CN112595312B - Method and system for filtering pseudo star target of large-field-of-view star sensor - Google Patents
Method and system for filtering pseudo star target of large-field-of-view star sensor Download PDFInfo
- Publication number
- CN112595312B CN112595312B CN202011243133.3A CN202011243133A CN112595312B CN 112595312 B CN112595312 B CN 112595312B CN 202011243133 A CN202011243133 A CN 202011243133A CN 112595312 B CN112595312 B CN 112595312B
- Authority
- CN
- China
- Prior art keywords
- target
- sequence
- targets
- star
- frame
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/02—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by astronomical means
Abstract
The invention relates to a method and a system for filtering a pseudo star target of a large-view-field star sensor, which respectively calculate the angular distance between any two targets in a current frame and a previous frame, a current frame and a previous two frames, and a previous frame and a previous two frames and the gray level deviation between the two targets, and judge whether all the targets in the current frame are related to the targets in the previous two frames or not; aiming at the target with inter-frame correlation, carrying out one-time filtering on a pseudo-star target based on the characteristic that the angular distance between the frames of the real star target is stable and unchanged; and for the target result subjected to the primary filtering, carrying out secondary filtering on the pseudo-star target based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target, and outputting the residual target serving as the real star target. According to the method, inter-frame correlation processing is performed by combining the motion characteristics of the target, so that the pseudo star target is effectively filtered, the attitude output efficiency of the star sensor is ensured, and the use requirement of a control system is met.
Description
Technical Field
The invention relates to the technical field of astronomical navigation, in particular to a method and a system for filtering a pseudo star target of a large-view-field star sensor based on target frame correlation processing.
Background
In an astronomical photoelectric observation system, a star measurement image has the problem of pseudo star targets in different degrees, and the pseudo star targets are spot targets formed by gathering small part of energy on a detector target surface by an optical system due to the irradiation of engine plumes or non-fixed star space targets and the like by sunlight. The star sensor can filter the pseudo star target through the characteristic difference between the non-fixed star target and the real star target, but the pseudo star target with the characteristic similar to the real star target is difficult to eliminate through single-frame image processing.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a method and a system for filtering a pseudo star target of a large-view-field star sensor based on target frame correlation processing, which combine the motion characteristics of a target to carry out frame correlation processing, realize effective filtering of the pseudo star target, ensure the attitude output efficiency of the star sensor and meet the use requirements of a control system.
The technical scheme for solving the technical problems is as follows:
on one hand, the invention provides a method for filtering a pseudo star target of a large-field-of-view star sensor, which comprises the following steps:
setting targets detected by a current frame, a previous frame and a previous two frames as a sequence A, a sequence B and a sequence C respectively, calculating the angular distance between any two targets between the sequence A and the sequence B, between the sequence A and the sequence C and between the sequence B and the sequence C and the gray level deviation between the two targets respectively, and judging whether all the targets in the sequence A are related to the targets in the sequence B and the sequence C or not;
aiming at the target with inter-frame correlation, carrying out one-time filtering on a pseudo-star target based on the characteristic that the angular distance between the frames of the real star target is stable and unchanged;
and for the target result subjected to the primary filtering, carrying out secondary filtering on the pseudo-star target based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target, and outputting the residual target serving as the real star target.
Further, the calculating the angular distance between any two targets between the sequence a and the sequence B, between the sequence a and the sequence C, and between the sequence B and the sequence C, and the gray level deviation between the two targets respectively includes:
for the calculation method of the angular distance between any two targets between the sequence A and the sequence B and the gray scale deviation between the two targets,
dSeqTarget＝acos(xnow·xbefore+ynow·ybefore+znow·zbefore)
lumSeqDev＝abs(lumnow-lumbefore)
in the formula, xnow、ynow、znowVector, x, representing the sequence A object in the star sensor coordinate systembefore、ybefore、zbeforeRepresenting a vector of the sequence B target in a star sensor coordinate system, wherein the dSeqTarget is angular distance information of two frame targets; lumnowFor the gray scale of the sequence A object, lumbeforeFor the gray level of the sequence B object, lumSeqDev is the difference in gray level between the sequence A object and the sequence B object;
and calculating the angular distance between any two targets between the sequence A and the sequence C, between the sequence B and the sequence C and the gray level deviation between the two targets in the same way.
