阅读说明：本技术 差异化导航星与空间目标提取方法、装置和存储介质 (Method and device for extracting differentiated navigation satellite and space target and storage medium ) 是由 王立 张俊 袁利 武延鹏 郑然� 赵春晖 卢欣 高秀娟 于 2021-01-12 设计创作，主要内容包括：本发明公开一种差异化导航星与空间目标提取方法,包括：在每一局部天区块中确定1-2颗符合预定条件的星；将所述符合预定条件的星和全天星表进行比对,将在所述全天星表里且符合所述预定条件的星确定为导航星,获取并利用所述导航星的位置参数确定卫星的姿态信息；在每一局部天区块中搜索空间目标,将搜索到的每一空间目标加入到第一目标队列或第二目标队列；对所述第一目标队列和第二目标队列进行帧间关联生成每一局部天区块的局部轨迹；将在预定时间内生成的每一局部天区块的多个局部轨迹进行关联操作生成局部轨迹的天区关联。通过将相平面划分为多个局部天区块,基于每一局部天区块提取空间目标,可以提高目标提取的效率和质量。(The invention discloses a method for extracting differentiated navigation satellites and space targets, which comprises the following steps: determining 1-2 stars meeting a preset condition in each local day block; comparing the star meeting the preset condition with a full celestial star table, determining the star meeting the preset condition in the full celestial star table as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star; searching a space target in each local day block, and adding each searched space target into a first target queue or a second target queue; performing inter-frame association on the first target queue and the second target queue to generate a local track of each local day block; and performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate day zone association of the local tracks. By dividing the phase plane into a plurality of local day blocks, the efficiency and quality of target extraction can be improved by extracting the spatial target based on each local day block.)

1. A method for extracting a differential navigation satellite and a space target is characterized by comprising the following steps:

determining 1-2 stars meeting a preset condition in each local day block, wherein the local day blocks are obtained by dividing a focal plane of a camera according to a preset rule;

comparing the star meeting the preset condition with a full celestial star table, determining the star meeting the preset condition in the full celestial star table as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star;

searching a space target in each local day block according to the attitude information of the satellite, and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition;

performing inter-frame association on the first target queue and the second target queue to generate a local track of each local day block;

and performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate day zone association of the local tracks.

2. The method as claimed in claim 1, wherein the step of adding each searched space target to the first target queue or the second target queue corresponding to each local day block according to a predetermined determination condition comprises:

judging whether the space target belongs to a determined suspected target or not;

when the space target does not belong to the determined suspected target, adding the space target into the first target queue;

and when the space target belongs to the determined suspected target, adding the space target into the second target queue.

3. The method of claim 1, further comprising:

entering an all-day capture mode after the camera initialization is completed;

setting the integration time to enable the product sensitivity to reach 5 Mv;

the step of determining 1-2 stars meeting the predetermined condition in each local day block comprises the following steps:

and calculating a background threshold value of each local day block, and determining 1-2 stars meeting a preset condition in each local day block according to the background threshold value.

4. The method as claimed in claim 3, wherein the determining 1-2 satellites meeting a predetermined condition in each local day block according to the background threshold comprises:

when the pixel point value of the star in the local sky plot satisfies g_i≥μ_i+ησ_iDetermining the star as a star meeting a preset condition; wherein, g_iFor pixel point values, μ_iFor the convolution calculation result, η is the coefficient, σ_iIs the standard deviation of the measured data to be measured,kernel is a function of a Kernel that is,

5. the method for extracting the differential navigation star and the space target according to claim 1, wherein: performing inter-frame association on the first target queue and the second target queue comprises:

spatial target information X in the first target queue and the second target queue_j、X_iAs input, by formula P_link,ji＝A_pos(X_j|X_i)A_size(X_j|X_i) Calculating the similarity between the space targets, and setting the segment with the similarity value larger than a preset threshold value as a track segment, wherein P is_link,jiSimilarity of track segments, A_pos(X_j|X_i) Being similarity of position, A_size(X_j|X_i) Similarity of the target morphology;

finding track segments associated with a pre-stored track from all the generated track segments by using a scene self-adaptive method, and generating the local track;

wherein the content of the first and second substances,X_j、X_icoordinate positions, σ, of the target points j and i on the celestial sphere, respectively_ijFor the uncertainty of position, X ═ cos α cos δ, sin α cos δ, sin δ, α is the right ascension and δ is the declination,wherein Δ_j、Δ_iAre each X_j、X_iThe feature size of (2).

