CN-121981925-A - Charged particle track tracking and data acquisition method and system based on bidirectional vision

CN121981925ACN 121981925 ACN121981925 ACN 121981925ACN-121981925-A

Abstract

The invention discloses a charged particle track tracking and data acquisition method and system based on bidirectional vision, which relate to the technical field of machine vision, and comprise the steps of acquiring a two-dimensional image sequence of charged particles through a plurality of image devices, establishing a geometric transformation model containing lens distortion compensation and dynamic baseline correction for multi-view fusion, and realizing sub-pixel precision three-dimensional positioning; and (3) carrying out time sequence association based on the time stamp and the space continuity to obtain track data, calculating track quality according to the motion parameters and the field intensity information, and reversely correcting the geometric parameters to form a closed-loop optimization mechanism.

Inventors

WU WEI
HU LINYI
WANG PU
TAO YI
ZHANG WENPAN
CHEN TING
XIE YI
CHEN PINGXING

Assignees

中国人民解放军国防科技大学

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The charged particle track tracking and data collecting method based on bidirectional vision is characterized by comprising the following steps: synchronously acquiring a two-dimensional image sequence of charged particles in an electric field environment through a plurality of image acquisition devices with spatial position relations; Based on the spatial position relation of the image acquisition device, carrying out multi-view fusion transformation on the pixel coordinates of the charged particles in the two-dimensional image sequence by establishing a geometric transformation model comprising lens distortion compensation and dynamic baseline correction to obtain a position coordinate sequence of the charged particles with sub-pixel level precision in a three-dimensional space; According to the time stamp information and the space continuity constraint in the position coordinate sequence, carrying out time sequence association on three-dimensional space position coordinates at different moments to obtain three-dimensional track data of charged particles; calculating to obtain a track quality evaluation index through a stress balance relation according to the motion state parameters in the three-dimensional track data and field intensity distribution information of an electric field environment; And reversely deducing the error amount of the geometric transformation parameters in the geometric transformation model based on the track quality evaluation index and the physical constraint relation, correcting the geometric transformation parameters, synchronously applying the corrected geometric transformation parameters to the multi-view fusion transformation and the time sequence to update a position coordinate sequence, and forming a closed-loop image processing optimization mechanism.
2. The method of claim 1, wherein building a geometric transformation model comprising lens distortion compensation and dynamic baseline correction comprises: Synchronously solving lens distortion parameters and relative position posture parameters of a plurality of image acquisition devices by acquiring a calibration image sequence containing known spatial position calibration points, integrating the lens distortion parameters and the relative position posture parameters into a unified parameter matrix, and establishing a geometric transformation model for coupling lens distortion compensation and baseline correction; The pixel position change of the standard point in the plurality of image acquisition devices is monitored in real time, and when the pixel position change is detected to exceed a preset change threshold value, the relative position change among the plurality of image acquisition devices is judged; and calculating the re-projection errors of the calibration points under different viewing angles based on the pixel position changes, and carrying out joint optimization updating on lens distortion parameters and relative position posture parameters in the unified parameter matrix by minimizing the re-projection errors to obtain an updated geometric transformation model.
3. The method of claim 1, wherein performing a multi-view fusion transform on the charged-particle pixel coordinates in the two-dimensional image sequence to obtain a sequence of position coordinates of the charged particles in three-dimensional space with sub-pixel level precision comprises: carrying out distortion correction and visual angle transformation on pixel coordinates of charged particles in the two-dimensional image sequence by using the geometric transformation model to obtain corrected pixel coordinates corresponding to each image acquisition device; Acquiring field intensity direction information of an electric field environment, calculating an included angle between the respective observation directions and the field intensity directions of a plurality of image acquisition devices, and distributing confidence degree weights to each image acquisition device according to the included angle; weighting and fusing the corrected pixel coordinates according to the confidence coefficient weight, and solving a candidate position coordinate set of the charged particles in a three-dimensional space through multi-view geometric constraint; Based on the motion physical constraint of the charged particles under the action of electric field force in an electric field environment, calculating the consistency measure between the electric field force direction corresponding to each candidate position coordinate in the candidate position coordinate set and the motion direction of the charged particles, eliminating the candidate position coordinate with the consistency measure lower than a consistency threshold value, and reserving the candidate position coordinate meeting the stress balance condition as an optimal position coordinate; and carrying out time sequence interpolation and sub-pixel level refinement on the optimal position coordinates to obtain a position coordinate sequence of the charged particles with sub-pixel level precision in a three-dimensional space.
4. A method according to claim 3, wherein the steps of obtaining field strength direction information of the electric field environment, calculating an angle between the respective observation direction of the plurality of image capturing devices and the field strength direction, and assigning a confidence weight to each image capturing device according to the angle comprise: Acquiring local field intensity vectors of positions of charged particles in the electric field environment, extracting the direction of the local field intensity vectors as field intensity direction information, and respectively calculating included angles between the optical axis directions of the plurality of image acquisition devices and the field intensity directions; determining an angle weight coefficient according to a numerical interval of the included angle, wherein the angle weight coefficient monotonically increases along with the increase of the included angle when the included angle is larger than a preset vertical angle threshold value, and monotonically decreases along with the decrease of the included angle when the included angle is smaller than or equal to the preset vertical angle threshold value; and carrying out combination operation on the angle weight coefficient and the image quality parameter to obtain an initial confidence coefficient weight, and carrying out normalization processing on the initial confidence coefficient weight to obtain a confidence coefficient weight.