Further, the determining whether inter-frame correlation exists between all the targets in the sequence a and the targets in the sequences B and C includes:
determining a target inter-frame correlation condition according to the maximum angular velocity information of the star sensor movement, and judging and counting target pairs meeting the inter-frame correlation condition between the sequence A and the sequence B, between the sequence A and the sequence C, and between the sequence B and the sequence C according to the target inter-frame correlation condition;
for target i in sequence A, target j in sequence B, and target k in sequence C,
if i is related to j, i is related to k, and j is related to k, judging that i, j and k are related;
if i is related to j, i is related to k, and j is related to k', then i is judged to be related to j; k' is a different target from k in sequence B;
if i is associated with j, i is associated with k, and no target associated with j exists in the sequence C, it is determined that i is associated with k.
Further, the determining whether inter-frame correlation exists between all targets in the sequence a and targets in the sequences B and C further includes:
for target i in sequence A, target j in sequence B, and target k in sequence C,
if the sequence B and the sequence C do not have the target related to the i, the i has no target inter-frame correlation;
if the target related to i does not exist in the sequence B and i is only related to the target k in the sequence C, i is considered to be related to k;
if the target related to i does not exist in the sequence B, i is related to other targets in the sequence C besides k, and the target with higher possibility of being related to the inter-frame is determined as the inter-frame related target of i by weighting the target inter-frame by the gray change rate and the motion angle distance.
Further, the calculation method for determining the correlation possibility between the target frames by weighting the gray scale change rate and the motion angular distance amount between the target frames is as follows:
lumSeqProb＝1-lumSeqDev/lumbefore
dSeqProb＝1-dSeqNow/dSeqMax
prob＝a×lumSeqProb+b×dSeqProb
wherein lumSeqDev is the target inter-frame gray scale variation, lumbeforeThe gray value of a previous frame of the target is obtained, the lumSeqProb is the correlation of the target gray level, the dSeqNow is the angular distance of motion between the target frames, the dSeqMax is the maximum angular distance value of motion between the target frames, the dSeqprob is the angular distance between the target frames, the prob is the probability value of the correlation between the target frames, and a and b are the gray level change rate and the weighting coefficient of the angular distance between the frames.
Further, the determining of the target inter-frame correlation condition according to the maximum angular velocity information of the star sensor motion includes:
in the formula (d)T1A first angular distance threshold representing a relative motion between target frames; lumTIndicating the gray change rate threshold of the frames before and after the target.
Further, aiming at the target with inter-frame correlation, the pseudo star target is filtered once based on the stable and invariable characteristic of the inter-frame angular distance of the real star target, and the method comprises the following steps:
suppose the angular distance between the two targets of target i and target j in sequence A is dAijTarget i and target j in sequence BAngular separation between the standards being dBijThe second angular distance threshold is dT2；
For the object i in the sequence A, the angular distance between the object i and all other objects in the sequence A is counted to satisfy | dAij-dBij|＜dT2When the number of times is greater than or equal to the number threshold, the target is judged to be a real star target, and other targets are filtered as pseudo star targets.
For the target result subjected to the primary filtering, performing secondary filtering on a pseudo star target based on the characteristic that the inter-frame real star target motion angular distance is relatively concentrated, and outputting the residual target as a real star target, wherein the secondary filtering comprises the following steps:
calculating the amount of angular distance motion between target frames for the target result subjected to primary filtering, and arranging the target result in ascending order according to the angular distance between the frames;
counting the number of frame matching related targets in a preset angular distance interval range;
and analyzing the moving direction of the target contained in the angular distance interval range with the largest number of the frame matching related targets, eliminating the target with larger difference of the moving direction, and outputting the residual targets.
Further, the performing a pseudo star target secondary filtering on the target result subjected to the primary filtering based on the relative concentration characteristic of the inter-frame real star target motion angular distance, and outputting the remaining target as a real star target further includes:
sorting the remaining targets according to the gray value size, and outputting the sorted remaining targets as a first priority; sorting the targets with continuous tracks but not meeting the motion characteristics of the real star target according to the gray value, and outputting the targets as a second priority; and sequencing the targets with discontinuous tracks according to the gray value size, and outputting the targets as a third priority.