6. The method of claim 5, wherein associating a plurality of local tracks for each local day block generated within a predetermined time to generate a day block association of local tracks comprises:

arranging P in numerical order from high to low_link,ji；

For each space target of each local day block, when a space target associated with the space target is found from pre-stored space target information, correcting the position and the size of the local track by using Kalman filtering, and extracting the speed characteristic of the local track according to the position and the size of the local track;

when the space target related to the space target is not found from the pre-stored space target information, using a formula P_link,kl＝A_{pos_curve}(P_link,k|P_link,l)A_velo(P_link,k|P_link,l) Calculating the connection probability among the local tracks;

wherein, P_link,klTo total probability after linking, P_link,l，P_link,kRespectively the connection probability, A, of the local trajectories l, k_{pos_curve}Similarity of position curves, A_veloSimilarity of velocity curves;

let P_link,kHas a fitting curve of f (t) and an error ofP_link,lThe fitted curve of (a) is g (t) and the error isA_{pos_curve}(P_link,k|P_link,l) The expression is as follows:

let P_link,kThe fitting speed of (d) is v (t), P_link,lThe fitted curve of (a) is u (t), A_velo(P_link,k|P_link,l) The expression is as follows:

7. the utility model provides a device that realizes differentiation navigation star and space target extraction which characterized in that includes:

the satellite determining module is used for determining 1-2 satellites meeting the preset conditions in each local day block, and the local day blocks are obtained by dividing the focal plane of the camera according to preset rules;

the satellite attitude information determining module is used for comparing the star meeting the preset condition with the all-celestial star chart, determining the star in the all-celestial star chart and meeting the preset condition as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star;

the tracking queue determining module is used for searching a space target in each local day block according to the attitude information of the satellite and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition;

the inter-frame correlation module is used for performing inter-frame correlation on the first target queue and the second target queue to generate a local track of each local day block;

and the local track association module is used for performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate the day zone association of the local tracks.

8. The apparatus of claim 7, wherein the tracking queue determining module is further configured to determine whether the spatial target belongs to a certain suspected target; when the space target does not belong to the determined suspected target, adding the space target into the first target queue; and when the space target belongs to the determined suspected target, adding the space target into the second target queue.

9. The device of claim 7, further comprising a configuration module for entering a full-day capture mode after the initialization of the camera is completed, and setting an integration time so that the product sensitivity reaches 5 Mv;

the star determination module is further configured to calculate a background threshold value of each local day block, and determine 1-2 stars meeting a predetermined condition in each local day block according to the background threshold value.

10. According to claimThe apparatus of claim 9, wherein the star determination module is further configured to determine the star pixel point value in the local day block satisfies g_i≥μ_i+ησ_iDetermining the star as a star meeting a preset condition; wherein, g_iFor pixel point values, μ_iFor the convolution calculation result, η is the coefficient, σ_iIs the standard deviation of the measured data to be measured,kernel is a function of a Kernel that is,

11. the apparatus of claim 7, wherein: the inter-frame association module is further used for

Spatial target information X in the first target queue and the second target queue_j、X_iAs input, by formula P_link,ji＝A_pos(X_j|X_i)A_size(X_j|X_i) Calculating the similarity between the space targets, and setting the segment with the similarity value larger than a preset threshold value as a track segment, wherein P is_link,jiSimilarity of track segments, A_pos(X_j|X_i) Being similarity of position, A_size(X_j|X_i) Similarity of the target morphology;

finding track segments associated with a pre-stored track from all the generated track segments by using a scene self-adaptive method, and generating the local track;

wherein the content of the first and second substances,X_j、X_icoordinate positions, σ, of the target points j and i on the celestial sphere, respectively_ijFor the uncertainty of position, X ═ cos α cos δ, sin α cos δ, sin δ, α is the right ascension and δ is the declination,wherein Δ_j、Δ_iAre each X_j、X_iThe feature size of (2).

