 includes: Calculate the center points of all markers on the surgical instrument fixture: Among them, P i =(x i ,y i ,z i ) is the three-dimensional coordinate of the i-th marker point, P c The center point of the marker.

4. The multi-instrument tracking method for an optical surgical navigation system according to claim 3, wherein: The specific process of step S42 marking point numbers includes: Calculate each marker point P i To the center point P c The Euclidean distance d i : d i =||P i -P c || And according to d i The points are numbered from smallest to largest value.

5. The multi-instrument tracking method for an optical surgical navigation system according to claim 4, wherein: The specific process of calculating the distance between the marking points in step S43 includes: Compute the Euclidean distance between markers: d ij =‖P i -P j ‖ Further construct the Euclidean distance feature vector of the marked points: d=[d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 ] Among them, d is the Euclidean distance feature vector of the marker point, d 12 ,d 13 ,d 14 ,d 23 ,d 24 ,d 34 is the Euclidean distance between the four markers.

6. The multi-instrument tracking method for an optical surgical navigation system according to claim 5, wherein: The specific process of calculating the angle feature between the marking points in step S44 includes: The vector angle formed by the three points is defined as: For the four marker points on the fixture, select all four triangles formed by three marker points; calculate two independent internal angles for each triangle and ignore the third internal angle; by eliminating redundant angles between different triangles, finally generate a feature vector containing 6 global independent angles: θ=[θ 123 ,i 124 ,i 134 ,i 234 ,i 314 ,i 412 ] Among them, θ is the angle feature vector of the calculated mark point, θ 123 ,θ 124 ,θ 134 ,θ 234 ,θ 314 ,θ 412 There are six global independent angles respectively.

7. The multi-instrument tracking method for an optical surgical navigation system according to claim 6, wherein: The specific process of selecting the model in step S46 includes: Calculate the error between the distance and angle features of the currently detected fixture markers and the standard model: Among them, d new and θ new is the distance and angle characteristics of the currently detected fixture mark point; d k and θ k is the corresponding feature of the kth standard model; by calculating the error err(k) between the currently detected fixture and different standard models, the model with the smallest error is selected as the matching result.

8. The multi-instrument tracking method for an optical surgical navigation system according to claim 7, wherein: The specific process of real-time tracking of the marker points in step S6 includes: A Kalman filter is constructed for each marker point on the surgical instrument, and the state variables of the marker point are initialized, including dynamic parameters such as position, velocity, acceleration, etc. The state vector x k Initial value of: x k =[x y z v x v y v z a x a y a z ] T Among them, (x, y, z) is the three-dimensional position of the marker point, (v x ,a y ,v z ) is the velocity component, (a x ,a y ,a z ) is the acceleration component; According to the state vector of the previous moment, the state of the marked point at the current moment is predicted through the state transition model: in, is the predicted current state, x k-1 is the state at the previous moment, A is the state transfer matrix; The state transfer matrix A is set as follows: Among them, Δt is the time step, which represents the time interval from the previous moment to the current moment; The prediction formula of the prior estimated covariance matrix at each moment is: P k|k-1 =A·P k-1 ·A T +Q Among them, P k-1 is the covariance matrix of the posterior state estimate at the previous moment, which represents the uncertainty of the system state estimate at the previous step; Q is the process noise matrix, which represents the uncertainty of the system prediction; Use the actual position of the marker detected by the binocular camera as the observation value z k , calculate the residual: Where H is the observation matrix; The measurement matrix H is set as follows: Calculate the Kalman gain: K k =P k|k-1 ·H T ·(H·P k|k-1 ·H T +R) -1 Where R is the measurement noise covariance matrix; According to the residual y k and the Kalman gain K k Revised forecast value: Update the covariance matrix: P k =(I-K k ·H)·P k|k-1 Among them, P k is the posterior estimated covariance matrix; the updated state is used as the input for the next moment, and the prediction and correction steps are executed cyclically to achieve real-time tracking of the markers.

9. The multi-instrument tracking method for an optical surgical navigation system according to claim 8, wherein: The steps The specific process of S7 target correction includes: S71, constructing a cost matrix: the elements of the cost matrix are based on the Euclidean distance cost matrix between the detection markers and the predicted positions; S72, optimal allocation: optimally allocate the cost matrix obtained in step S71, with the goal of minimizing the total matching cost, and output the matching result; S73. Correct the target: classify the target according to the matching results: for newly added targets, initialize the filter; for unmatched targets, update their status; for unmatched targets, use historical predictions for correction; consecutive unmatched targets are judged as lost and removed.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, any step of the multi-instrument tracking method for an optical surgical navigation system according to any one of claims 1 to 9 is implemented.

Images