On the other hand, the invention also provides a system for filtering the pseudo star target of the large-field-of-view star sensor, which comprises the following steps:
the inter-frame correlation judging module is used for setting the targets detected by the current frame, the previous frame and the previous two frames as a sequence A, a sequence B and a sequence C respectively, and judging whether all the targets in the sequence A are in inter-frame correlation with the targets in the sequence B and the sequence C respectively by setting the angular distance between any two targets between the sequence A and the sequence B, between the sequence A and the sequence C and gray level deviation between the two targets;
the primary filtering module is used for carrying out primary filtering on a pseudo star target based on the characteristic that the angular distance between frames of the real star target is stable and unchanged aiming at the target with inter-frame correlation;
and the secondary filtering and outputting module is used for secondarily filtering the pseudo star target based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target for the target result subjected to primary filtering, and outputting the residual target serving as the real star target.
The invention has the beneficial effects that: the single-frame image processing is difficult to completely filter pseudo-star targets in the images, especially pseudo-star targets with similar imaging characteristics to real star targets, and the motion information of multi-frame image targets must be utilized to combine the motion characteristics of the targets with the continuity and consistency of motion tracks for consideration. The movement of the real star target has the continuity of movement and the consistency of tracks, and is regular; the motion of the noise points is random and cannot form a continuous motion track, and although the pseudo star target can form a continuous track, the motion direction and the motion speed of the pseudo star target are different from those of the real star target. After the target detection part carries out pseudo-star target filtering processing, the pseudo-star target is further filtered based on the inconsistency of the motion characteristics of the real star target and the pseudo-star target.
Drawings
FIG. 1 is a remote star map of an actual flight test of a star sensor;
FIG. 2 is a schematic diagram of filtering a pseudo-star target based on the stable and invariant angular distance characteristic of a real star target;
FIG. 3 is a demonstration result of 1 frame correlation processing of actual flight star measurement data of a certain type of star sensor;
FIG. 4 is a 2-frame correlation processing demonstration result of actual flight star measurement data of a certain type of star sensor;
FIG. 5 is a working flow of a pseudo star target filtering method of a large-field-of-view star sensor based on a frame correlation technique.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth by way of illustration only and are not intended to limit the scope of the invention.
The embodiment provides a system for filtering a pseudo star target of a large-view-field star sensor, which comprises:
the inter-frame correlation judging module is used for setting the targets detected by the current frame, the previous frame and the previous two frames as a sequence A, a sequence B and a sequence C respectively, and judging whether all the targets in the sequence A are in inter-frame correlation with the targets in the sequence B and the sequence C respectively by setting the angular distance between any two targets between the sequence A and the sequence B, between the sequence A and the sequence C and gray level deviation between the two targets;
the primary filtering module is used for carrying out primary filtering on the pseudo star target based on the stable and invariable characteristic of the angular distance between the actual star target frames aiming at the target with the inter-frame correlation;
and the secondary filtering and outputting module is used for secondarily filtering the pseudo star target based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target for the target result subjected to primary filtering, and outputting the residual target serving as the real star target.
The system performs frame correlation processing by combining the motion characteristics of the target, realizes effective filtering of a pseudo star target, ensures the attitude output efficiency of the star sensor, and meets the use requirements of a control system.
Specifically, based on the above system, an embodiment of the present invention further provides a method for filtering a pseudo star target of a large-field-of-view star sensor, as shown in fig. 5, the method specifically includes the following steps:
the method comprises the following steps: and performing interframe correlation processing on all targets of the current frame, calculating the angular distance and the gray level deviation between each target of the current frame and all targets of the previous two frames, judging whether the target is interframe correlated with the targets of the previous two frames, and when the angular distance and the gray level deviation between the target and the target of the previous two frames meet frame correlation conditions, considering that the target is interframe correlated with the targets of the previous two frames, wherein interframe correlation thresholds are determined by the camera parameters of the star sensor and angular motion characteristics.
When performing inter-frame correlation processing, for convenience of explanation, the detected targets of the current frame, the previous frame and the previous two frames are respectively set as a sequence a, a sequence B and a sequence C. In order to judge whether the target i in the sequence A is related to the target in the sequence B or not, firstly, the angular distance between the target i in the sequence A and all the targets in the sequence B and the gray level deviation between the two targets are calculated, then, whether the target is related to the target in the sequence B or not is judged, and the calculation formula of the angular distance between the two targets and the gray level deviation between the two targets is as follows:
dSeqTarget＝acos(xnow·xbefore+ynow·ybefore+znow·zbefore)
lumSeqDev＝abs(lumnow-lumbefore)
in the formula, xnow、ynow、znowVector, x, representing the sequence A object in the star sensor coordinate systembefore、ybefore、zbeforeRepresenting a vector of the sequence B target in a star sensor coordinate system, wherein the dSeqTarget is angular distance information of two frame targets; lumnowGray scale of sequence A object, lumbeforeFor the grayscale of the sequence B object, lumSeqDev is the difference in grayscale between the sequence A object and the sequence B object.