12. The apparatus of claim 11, wherein the local track association module is further configured to:

arranging P in numerical order from high to low_link,ji；

For each space target of each local day block, when a space target associated with the space target is found from pre-stored space target information, correcting the position and the size of the local track by using Kalman filtering, and extracting the speed characteristic of the local track according to the position and the size of the local track;

when the space target related to the space target is not found from the pre-stored space target information, using a formula P_link,kl＝A_{pos_curve}(P_link,k|P_link,l)A_velo(P_link,k|P_link,l) Calculating the connection probability among the local tracks;

wherein, P_link,klTo total probability after linking, P_link,l，P_link,kRespectively the connection probability, A, of the local trajectories l, k_{pos_curve}Similarity of position curves, A_veloSimilarity of velocity curves;

let P_link,kHas a fitting curve of f (t) and an error ofP_link,lThe fitted curve of (a) is g (t) and the error isA_{pos_curve}(P_link,k|P_link,l) The expression is as follows:

let P_link,kThe fitting speed of (d) is v (t), P_link,lThe fitted curve of (a) is u (t), A_velo(P_link,k|P_link,l) The expression is as follows:

13. a non-volatile storage medium for storing a software program which, when executed, is adapted to perform the method of any of claims 1 to 6.

Technical Field

The invention relates to the field of target perception measurement, in particular to a method, a device and a storage medium for realizing extraction of a differentiated navigation satellite and a spatial target.

Background

In the face of the urgent need of national security for the development of spatial target situation perception capability in the new period, it is urgent to establish a space-based spatial target monitoring system and establish a space-based spatial target monitoring backbone network in China. At present, the size of a detected target is smaller and smaller, and in order to meet the detection requirements of 10cm @ GEO and 2cm @ LEO, detection stars and the like are also darker and darker, the corresponding sensitivity is superior to 14 stars and the like, and the highest detection sensitivity reaches 19 stars and the like. Overall, the following objectives are to be achieved: (1) real-time performance, the processing process of a single frame image comprises image preprocessing, target identification and position calculation, and the processing time of the processing process is shorter than the exposure time interval of the telescope; (2) reliability, a processing result of a space target observation image, a lower false alarm rate and higher detection efficiency; (3) long-time association capability and data validity, namely capability of successfully associating complex track images.

Disclosure of Invention

The embodiment of the invention provides a method, a device and a storage medium for extracting a differentiated navigation satellite and a space target, so as to improve the extraction efficiency and quality of the navigation satellite and the space target.

The embodiment of the invention provides a method for extracting a differentiated navigation satellite and a space target, which comprises the following steps:

determining 1-2 stars meeting a preset condition in each local day block, wherein the local day blocks are obtained by dividing a focal plane of a camera according to a preset rule;

comparing the star meeting the preset condition with a full celestial star table, determining the star meeting the preset condition in the full celestial star table as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star;

searching a space target in each local day block according to the attitude information of the satellite, and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition;

performing inter-frame association on the first target queue and the second target queue to generate a local track of each local day block;

and performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate day zone association of the local tracks.

The embodiment of the invention provides a device for realizing extraction of differentiated navigation satellites and space targets, which comprises:

the satellite determining module is used for determining 1-2 satellites meeting the preset conditions in each local day block, and the local day blocks are obtained by dividing the focal plane of the camera according to preset rules;

the satellite attitude information determining module is used for comparing the star meeting the preset condition with the all-celestial star chart, determining the star in the all-celestial star chart and meeting the preset condition as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star;

the tracking queue determining module is used for searching a space target in each local day block according to the attitude information of the satellite and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition;

the inter-frame correlation module is used for performing inter-frame correlation on the first target queue and the second target queue to generate a local track of each local day block;

and the local track association module is used for performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate the day zone association of the local tracks.

Embodiments of the present invention provide a non-volatile storage medium for storing a software program, where the software program is used to execute the steps of the method described above when executed.

In the embodiment of the invention, 1-2 stars meeting the preset condition are determined in each local day block, and the local day block is obtained by dividing the focal plane of a camera according to the preset rule; comparing the star meeting the preset condition with the all-celestial star table, determining the star meeting the preset condition in the all-celestial star table as a navigation star, and acquiring and determining the attitude information of the satellite by using the position parameters of the navigation star; searching a space target in each local day block according to the attitude information of the satellite, and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition; performing inter-frame association on the first target queue and the second target queue to generate a local track of each local day block; and performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate day zone association of the local tracks. By dividing the focal plane of the camera into local day blocks and determining local tracks and local track day area associations for each day block, the embodiment of the invention has a larger field of view compared with the existing method for determining the local track of a space target for the whole focal plane. Because the field of view is bigger, the number of navigation stars and space targets processed at one time is more, and the efficiency is higher. In addition, because the field of view is larger, the information of the navigation star and the space target in the field of view is more accurate and better in quality.