5. The method of claim 1, wherein the step of time-sequentially correlating three-dimensional spatial position coordinates at different times according to time stamp information and spatial continuity constraints in the sequence of position coordinates to obtain three-dimensional trajectory data of the charged particles comprises: extracting time stamp information from the position coordinate sequence, and sequencing according to a time stamp order to obtain a time sequence arranged position coordinate sequence; Calculating the space distance and the speed variation between adjacent position coordinates in the time sequence arranged position coordinate sequence, and judging that the track is broken when the space distance exceeds a position deviation threshold calculated based on the motion speed or the speed variation exceeds an acceleration threshold calculated based on the electric field force direction; For the coordinates with track fracture, calculating interpolation position coordinates between fracture moments based on the speed direction of the position coordinates before fracture and the acting direction of the electric field force, and inserting the interpolation position coordinates into the position coordinate sequence of the time sequence arrangement; Calculating the similarity between the spatial position distribution and the motion direction among different track segments in the repaired time sequence arranged position coordinate sequence, splicing the track segments belonging to the same charged particle according to the time stamp sequence when the similarity is higher than the association threshold value, and performing smoothing filter processing on the spliced position coordinate sequence to obtain the three-dimensional track data of the charged particle.
6. The method according to claim 1, wherein the step of calculating the track quality evaluation index by a force balance relation according to the motion state parameter in the three-dimensional track data and the field intensity distribution information of the electric field environment comprises: extracting position coordinates and time stamp information of the charged particles at different moments from the three-dimensional track data, calculating the instantaneous speed of the charged particles by means of difference between the position coordinates at adjacent moments, and calculating to obtain instantaneous acceleration; Acquiring field intensity distribution information of the electric field environment, and inquiring local field intensity vectors corresponding to the positions of the charged particles according to the position coordinate sequence; Calculating theoretical acceleration of the charged particles according to the local field intensity vector based on the stress balance relation of the charged particles under the action of electric field force, gravity and air resistance in the electric field environment; And calculating the deviation between the instantaneous acceleration and the theoretical acceleration as a track quality evaluation index.
7. The method of claim 1, wherein the steps of inversely deriving the error amount of the geometric transformation parameters in the geometric transformation model based on the track quality assessment index and physical constraint relation, correcting the geometric transformation parameters, synchronously applying the corrected geometric transformation parameters to the multi-view fusion transformation to correlate with the time sequence to update a position coordinate sequence, and forming a closed-loop image processing optimization mechanism comprise: When the track quality evaluation index exceeds a preset quality threshold, judging that errors exist in geometric transformation parameters in the geometric transformation model; Establishing a physical constraint equation based on a stress balance relation of the charged particles in an electric field environment, wherein the physical constraint equation describes an association relation between a motion state parameter and a position coordinate of the charged particles; substituting the track quality evaluation index into the physical constraint equation, calculating the error amount of the position coordinate sequence through reverse deduction, and deducing the error amount of the geometric transformation parameters in the geometric transformation model based on the error amount of the position coordinate sequence; correcting the geometric transformation parameters based on the geometric transformation parameters to obtain corrected geometric transformation parameters; Synchronously applying the corrected geometric transformation parameters to multi-view fusion transformation of charged particle pixel coordinates in the two-dimensional image sequence, recalculating to obtain an updated position coordinate sequence, and applying the updated position coordinate sequence to the time sequence association to obtain updated three-dimensional track data; and recalculating a track quality evaluation index based on the updated three-dimensional track data to form a closed-loop image processing optimization mechanism.
8. A bi-directional vision based charged particle trajectory tracking and data acquisition system for implementing the method of any of the preceding claims 1-7, comprising: The first unit is used for synchronously acquiring a two-dimensional image sequence of the charged particles in an electric field environment through a plurality of image acquisition devices with spatial position relations; The second unit is used for carrying out multi-view fusion transformation on the pixel coordinates of the charged particles in the two-dimensional image sequence by establishing a geometric transformation model comprising lens distortion compensation and dynamic baseline correction based on the spatial position relation of the image acquisition device to obtain a position coordinate sequence of the charged particles with sub-pixel level precision in a three-dimensional space; A third unit, configured to perform time sequence association on three-dimensional spatial position coordinates at different moments according to time stamp information and space continuity constraint in the position coordinate sequence, so as to obtain three-dimensional trajectory data of the charged particles; a fourth unit, configured to calculate, according to the motion state parameter in the three-dimensional track data and the field intensity distribution information of the electric field environment, a track quality evaluation index through a stress balance relationship; And a fifth unit, configured to reversely derive an error amount of a geometric transformation parameter in the geometric transformation model based on the track quality evaluation index and a physical constraint relation, correct the geometric transformation parameter, and apply the corrected geometric transformation parameter to the multi-view fusion transformation in synchronization with the time sequence to update a position coordinate sequence, so as to form a closed-loop image processing optimization mechanism.
9. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 7.
10. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 7.