Determining a motion angular distance threshold d related to target frames according to the maximum angular velocity information of the motion of the star sensorT1And a gray change rate threshold value lum of frames before and after the targetTIf the target i in the sequence A and the target j in the sequence B meet the target inter-frame correlation condition, the two targets are considered to have inter-frame correlation, and the inter-frame correlation condition is shown as the following formula.
For each object in the sequence A, counting all objects of the sequence B and the objects i possibly related to the sequence A, and then further analyzing and processing the possible inter-frame related objects.
For the object i in the sequence a, if no object in the sequence B satisfies the inter-frame correlation condition, the processing method is as follows: performing interframe correlation on the sequence A target i and the targets in the sequence C, and if one target in the sequence C meets interframe correlation conditions, determining that the sequence A target i and the sequence C are interframe correlated; if no target in the sequence C meets the inter-frame correlation condition, the target i in the sequence A is considered to have no target inter-frame correlation; and if a plurality of targets in the sequence C are matched with the target i in the sequence A, weighting and determining the target with higher possibility of being related to the target between frames by using the gray change rate and the motion angle distance quantity between the target frames. The inter-frame correlation probability calculation formula of the weighted determination target inter-frame "gray degree change rate + motion angle distance amount" is as follows:
lumSeqProb＝1-lumSeqDev/lumbefore
dSeqProb＝1-dSeqNow/dSeqMax
prob＝a×lumSeqProb+b×dSeqProb
wherein lumSeqDev is the target inter-frame gray scale variation, lumbeforeThe gray value of a previous frame of the target is obtained, the lumSeqProb is the correlation of the target gray level, the dSeqNow is the angular distance of motion between the target frames, the dSeqMax is the maximum angular distance value of motion between the target frames, the dSeqprob is the angular distance between the target frames, the prob is the probability value of the correlation between the target frames, and a and b are the gray level change rate and the weighting coefficient of the angular distance between the frames.
For the target i in the sequence A, if the target j in the sequence B meets the inter-frame correlation condition, the processing mode is as follows: performing interframe correlation on the target i in the sequence A and the target in the sequence C, and if no target in the sequence C meets interframe correlation conditions, successfully pairing the target i in the sequence A and the target j in the sequence B; if the target k in the sequence C meets the interframe correlation condition, determining the interframe correlation condition of the target i in the sequence A according to the interframe correlation condition of the target j in the sequence B and the target in the sequence C, if the target in the sequence C has interframe correlation with the target j in the sequence B and is the target k, then the target in the sequence A is interframe correlated with both the target j in the sequence B and the target k in the sequence C, and if the target in the sequence C has interframe correlation with the target j in the sequence B but is not the target k, then the target i in the sequence A is interframe correlated with the target j in the sequence B; if no object in sequence C is inter-correlated with object j in sequence B, then object i in sequence A is inter-correlated with object k in sequence C.
Step two: and according to the target frame correlation processing result, dividing the frame correlation target and the non-frame correlation target in the current frame, and separating out the inter-frame correlation target and the non-inter-frame correlation target.
Step three: and for the inter-frame related targets, further filtering the pseudo-star targets by using the characteristic that the angular distance between the real star targets is stable and unchanged, and counting the number of the current frame each target and other targets of the current frame which meet the characteristic that the angular distance between the current frame and the other targets of the current frame is unchanged, namely the angular distances between the two targets in the current frame are the same as the angular distance between the previous frame or the previous two frames. And analyzing the statistical result, determining a frequency threshold value meeting the real star target, selecting the targets with the number larger than the threshold value as the real star target, and filtering the other targets serving as pseudo star targets.
After the frame correlation processing is performed on the target, the pseudo star target also has a frame correlation condition, and the motion track has continuity, so that the pseudo star target needs to be further filtered by utilizing the consistency of the motion track of the real star target and the relative stability of the inter-frame angular distance, and the reliability of the output star target data is ensured.