Drawings

Fig. 1 is a schematic flowchart of a method for extracting a differentiated navigation satellite and a spatial target according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of another method for extracting a differentiated navigation satellite and a spatial target according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a focal plane formed by splicing eight local sky patches according to an embodiment of the present invention;

fig. 4 is a flowchart illustrating a method for determining an attitude of a satellite in an attitude determination mode according to an embodiment of the present invention;

fig. 5 is a schematic flowchart of a method for generating a sky-region association of a local track in a target detection mode according to an embodiment of the present invention;

fig. 6 is a schematic flowchart of a method for associating local tracks to generate day-zone associations of local tracks according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of a local day block in accordance with an embodiment of the present invention;

FIG. 8 is a schematic diagram of focal plane target recognition function assignment for an embodiment of the present invention;

fig. 9 is a device for implementing extraction of a differential navigation star and a spatial target according to an embodiment of the present invention.

Detailed Description

The invention is described in detail below with reference to the figures and specific examples.

Fig. 1 is a schematic flow chart of a method for extracting a differentiated navigation star and a spatial target according to an embodiment of the present invention. As shown in fig. 1, the method includes:

step 101: and determining 1-2 stars meeting a preset condition in each local day block, wherein the local day blocks are obtained by dividing the focal plane of the camera according to a preset rule.

Step 102: comparing the star meeting the preset condition with the all-celestial star chart, determining the star meeting the preset condition in the all-celestial star chart as a navigation star, and acquiring and utilizing the position parameters of the navigation star to determine the attitude information of the satellite.

Step 103: and searching a space target in each local day block according to the attitude information of the satellite, and adding each searched space target into a first target queue or a second target queue corresponding to each local day block according to a preset judgment condition.

Step 104: and performing inter-frame association on the first target queue and the second target queue to generate a local track of each local day block.

Step 105: and performing association operation on a plurality of local tracks of each local day block generated in a preset time to generate day zone association of the local tracks.

In the embodiment of the present invention, adding each searched space target to the first target queue or the second target queue corresponding to each local day block according to the preset determination condition includes: and judging whether the space target belongs to a determined suspected target, adding the space target to the first target queue when the space target does not belong to the determined suspected target, and adding the space target to the second target queue when the space target belongs to the determined suspected target.

The embodiment of the invention provides another flow schematic diagram of a method for extracting a differentiated navigation satellite and a space target. As shown in fig. 2, the specific steps include:

step 201, initializing the camera. And after the initialization of the camera is completed, determining that the camera is in a normal working state.

In step 202, the focal plane of the camera is partitioned according to a predetermined rule, for example, the focal plane may be partitioned into a plurality of small blocks according to the size of the field of view, each small block being called a local day block. For example, the divided local day blocks have a fixed field size (e.g., 1 ° × 1 °) or a proportion (e.g., 10% × 10%), as shown in fig. 3, and fig. 3 is a schematic diagram of a focal plane composed of eight local day blocks in a mosaic provided by the embodiment of the present invention. In subsequent operations, each local day block is processed in a pose determination mode and a target detection mode, respectively. The attitude determination mode mainly completes the functions of bright star extraction (namely, star meeting the preset conditions), star map recognition, tracking and satellite attitude determination. 1-2 bright stars are extracted from each local day block to participate in the subsequent all-day attitude identification and determination, the attitude of the satellite is calculated after the star map identification is completed, and after the attitude of the satellite enters a tracking mode, the processing of the technical scheme of the invention is switched into a target detection mode. In the embodiment of the present invention, other numbers of bright stars, for example, more than 2 bright stars, may also be extracted from each local day block according to actual needs, and the number of extracted bright stars is not limited in the present invention. In the target detection mode, the functions of extracting, capturing, identifying, authenticating, calculating parameters, correlating between frames, correlating in local day zones and the like of the space target in each local day zone are mainly completed. In specific operation, each local day block is divided into a capture area and a tracking area. And extracting internal bright spots in a capture area in real time, and completing the establishment of a suspected queue and the capture of a moving space target. Extracting window information of the space target in the tracking area, updating target queue information, generating a local track, and finishing local day area association after accumulating a certain local track.

And step 203, setting the integration time in the all-day capture mode so that the product sensitivity reaches 5 Mv.