Description

Charged particle track tracking and data acquisition method and system based on bidirectional vision Technical Field The invention relates to a machine vision technology, in particular to a charged particle track tracking and data acquisition method and system based on bidirectional vision. Background The conventional charged particle track tracking technology mainly relies on a single-view imaging and fixed-parameter image processing method, and is difficult to cope with the tracking requirement of high-speed particle motion in a complex electric field environment. The three problems of the prior art are that the three-dimensional space positioning accuracy is insufficient due to lack of depth information in single-view imaging, particularly, the tracking effect is poor when charged particles move along the optical axis direction, the relative position change of the device caused by environmental disturbance is not considered in a geometric transformation model, the positioning accuracy is obviously reduced due to geometric parameter drift in the long-time working process, and the trajectory optimization and parameter correction cannot be performed by utilizing the motion rule of the charged particles in an electric field due to lack of physical constraint and organic combination of image processing. Disclosure of Invention The invention provides a charged particle track tracking and data acquisition method and system based on bidirectional vision, which can solve the problems in the prior art. In a first aspect of an embodiment of the present invention, a method for tracking and collecting trajectories of charged particles based on bidirectional vision is provided, including: synchronously acquiring a two-dimensional image sequence of charged particles in an electric field environment through a plurality of image acquisition devices with spatial position relations; Based on the spatial position relation of the image acquisition device, carrying out multi-view fusion transformation on the pixel coordinates of the charged particles in the two-dimensional image sequence by establishing a geometric transformation model comprising lens distortion compensation and dynamic baseline correction to obtain a position coordinate sequence of the charged particles with sub-pixel level precision in a three-dimensional space; According to the time stamp information and the space continuity constraint in the position coordinate sequence, carrying out time sequence association on three-dimensional space position coordinates at different moments to obtain three-dimensional track data of charged particles; calculating to obtain a track quality evaluation index through a stress balance relation according to the motion state parameters in the three-dimensional track data and field intensity distribution information of an electric field environment; And reversely deducing the error amount of the geometric transformation parameters in the geometric transformation model based on the track quality evaluation index and the physical constraint relation, correcting the geometric transformation parameters, synchronously applying the corrected geometric transformation parameters to the multi-view fusion transformation and the time sequence to update a position coordinate sequence, and forming a closed-loop image processing optimization mechanism. Optionally, establishing the geometric transformation model including lens distortion compensation and dynamic baseline correction includes: Synchronously solving lens distortion parameters and relative position posture parameters of a plurality of image acquisition devices by acquiring a calibration image sequence containing known spatial position calibration points, integrating the lens distortion parameters and the relative position posture parameters into a unified parameter matrix, and establishing a geometric transformation model for coupling lens distortion compensation and baseline correction; The pixel position change of the standard point in the plurality of image acquisition devices is monitored in real time, and when the pixel position change is detected to exceed a preset change threshold value, the relative position change among the plurality of image acquisition devices is judged; and calculating the re-projection errors of the calibration points under different viewing angles based on the pixel position changes, and carrying out joint optimization updating on lens distortion parameters and relative position posture parameters in the unified parameter matrix by minimizing the re-projection errors to obtain an updated geometric transformation model. Optionally, performing multi-view fusion transformation on the pixel coordinates of the charged particles in the two-dimensional image sequence to obtain a position coordinate sequence of the charged particles with sub-pixel level precision in a three-dimensional space includes: carrying out distortion correction and visual angle tran