And dividing the frame related target and the non-frame related target according to the target frame related processing result, and separating a target with inter-frame correlation between the current frame and the previous two frames and a target without inter-frame matching correlation. And filtering pseudo star targets of the star targets related to the frames based on the stable and invariable characteristic of the angular distance of the real star targets, wherein the angular distance between the real star targets of the current frame is consistent with the angular distance between the related targets of the corresponding frames of the previous two frames. Suppose the angular distance between the two targets of target i and target j in sequence A is dAijThe angular distance between the target i and the target j in the sequence B is dBij. If the target i and the target j in the sequence A are both real star targets, then: | dAij-dBij|＜dT2Wherein i ≠ j; if one of the targets i and j in the sequence A is a pseudo star target, | dAij-dBijI very probable rate is greater than the angular distance threshold dT2. Fig. 2 shows a method for filtering pseudo star targets based on the characteristic that the angular distance of a real star target is stable and invariant.
As shown in FIG. 2, object 3 in sequence A is correlated with the previous frame, i.e., object in sequence B, and if there is no true match correlation with the previous two frames, i.e., objects in sequence C, there is | dA13-dB13|＜dT2. Similarly, if there is no frame matching correlation between target 4 in sequence A and the target in the previous frame, i.e. sequence B, and there is frame matching correlation between target 4 in the previous two frames, i.e. sequence C, then | dA exists14-dC14|＜dT2。
For each target in the sequence A, the angular distance between the target and all the targets in the sequence A is counted to satisfy | dAij-dBij|＜dT2The angular distance between the real star target and other targets satisfies | dAij-dBij|＜dT2The larger the number of times, the angular distance between the pseudo star target and other targets satisfies | dAij-dBij|＜dT2The smaller the number of times of (a), even 0. And analyzing the statistical result, determining a frequency threshold value meeting the real star target, selecting the target meeting the frequency threshold value as the real star target, and filtering the rest targets serving as pseudo star targets.
Step four: and separating the target which meets the inter-frame angular distance invariant characteristic from the target which does not meet the inter-frame angular distance invariant characteristic according to the processing result of filtering the pseudo-star target based on the inter-frame angular distance invariant characteristic of the real star target.
Step five: and calculating inter-frame angular distance values of frames before and after the target for the target meeting the inter-frame angular distance invariant characteristic, and arranging the inter-frame angular distance values in ascending order. And analyzing the angular distance information among the sequenced frame related targets, counting the number of the frame related targets in a certain angular distance interval range, wherein the angular distance interval range with the largest number of the frame related targets has the largest probability of containing the real star target, and filtering the pseudo star target with the large change of the inter-frame motion angular distance.
Step six: and analyzing the motion direction of the determined targets, eliminating the targets with larger motion direction difference, and regarding the rest targets as real star targets.
Step seven: sorting possibly real star targets according to the gray value according to the processing result, and outputting the sorted possibly real star targets as a first priority; sorting the targets with continuous tracks but not meeting the motion characteristics of the real star target according to the gray value, and outputting the targets as a second priority; and sequencing the targets with discontinuous tracks according to the gray value, outputting the targets as a third priority, and selecting a certain number of targets to participate in star map recognition according to actual needs.
And measuring the star target data in the actual flight test of the star sensor at a certain time, and testing and analyzing the actual flight target data of the star sensor by adopting an interframe correlation processing algorithm. As shown in fig. 3 and 4, the star sensor has maneuvering gesture adjusting operation in the process of measuring the star, the left side in the figure is a target track after filtering the pseudo star target by adopting a frame correlation processing method, and the right side is an original track of the target in the star measurement data. Simulation results show that:
(1) the real star target can be correctly and preferentially output when more than two times of continuous 3 frames of images of the real star target are detected, no less than 3 real star targets in the images are continuously tracked, the real star target can be correctly and preferentially output,
(2) the star sensor measures 280 frames of data of the star in total, the attitude adjustment exists in the middle of the bomb body, after the processing of the frame correlation algorithm, 1 pseudo star target is output in 7 frames of images, and no pseudo star target is output in the other images. Analyzing 7 frames of images with pseudo-star targets, and finding that the number of real star targets in the images is less than 3, and the pseudo-star targets are close to the angular distances of the front and rear frames of the real star targets, so that the pseudo-star targets are not correctly filtered and are output as the real star targets; when the number of real star targets in the image is large, the filtering purity of the pseudo star target is very high and can reach more than 98%.