Step 204, determining the satellite attitude. In the step, a background threshold value of a local sky block is calculated, and a hardware pipeline technology is adopted to extract highlight pixel points. On the basis of obtaining highlight pixel points (namely stars meeting preset conditions), 1-2 stars meeting the preset conditions, namely bright stars, are determined in each local day block, the bright stars are compared with a whole-day star table, star map recognition is completed, the bright stars in the whole-day star table are found out and determined as navigation stars, and attitude information of the satellite is calculated according to position parameters of the navigation stars.

Step 205, determine whether the calculation of the attitude information of the satellite is successful, and the number of times reaches 2. If so, the operation in the object detection mode is performed in step 206, otherwise, the process returns to step 203.

Fig. 4 is a flowchart illustrating a method for determining an attitude of a satellite in an attitude determination mode according to an embodiment of the present invention. As shown in fig. 4, the method includes:

step 401, initializing a camera.

At step 402, it is determined whether the camera is in an all-day recognition mode. If so, step 403 is performed, otherwise, step 409 is performed, ending the pose determination mode.

And step 403, setting the integration time so that the product sensitivity reaches 5 Mv.

In step 404, a detection unit is selected.

In step 405, hardware pixel extraction is performed. In this step, a hardware pipeline technique is used to extract the highlight pixel points. The method for extracting the internal highlight pixel points from the capture area in real time comprises the following steps: estimating the background mean value and standard deviation, and when the pixel point value of the sampled star point meets g_i≥μ_i+ησ_iWhen is at timeThe star point is extracted as a highlight pixel point. Wherein, g_iFor pixel point values, μ_iFor the convolution calculation result, the formula is:kernel is Kernel function, η is coefficient, σ_iFor standard deviation, the calculation formula is:

step 406, extracting the star point information of the bright star in each local day block.

Step 407, performing a star map identification operation by using the star point information of the bright star and the all-day star table, and determining the navigation star in the current local day block according to the star map identification result.

And step 408, determining the attitude information of the satellite according to the position information of the navigation satellite.

Step 409, the pose determination mode is ended.

Fig. 5 is a flowchart illustrating a method for generating a sky-region association of a local track in a target detection mode according to an embodiment of the present invention. As shown in fig. 5, the method includes the steps of:

step 501, confirming whether the camera is in the target detection mode, if so, executing step 502, otherwise, executing step 508, and entering a posture determination mode.

And 502, predicting the position of the current frame navigation star according to the attitude information of the satellite acquired in the attitude determination mode, realizing the functions of star extraction and tracking matching, and determining the attitude of the satellite after matching is successful.

In step 503, for the tracking target queue (referred to as the first target queue) already established, the current frame target position is predicted according to the information of the tracking target queue, and a spatial target is searched in each local day block.

In step 504, each time a space object is searched, it is determined that the space object does not belong to the determined suspected object, and the space object is updated to the first object queue.

And 505, for the condition that a tracking target queue is not established, firstly establishing a capture area according to the attitude information of the satellite, entering a capture authentication mode, extracting highlight pixel points, namely the highlight satellite parameters, by using a hardware pipeline technology, and completing the extraction and capture authentication of the space target. And when the space target is successfully authenticated, determining the parameters of the space target, judging that the space target belongs to a determined suspected target, establishing a tracking target queue (called as a second target queue), and adding the space target into the newly-established second target queue.

Step 506, inter-frame association is performed on the first target queue and the second target queue, and a local track of each local day block is generated.

In step 507, after accumulation of a predetermined time, after accumulating a plurality of local tracks (called segment local tracks), the local tracks are associated to generate day-area association of the local tracks.

Fig. 6 is a flowchart illustrating a method for associating local tracks to generate day and area associations of local tracks according to an embodiment of the present invention. As shown in fig. 6, the method includes the steps of:

step 601, performing inter-frame association operation, specifically including:

step 6011, target point information X_j、X_iAs an input.

Step 6012, calculating similarity of the positions. In this step, the formula is usedAnd calculating the similarity of the positions. Wherein, X_j、X_iCoordinate positions, σ, of the target points j and i on the celestial sphere, respectively_ijFor the uncertainty of position, X ═ cos α cos δ, sin α cos δ, sin δ, α is right ascension and δ is declination.

Step 6013, the similarity of the sizes is calculated. In this step, use is made ofCalculating the similarity of the sizes, wherein_j、Δ_iAre each X_j、X_iThe feature size of (2).