(3) The more real star targets in the image, the higher the reliability of the star targets output through frame correlation processing.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (8)
1. A method for filtering a pseudo star target of a large-field-of-view star sensor is characterized by comprising the following steps:
setting targets detected by a current frame, a previous frame and a previous two frames as a sequence A, a sequence B and a sequence C respectively, calculating the angular distance between any two targets between the sequence A and the sequence B, between the sequence A and the sequence C and between the sequence B and the sequence C and the gray level deviation between the two targets respectively, and judging whether all the targets in the sequence A are related to the targets in the sequence B and the sequence C or not;
aiming at the target with inter-frame correlation, carrying out one-time filtering on a pseudo-star target based on the characteristic that the angular distance between frames of a real star target is stable and unchanged;
performing secondary filtering on a pseudo star target based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target for the target result subjected to the primary filtering, and outputting the residual target serving as a real star target;
the judging whether the inter-frame correlation exists between all the targets in the sequence A and the targets in the sequence B and the sequence C comprises the following steps:
determining a target inter-frame correlation condition according to the maximum angular velocity information of the star sensor movement, and judging and counting target pairs meeting the inter-frame correlation condition between the sequence A and the sequence B, between the sequence A and the sequence C, and between the sequence B and the sequence C according to the target inter-frame correlation condition;
for target i in sequence A, target j in sequence B, and target k in sequence C,
if i is related to j, i is related to k, and j is related to k, judging that i, j and k are related;
if i is related to j, i is related to k, and j is related to k', judging that i is related to j; k' is a different target from k in sequence B;
if i is related to j, i is related to k, and the sequence C does not have a target related to j, judging that i is related to k;
if the sequence B and the sequence C do not have the target related to the i, the i has no target inter-frame correlation;
if the target related to i does not exist in the sequence B and i is only related to the target k in the sequence C, i is considered to be related to k;
if the target related to i does not exist in the sequence B, i is related to other targets in the sequence C besides k, and the target with higher possibility of being related to the inter-frame is determined as the inter-frame related target of i by weighting the target inter-frame by the gray change rate and the motion angle distance.
2. The method of claim 1, wherein the calculating the angular distance between any two targets between sequence a and sequence B, sequence a and sequence C, and sequence B and sequence C, and the gray scale deviation between the two targets respectively comprises:
for the calculation method of the angular distance between any two targets between the sequence A and the sequence B and the gray scale deviation between the two targets,
dSeqTarget＝acos(xnow·xbefore+ynow·ybefore+znow·zbefore)
lumSeqDev＝abs(lumnow-lumbefore)
in the formula, xnow、ynow、znowVector, x, representing the sequence A object in the star sensor coordinate systembefore、ybefore、zbeforeRepresenting a vector of the sequence B target in a star sensor coordinate system, wherein the dSeqTarget is angular distance information of two frame targets; lumnowGray scale of sequence A object, lumbeforeFor the gray level of the sequence B object, lumSeqDev is the difference in gray level between the sequence A object and the sequence B object;
and similarly, calculating the angular distance between any two targets between the sequence A and the sequence C, between the sequence B and the sequence C and the gray level deviation between the two targets.
3. The method according to claim 1, wherein the calculation method for determining the correlation possibility between the target frames by weighting the gray change rate + the motion angle distance amount between the target frames is as follows:
lumSeqProb＝1-lumSeqDev/lumbefore
dSeqProb＝1-dSeqNow/dSeqMax
prob＝a×lumSeqProb+b×dSeqProb
wherein lumSeqDev is the target inter-frame gray scale variation, lumbeforeThe gray value of a previous frame of the target, lumSeqProb, dSeqNow, dSeqMax, and dSeqProb are the gray value of the previous frame of the target, the lumSeqProb is the correlation of the gray value of the target, the dSeqNow is the angular distance of the motion between the target frames, the dSeqMax is the maximum angular distance value of the motion between the target frames, and the dSeqProb is the angular distance between the target framesProb is the probability value related to the target frame, and a and b are the weighting coefficients of the gray scale change rate and the inter-frame motion angular distance.
4. The method according to claim 2, wherein the determining the target frame-to-frame correlation condition according to the maximum angular velocity information of the star sensor motion comprises:
in the formula (d)T1A first angular distance threshold representing a target inter-frame correlation process; lumTRepresenting the gray change rate threshold of the frames before and after the target.
5. The method of claim 1, wherein for the target with inter-frame correlation, performing one-time pseudo-star target filtering based on the characteristic that the angular distance between the actual star targets is stable and invariant comprises:
suppose the angular distance between the two targets of target i and target j in sequence A is dAijThe angular distance between the target i and the target j in the sequence B is dBijThe second angular distance threshold is dT2；
For the object i in the sequence A, the angular distance between the object i and all other objects in the sequence A is counted to satisfy | dAij-dBij|＜dT2When the number of times is greater than or equal to the number threshold, the target is judged to be a real star target, and other targets are filtered as pseudo star targets.
6. The method according to claim 1, wherein the performing pseudo star target secondary filtering on the once filtered target result based on the relative concentration characteristics of the motion angular distances of the inter-frame real star targets and outputting the remaining targets as the real star targets comprises:
calculating the amount of angular distance motion between target frames for the target result subjected to primary filtering, and arranging the target result in ascending order according to the angular distance between the frames;
counting the number of frame matching related targets in a preset angular distance interval range;
and analyzing the moving direction of the target contained in the angular distance interval range with the largest number of the frame matching related targets, eliminating the target with larger difference of the moving direction, and outputting the residual targets.
7. The method according to claim 6, wherein the second filtering of the pseudo star object is performed on the result of the object subjected to the first filtering based on the relative concentration characteristics of the angular distances of motion of the inter-frame real star object, and the remaining objects are output as real star objects, further comprising:
sorting the remaining targets according to the gray value size, and outputting the sorted remaining targets as a first priority; sorting the targets with continuous tracks but not meeting the motion characteristics of the real star target according to the gray value, and outputting the targets as a second priority; and sequencing the targets with discontinuous tracks according to the gray value size, and outputting the targets as a third priority.
8. A system for filtering a pseudo star target of a large-field-of-view star sensor is characterized by comprising:
the inter-frame correlation judging module is used for respectively calculating the angular distance between any two targets between the sequence A and the sequence B, between the sequence A and the sequence C, between the sequence B and the sequence C and the gray level deviation between the two targets, and judging whether all the targets in the sequence A are inter-frame correlated with the targets in the sequence B and the sequence C;
the primary filtering module is used for carrying out primary filtering on the pseudo star target based on the stable and invariable characteristic of the angular distance between the actual star target frames aiming at the target with the inter-frame correlation;
the secondary filtering and outputting module is used for carrying out secondary filtering on the pseudo-star target on the target result subjected to the primary filtering based on the relative concentration characteristic of the movement angular distance of the inter-frame real star target and outputting the residual target serving as the real star target;
the judging whether the inter-frame correlation exists between all the targets in the sequence A and the targets in the sequence B and the sequence C comprises the following steps:
determining a target inter-frame correlation condition according to the maximum angular velocity information of the star sensor movement, and judging and counting target pairs meeting the inter-frame correlation condition between the sequence A and the sequence B, between the sequence A and the sequence C, and between the sequence B and the sequence C according to the target inter-frame correlation condition;
for target i in sequence A, target j in sequence B, and target k in sequence C,
if i is related to j, i is related to k, and j is related to k, judging that i, j and k are related;
if i is related to j, i is related to k, and j is related to k', then i is judged to be related to j; k' is a different target from k in sequence B;
if i is related to j, i is related to k, and the sequence C does not have a target related to j, judging that i is related to k;
if the sequence B and the sequence C do not have the target related to the i, the i has no target inter-frame correlation;
if the target related to i does not exist in the sequence B and i is only related to the target k in the sequence C, i is considered to be related to k;
if the target related to i does not exist in the sequence B, i is related to other targets in the sequence C besides k, and the target with higher possibility of being related to the inter-frame is determined as the inter-frame related target of i by weighting the target inter-frame by the gray change rate and the motion angle distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011243133.3A CN112595312B (en) | 2020-11-09 | 2020-11-09 | Method and system for filtering pseudo star target of large-field-of-view star sensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011243133.3A CN112595312B (en) | 2020-11-09 | 2020-11-09 | Method and system for filtering pseudo star target of large-field-of-view star sensor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112595312A CN112595312A (en) | 2021-04-02 |
| CN112595312B true CN112595312B (en) | 2022-07-05 |
Family
ID=75183077