Step 6014, by formula P_link,ji＝A_pos(X_j|X_i)A_size(X_j|X_i) And calculating the similarity between the space targets, and setting the segments with the similarity values larger than a preset threshold value as track segments, namely reliable track segments. Wherein, P_link,jiSimilarity of track segments, A_pos(X_j|X_i) Being similarity of position, A_size(X_j|X_i) The similarity of the target features. In this step, it may be further determined whether the similarity of the track segments reaches the maximum, if so, step 602 is executed, otherwise, step 6012 is executed.

Step 602, finding out track segments associated with a pre-stored track (also called an existing track) from all the generated track segments by using a scene adaptive method, generating the local track,

step 603, the plurality of local tracks of each local day block generated in the preset time are correlated to generate the day zone correlation of the local track.

The method specifically comprises the following steps:

step 6031, arrange P by number from high to low_link,ji. Will P_link,jiThe method is used as an important parameter of a local track (segment) to participate in local multi-track calculation, namely track segment association, and the two conditions of the associated local track and the unassociated local track are respectively processed.

Step 6032, for each space target of each local day block, when the space target associated with the space target is found from the pre-stored space target information, performing feature information integration of the track segment of the associated target (i.e., the track segment where the associated target is located).

For example, for the associated target, the position and the size of the track segment are corrected by using kalman filtering, and thus the speed characteristic of the track segment is extracted.

Step 6033, for each space target of each local day block, when the space target associated with the space target is not found from the pre-stored space target information, performing track segment association calculation of the unassociated target.

In this step, the link probability calculation formula between track segments is as follows:

P_link,kl＝A_{pos_curve}(P_link,k|P_link,l)A_velo(P_link,k|P_link,l) Wherein P is_link,klTo total probability after linking, P_link,l，P_link,kRespectively, are track segments l, k, A_{pos_curve}Similarity of position curves, A_veloSimilarity of velocity curves.

Let P_link,kThe fitted curve of the segment is f (t) with an error ofP_link,lThe fitted curve of the segment is g (t) with an error ofA_{pos_curve}(P_link,k|P_link,l) Can be expressed as:

let P_link,kThe fitting speed of the segment is v (t), P_link,lThe fitted curve of the segment is u (t), A_velo(P_link,k|P_link,l) Can be expressed as:

the technical scheme of the invention is mainly divided into the following aspects:

(1) and (4) function partitioning. The focal plane is divided into a predetermined number of small blocks called local day blocks according to the size of the field of view, as shown in fig. 7, fig. 7 is a schematic diagram of a local day block according to an embodiment of the present invention. And carrying out different processing on the local day block according to the attitude determination and the target measurement process. For attitude determination, each day block is treated equally, and the brightest 2 stars in each day block are calculated to participate in the subsequent star map identification process.

For target measurement, the target measurement is processed respectively according to the size of the running speed, and is divided into a capture area and a tracking area according to whether the target measurement enters a visual field or not. Fig. 8 is a schematic diagram of allocation of a focal plane target recognition function according to an embodiment of the present invention, as shown in fig. 8, a: 2 degrees (45 degrees direction), B: not more than 4 degree (10 degree direction)

The part with the arrow is a capture area, extraction of stars and targets is realized by using a hardware pipeline technology, and the other part is a tracking area, namely prediction is carried out according to known target measurement information, and the next frame of target is directly extracted from the image.

(2) Attitude determination procedure

The attitude estimation mainly adopts a bright star mechanism, and fixed stars larger than 9Mv of an image plane are extracted and directly participate in calculation. The characteristic that the angular rate of the camera changes slowly is mainly utilized to distinguish the all-day recognition capture mode from the tracking mode.

When the image is in the all-day capturing mode, a short integration time (such as 5-10ms) is adopted, an image plane covering a 15-degree multiplied by 10-degree view field of the detector is divided into 15-degree multiplied by 10-degree multiplied by 150 sections, only star points within 1-degree multiplied by 1-degree are searched on hardware, the brightest 2 stars are saved, and therefore, the maximum of no more than 50 bright stars are saved, and posture recognition is carried out. And (5) outputting in a stable posture, and entering a tracking mode.

In the tracking mode, the conventional scheme is adopted. The difference is that the integration time is recovered to 100ms, and the target detection function is realized. After processing is complete, an accurate local sky star map and predicted coordinates may be provided. The fixed installation relation of the detectors can be calibrated on the ground, and parameters of each detector image surface relative to other detector image surfaces are obtained.

(3) Target detection procedure

The above 150 intervals are still divided according to the characteristics of target detection, except that the target detection area is divided into an acquisition area and a tracking area.

The capture area mainly completes initial capture and authentication of the target, extracts internal bright spots in real time by adopting a hardware technology, and completes establishment of a suspected queue and capture of a moving target. The capture area is determined by determining the module requiring hardware processing based on the motion speed. As shown in fig. 7, the focal plane is divided into 150 blocks, each corresponding to approximately 1 ° × 1 ° space. When the moving speed in the image frame time is 1 degree (45 degree direction), the arrow block shown in the figure is used as a capture area, and other areas are used as tracking areas, and the like.

The tracking area will perform the tracking function of the moving object, which can greatly reduce the extraction time. The position of the target on the image plane can be estimated in real time according to the moving target information, and then a target window is directly extracted from the image to update the known target queue information.

(4) Inter-frame association process

And taking the obtained measured values (double-star or multi-star precise tracking data) as data points of the temporary flight path, recording the azimuth angle, the energy information and the target size of each observation point, and establishing a temporary flight path set.

The maximum speed and the minimum speed of the target in the respective directions are respectively taken as the radius of the product of the sampling period, and the associated circular ring-shaped area is formed, so that the purpose of judgment is to fully utilize the prior information, for example, in the initial stage of the target, the speed of the target in a certain direction is known to be higher, the speed in a certain direction is small, and the direction speed judgment is utilized, so that the formation of false tracks is reduced in the formation process of temporary tracks.

When there are no subsequent correlation points, a point is extrapolated using a one-step prediction. And if the observation which fails to meet the condition falls into the forecast wave gate in the next correlation process, considering the temporary track as a false track, and logging off the track.

For a temporary track (the number of observation points is more than 5) of which the track cannot be confirmed, calculating the speed of the space target at the azimuth angle by using least square fitting, checking whether the measurement data fall within an allowable deviation angle range, if the measurement point appears in the flight direction prediction area of the temporary track target, considering the point to be effective, and otherwise, considering that the measurement point is not a subsequent point of the temporary track.

If a plurality of observations satisfy the above correlation, determining the confidence level of the observation according to the range of the position deviation D (the difference value of the position of the observation point and the position of the predicted point) and the characteristic deviation (a characteristic correlation method) of the target, and taking the observation point with the maximum confidence level as the optimal subsequent point of the temporary flight path.

The above steps are repeated until the temporary track meets the output condition (track confirmation principle of variable window length).

The point information X is subjected to interframe preliminary association algorithm_iAs input, reliable track segments are generated by directly computing similarities: p_link,ji＝A_pos(X_j|X_i)A_size(X_j|X_i)

The location similarity is calculated as:

X＝(cosαcosδ,sinαcosδ,sinδ)

and calculating the similarity of all track segments, and associating the detection value of the current frame in the tracking process to the existing track based on a scene self-adaptive mode to generate a local track segment.

(5) Local track association process

Arranging the local track segments from high to low according to the maximum similarity, taking the local track segments as important parameters of the track segments to participate in the next round of calculation, namely, the track segment association process, and calculating the local queue which is associated and the local queue which is not associated.

a) Track segment feature information integration

And for the correlated target, correcting the position and the size of the track segment by using Kalman filtering, and extracting the speed characteristic of the track segment according to the position and the size.

b) Track segment correlation computation

The calculation formula of the link probability between the track segments is as follows:

P_link,kl＝A_{pos_curve}(P_link,k|P_link,l)A_velo(P_link,k|P_link,l)

let P_link,kThe fitted curve of the segment is f (t) with an error ofP_link,lThe fitted curve of the segment is g (t) with an error ofA_{pos_curve}(P_link,k|P_link,l) Can be expressed as:

let P_link,kThe fitting speed of the segment is v (t), P_link,lThe fitted curve of the segment is u (t), A_velo(P_link,k|P_link,l) Can be expressed as

The similarity of the track segments can be obtained, and the maximum similarity is used as the matching success basis of the segments.

According to the invention, the focal planes of different detectors are re-divided and functionally defined, so that the fast extraction of fixed stars and space targets under the background of a complex star field is realized, the real-time property of attitude determination is ensured, and the extraction capability of dim and weak targets is realized. According to the invention, the space area track is divided, the inter-frame association track is maximized to be a target in terms of similarity, and long-time track association is realized, so that the method is beneficial to discrimination, orbit determination and dynamic tracking of a space target state. The invention breaks through the bottleneck of the prior art, effectively utilizes the prior resources to the maximum extent, and realizes the real-time capturing, tracking and correlating capabilities of space-based targets under the condition of limited resources.

Fig. 9 is a device for extracting a differentiated navigation satellite and a spatial target according to an embodiment of the present invention, including:

a star determination module 901, configured to determine 1 to 2 stars meeting a predetermined condition in each local day block, where the local day block is obtained by dividing a focal plane of a camera according to a predetermined rule;

a satellite attitude information determining module 902, configured to compare the satellites meeting the predetermined condition with a full celestial star chart, determine the satellites in the full celestial star chart and meeting the predetermined condition as navigation satellites, and obtain and determine attitude information of the satellites by using position parameters of the navigation satellites;

a tracking queue determining module 903, configured to search a space target in each local day block according to the attitude information of the satellite, and add each searched space target to a first target queue or a second target queue corresponding to each local day block according to a preset determination condition;

an inter-frame association module 904, configured to perform inter-frame association on the first target queue and the second target queue to generate a local track of each local day block;

a local track association module 905, configured to perform an association operation on the multiple local tracks of each local day block generated within a predetermined time to generate a day zone association of the local tracks.

The tracking queue determining module 903 is further configured to determine whether the spatial target belongs to a determined suspected target; when the space target does not belong to the determined suspected target, adding the space target into the first target queue; and when the space target belongs to the determined suspected target, adding the space target into the second target queue.

The device further comprises a configuration module, a display module and a display module, wherein the configuration module is used for entering an all-day capture mode after the initialization of the camera is completed; setting the integration time to enable the product sensitivity to reach 5 Mv;

the star determination module 901 is further configured to calculate a background threshold value of each local day block, and determine 1 to 2 stars meeting a predetermined condition in each local day block according to the background threshold value.

The star determination modelA block 901 for further determining if pixel point values of stars in the local day block satisfy g_i≥μ_i+ησ_iDetermining the star as a star meeting a preset condition; wherein, g_iFor pixel point values, μ_iFor the convolution calculation result, η is the coefficient, σ_iIs the standard deviation of the measured data to be measured,kernel is a function of a Kernel that is,

the inter-frame association module 904 is further configured to

Spatial target information X in the first target queue and the second target queue_j、X_iAs input, by formula P_link,ji＝A_pos(X_j|X_i)A_size(X_j|X_i) Calculating the similarity between the space targets, and setting the segment with the similarity value larger than a preset threshold value as a track segment, wherein P is_link,jiSimilarity of track segments, A_pos(X_j|X_i) Being similarity of position, A_size(X_j|X_i) Similarity of the target morphology;

finding track segments associated with a pre-stored track from all the generated track segments by using a scene self-adaptive method, and generating the local track;

wherein the content of the first and second substances,X_j、X_icoordinate positions, σ, of the target points j and i on the celestial sphere, respectively_ijFor the uncertainty of position, X ═ cos α cos δ, sin α cos δ, sin δ, α is the right ascension and δ is the declination,wherein Δ_j、Δ_iAre each X_j、X_iThe feature size of (2).

The local track association module 905 is further configured to:

arranging P in numerical order from high to low_link,ji；

For each space target of each local day block, when a space target associated with the space target is found from pre-stored space target information, correcting the position and the size of the local track by using Kalman filtering, and extracting the speed characteristic of the local track according to the position and the size of the local track;

when the space target related to the space target is not found from the pre-stored space target information, using a formula P_link,kl＝A_{pos_curve}(P_link,k|P_link,l)A_velo(P_link,k|P_link,l) Calculating the connection probability among the local tracks;

wherein, P_link,klTo total probability after linking, P_link,l，P_link,kRespectively the connection probability, A, of the local trajectories l, k_{pos_curve}Similarity of position curves, A_veloSimilarity of velocity curves;

let P_link,kHas a fitting curve of f (t) and an error ofP_link,lThe fitted curve of (a) is g (t) and the error isA_{pos_curve}(P_link,k|P_link,l) The expression is as follows:

let P_link,kThe fitting speed of (d) is v (t), P_link,lThe fitted curve of (a) is u (t), A_velo(P_link,k|P_link,l) The expression is as follows:

an embodiment of the present invention provides a non-volatile storage medium for storing a software program, where the software program is used to execute the methods described in fig. 1, fig. 2, and fig. 4 to fig. 6.

The invention is not described in detail and is within the knowledge of a person skilled in the art.