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011243133.3A Active CN112595312B (en) | 2020-11-09 | 2020-11-09 | Method and system for filtering pseudo star target of large-field-of-view star sensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112595312B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114279463B (en) * | 2021-12-14 | 2023-08-29 | 江苏集萃智能光电系统研究所有限公司 | Rolling shutter distortion correction method based on single-frame star map angular velocity estimation |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20050105369A (en) * | 2004-04-30 | 2005-11-04 | 한국과학기술원 | Apparatus for attitude determination test of star-sensor and method for display of imaginary constellation |
| CN102840860A (en) * | 2012-08-30 | 2012-12-26 | 北京航空航天大学 | Hybrid particle swarm algorithm-based star graph recognition method |
| CN104567865A (en) * | 2014-12-29 | 2015-04-29 | 北京控制工程研究所 | Attitude capture method of star sensor under space particle interference condition |
| CN109724591A (en) * | 2019-02-27 | 2019-05-07 | 北京航空航天大学 | It is a kind of based on angular distance screening star sensor artificial star screen out method |
| WO2020117530A1 (en) * | 2018-12-03 | 2020-06-11 | Ball Aerospace & Technologies Corp. | Star tracker for multiple-mode detection and tracking of dim targets |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103674023B (en) * | 2013-12-26 | 2014-10-15 | 中国人民解放军国防科学技术大学 | Method for dynamically measuring attitude of star sensor based on top accurate angle relevance |
-
2020
- 2020-11-09 CN CN202011243133.3A patent/CN112595312B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20050105369A (en) * | 2004-04-30 | 2005-11-04 | 한국과학기술원 | Apparatus for attitude determination test of star-sensor and method for display of imaginary constellation |
| CN102840860A (en) * | 2012-08-30 | 2012-12-26 | 北京航空航天大学 | Hybrid particle swarm algorithm-based star graph recognition method |
| CN104567865A (en) * | 2014-12-29 | 2015-04-29 | 北京控制工程研究所 | Attitude capture method of star sensor under space particle interference condition |
| WO2020117530A1 (en) * | 2018-12-03 | 2020-06-11 | Ball Aerospace & Technologies Corp. | Star tracker for multiple-mode detection and tracking of dim targets |
| CN109724591A (en) * | 2019-02-27 | 2019-05-07 | 北京航空航天大学 | It is a kind of based on angular distance screening star sensor artificial star screen out method |
Non-Patent Citations (3)
| Title |
|---|
| Dynamical tides excited in rotating stars of different masses and ages and the formation of close in orbits;S. V. Chernov 等;《Monthly Notices of the Royal Astronomical Society》;20131231;第434卷(第2期);第1079-1097页 * |
| 基于嵌入式处理器的实用快速全天星图识别算法;艾国等;《光学与光电技术》;20161231;第14卷(第06期);第45-49、62页 * |
| 星图识别算法的优化设计与实现;蔡迎波等;《光学与光电技术》;20090430;第7卷(第02期);第37-39、69页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112595312A (en) | 2021-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105335955B (en) | Method for checking object and object test equipment | |
| CN104183127B (en) | Traffic surveillance video detection method and device | |
| CN111627044B (en) | Target tracking attack and defense method based on deep network | |
| CN107038713A (en) | A kind of moving target method for catching for merging optical flow method and neutral net | |
| CN112883819A (en) | Multi-target tracking method, device, system and computer readable storage medium | |
| EP1759334A2 (en) | Method and apparatus for visual odometry | |
| CN108520203B (en) | Multi-target feature extraction method based on fusion of self-adaptive multi-peripheral frame and cross pooling feature | |
| CN110058222B (en) | Double-layer particle filter tracking-before-detection method based on sensor selection | |
| CN104899866A (en) | Intelligent infrared small target detection method | |
| CN110555868A (en) | method for detecting small moving target under complex ground background | |
| WO2018152214A1 (en) | Event-based feature tracking | |
| CN105374049B (en) | Multi-corner point tracking method and device based on sparse optical flow method | |
| CN110345918B (en) | Space debris detection method and system based on star retrieval | |
| Cornall et al. | Aircraft attitude estimation from horizon video | |
| CN113744315B (en) | Semi-direct vision odometer based on binocular vision | |
| CN111160212A (en) | Improved tracking learning detection system and method based on YOLOv3-Tiny | |
| CN108334885A (en) | A kind of video satellite image space object detection method | |
| CN104331907B (en) | A kind of method based on ORB feature detections measurement bearer rate | |
| CN110853074A (en) | Video target detection network system for enhancing target by utilizing optical flow | |
| CN109165603B (en) | Ship detection method and device | |
| CN112595312B (en) | Method and system for filtering pseudo star target of large-field-of-view star sensor | |
| CN103942786A (en) | Self-adaptation block mass target detecting method of unmanned aerial vehicle visible light and infrared images | |
| CN116883457B (en) | Light multi-target tracking method based on detection tracking joint network and mixed density network | |
| RU2618927C2 (en) | Method for detecting moving objects | |
| CN108680940A (en) | A kind of automatic driving vehicle assisted location method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |