A Multi-Band Limb Detection Method for Dim Space Targets

Abstract

The present invention belongs to the technical field of space target detection, and discloses a multi-band limb detection method for dim space targets, including S101, using a multi-band infrared detection system for space target detection to image the limb or deep space area; S102, initializing the system integration time according to the angles between the detection system and the sun and the earth; S103, obtaining the image data collected by the multi-band infrared detection system; S104, performing target detection on the multi-band images; S105, performing registration transformation on the multi-band detection results; S106, associating and matching the targets after multi-band registration, and outputting the calculation result of the target feature information; S107, optimizing the integration time of the detection system according to the output target information and continuously observing the target. This method solves the problems that the existing single-band detection system has limited target information acquisition, weak anti-interference ability, and weak detection ability for dim targets.

Description

Technical Field

[0001] The present invention relates to the technical field of space target detection, and specifically to a multi-band limb detection method for dim space targets. Background Technique

[0002] In the observation scenario of space targets, the detection distance of the target is far, and the radiation energy of the target that the detector can receive is weak. At the same time, due to interference factors such as background fluctuations and cloud occlusion, the signal intensity received by the detector changes greatly. Therefore, the observation of space-based dim space targets has extremely high requirements for the sensitivity of the detection system.

[0003] In recent years, infrared long-wave detectors with low dark current levels in low-temperature environments have played a crucial role in detecting and tracking low-temperature and long-distance moving targets due to their advantages such as high sensitivity, high detectivity, and strong anti-interference ability, and are one of the ideal choices for space-based detection systems. However, the existing single-band detection system has limited target information acquisition, weak anti-interference ability, and insufficient detection ability for dim targets, which hinders the development of the field of infrared dim target search and tracking. Summary of the Invention

[0004] The present invention aims to provide a multi-band limb detection method for dim space targets, which integrates multi-band target energy information, improves the detection ability and sensitivity of the detection system for targets, and solves the problem of target detection at long operating distances.

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

[0006] A multi-band limb detection method for dim space targets, comprising the following steps:

[0007] S101. Use a multi-band infrared detection system for space target detection to image the limb or deep space area, where the limb area refers to the area at an altitude of 50 km - 500 km from the earth, and the deep space area refers to the area at an altitude of more than 500 km from the earth;

[0008] Among them, the multi-band infrared detection system has the following characteristics: the optical system is cryogenically cooled to below 120 K, the detection band is subdivided in the 5 - 16 micron band according to target characteristics, a planar array multi-band infrared detection system is constructed using no less than two bands, and the detection system shares the main optical system for imaging; the multi-band infrared detection system is a planar array image detector; the output frame rate of the detector is not less than 10 frames per second; the number of infrared channels is not less than two;

[0009] S102. Obtain the angle between the optical axis of the multi-band infrared detection system and the earth, and the angle between the optical axis of the multi-band infrared detection system and the sun, and set the initial imaging integration time for each band according to the above two angles;

[0010] S103. Use the multi-band infrared detection system initialized with the integration time in step S102 to observe and image the limb scene, and obtain multi-band infrared image data;

[0011] S104. Perform target detection on the image data obtained in step S103 to detect candidate targets in each band image;

[0012] S105. Match the candidate targets in each band to select valid targets. The inter-band target matching matrix is obtained by observing and registering stars with the multi-band infrared detection system. A valid target is a target that is observed in at least two bands among multiple bands;

[0013] S106. Calculate the neighborhood signal-to-noise ratio and background mean of the valid targets in each band, select the two bands with the highest signal-to-noise ratio for feature fusion of the valid targets, and package and output them together with the position information and angle information of the valid targets;

[0014] S107. Optimize the observation integration time using the average neighborhood signal-to-noise ratio and background mean of the observed targets in each band and continuously detect the targets.

[0015] Further, in S102, setting the initial imaging integration time for each band according to the above two angles includes establishing a mapping table by pre-recording the gray values of different integration times for each band under different angle combinations, and selecting the nearest neighbor integration time in the mapping table according to the current angle for setting.

[0016] Further, in S104, performing target detection on the image data obtained in step S103 includes extracting significant point targets in the image by local contrast or morphological filtering methods.

[0017] Further, in S105, matching the candidate targets in each band to select valid targets includes registering the candidate targets in each band to the image coordinate system of band 1, calculating the Euclidean distance between targets in different bands, and determining the targets with a Euclidean distance less than the set threshold as valid targets. The Euclidean distance M between targets in different bands is calculated according to the following formula:

[0018]

[0019] where (x1, y1) are the coordinate data of the candidate target in band 1, and (x2, y2) are the coordinate data of the candidate target in band 2 registered to the image coordinate system of band 1.

[0020] Further, in S106, the neighborhood signal-to-noise ratio SNR of candidate targets in each band L The calculation formula is as follows:

[0021]

[0022] Among them, M T is the maximum value of the pixels in the candidate target center area, and M B is the average value of the background in the target neighborhood, and S B is the standard deviation of the background in the target neighborhood. The calculation formula of M T is as follows:

[0023]

[0024] Among them, represents the central block S0;

[0025] The calculation formula of the background average value M B is as follows:

[0026]

[0027] Among them, R s represents the neighborhood background block S, and s represents the radius of S;

[0028] Among them, the calculation formula of S B is as follows:

[0029]

[0030] The regions of S and S0 are represented as follows:

[0031]

[0032] R S = {(p, q)|max(|p - x|, |q - y| ≤ s)}, s = 4, 7, 10, 13.

[0033] Furthermore, in S106, the two bands with the highest signal-to-noise ratio are selected for effective target feature fusion, including separately extracting the radiation intensity and energy concentration of the target, averaging the energy concentration and image plane position features of the two bands, and using the intensities of the two bands for fusion calculation of temperature and equivalent radiation area, and outputting multi-band target fusion features.

[0034] Furthermore, in S107, the average neighborhood signal-to-noise ratio and background average value of observing the target in each band are used to optimize the observation integration time, including calculating the average background average value of multiple targets. If the ratio of the background gray average value to the image saturation gray value is less than the set threshold, the integration time is increased in the smallest unit dT, otherwise it is decreased in the smallest unit.

[0035] The beneficial effects of the technical solution are:

[0036] A multi-band limb detection method for dim space targets of the present invention initializes the integration time of the detection system according to the angle between the optical axis of the multi-band infrared detection system and the Earth and the Sun, and obtains the grayscale image output by the multi-band image sensor; performs target detection and extraction on the image to obtain multi-band candidate target information; performs registration transformation on the multi-band detection results, and performs matching and association on the registered and transformed targets, fuses the multi-band target information to calculate and output the target feature information, and finally optimizes and adjusts the integration time of the detection system according to the target area and the neighborhood grayscale feature information to achieve the optimal observation of dim targets. In summary, this method uses a multi-band infrared detection system to obtain multi-dimensional target information, fuses the observation results of the detection system to enhance the target energy, improves the detection ability for dim targets, and solves the problem of target detection at long working distances. The integration time of the detection system is adaptively adjusted according to the target detection results to optimize the detection ability of the system to the greatest extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flowchart of a multi-band limb detection method for dim space targets of the present invention;

[0038] Figure 2 is a schematic diagram of the target detection results of the multi-band infrared detection system of a multi-band limb detection method for dim space targets of the present invention;

[0039] Figure 3 is a registration schematic diagram of the multi-band target detection results of a multi-band limb detection method for dim space targets of the present invention;

[0040] Figure 4 is a schematic diagram of the multi-band target detection results after the integration time is optimized for a multi-band limb detection method for dim space targets of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0041] The present invention will be further described in detail below with reference to the drawings and embodiments:

[0042] As Figure 1 shown, a multi-band limb detection method for dim space targets includes the following steps:

[0043] S101. Use a multi - band infrared detection system for space - target detection to image the limb or deep - space region. Among them, the limb region refers to the region at an altitude of 50 km - 500 km from the Earth, and the deep - space region refers to the region at an altitude of more than 500 km from the Earth. The multi - band infrared detection system has the following characteristics: The optical system is cryogenically cooled to below 120 K, and a two - long - wave and medium - long - wave band is used to construct a planar - array multi - band infrared detection system. The detection system shares the main optical system for imaging. The multi - band infrared detection system is a planar - array image detector. The output frame rate of the detector is 100 frames per second. The detection band is subdivided within the range of 5 - 16 microns.

[0044] S102. Obtain the angle between the optical axis of the multi - band infrared detection system and the Earth, and the angle between the optical axis of the multi - band infrared detection system and the Sun, and set the initial imaging integration time for each band according to the above two angles. Among them, setting the initial imaging integration time for each band according to the angle includes establishing a mapping table by pre - recording the gray - scale values of different integration times for each band under different angle combinations in advance, and selecting the nearest - neighbor integration time in the mapping table according to the current angle for setting, with the initial setting being 3 ms. The mapping table is shown in Table 1:

[0045] Table 1 Mapping table of solar angle, earth angle and integration time

[0046] Serial number Solar angle Earth angle Integration time 1 60° 60° 3 ms 2 90° 90° 5 ms 3 120° 120° 10 ms 4 150° 150° 20 ms

[0047] S103. Use the multi - band infrared detection system initialized with the integration time in step S102 to observe and image the limb scene, and obtain multi - band infrared image data.

[0048] S104. Perform target detection on the image data obtained in step S103 to detect candidate targets in each - band image. Performing target detection on the image data includes using local contrast or morphological filtering methods to extract significant point targets in the image.

[0049] S105. Match the candidate targets in each band to select effective targets. The inter - band target matching matrix is obtained through the registration calculation of observing stars by the multi - band infrared detection system. An effective target is a target that is observed by at least two bands among multiple bands.

[0050] Among them, matching the candidate targets in each band to select effective targets includes registering the candidate targets in each band to the image coordinate system of band 1, calculating the Euclidean distance between targets in different bands, and determining a target with a Euclidean distance less than five pixels as an effective target. The Euclidean distance M between targets in different bands is calculated according to the following formula:

[0051]

[0052] Among them, (x1, y1) are the coordinate data of the candidate target in Band 1, and (x2, y2) are the coordinate data of the candidate target registered in Band 2 in the image coordinate system of Band 1.

[0053] S106. Calculate the neighborhood signal-to-noise ratio and background mean of the effective target in each band, select the two bands with the highest signal-to-noise ratio for effective target feature fusion, and package and output them together with the position information and angle information of the effective target; among them, selecting the two bands with the highest signal-to-noise ratio of the effective target for target feature fusion includes separately extracting the radiation intensity and energy concentration of the target, averaging the energy concentration and image plane position features of the two bands, and using the intensities of the two bands for fusion calculation of temperature and equivalent radiation area, and outputting multi-band target fusion features;

[0054] The neighborhood signal-to-noise ratio SNR of the candidate target in each band L The calculation formula is as follows:

[0055]

[0056] Among them, M T is the maximum value of the pixels in the central area of the candidate target, M B is the background mean of the target neighborhood, S B is the background standard deviation of the target neighborhood, M T The calculation formula is as follows:

[0057]

[0058] Among them, represents the central block S0;

[0059] The background mean M B The calculation formula is as follows:

[0060]

[0061] Among them, R s represents the neighborhood background block S, and s represents the radius of S.

[0062] Among them, S B The calculation formula is as follows:

[0063]

[0064] The area representations of S and S0 are as follows:

[0065]

[0066] R S ={(p, q)|max(|p - x|, |q - y|≤s)}, s = 4, 7, 10, 13.

[0067] The detection results obtained by the method of the present invention are as Figure 2 shown, Figure 2 (a) is the target detection result of band 1. The signal-to-noise ratio of target 1 is 5.6, the signal-to-noise ratio of target 2 is 5.11, and the average background mean is 5010. Figure 2 (b) is the target detection result of band 2. The signal-to-noise ratio of target 1 is 4.86, and the average background mean is 4570. According to the calculation results, the integration time is adjusted to 5 ms.

[0068] The schematic diagram of multi-band target detection result registration provided by the embodiment of the present invention is as Figure 3 shown, Figure 3 (a) is the target detection result of band 1. The coordinates of target 1 are (431, 204), and the coordinates of target 2 are (167, 454). Figure 3 (b) is the schematic diagram of the position of the target registration in band 2 transformed into the image coordinate system of band 1. The coordinates of target 1 are (434, 206). The Euclidean distance between target 1 in band 1 and target 1 in band 2 is 3.6, and they are determined to be the same target.

[0069] S107. Optimize the observation integration time by using the average neighborhood signal-to-noise ratio and background mean of the target observed in each band and continuously detect the target; specifically, calculate the average background mean of multiple targets. If the ratio of the background gray mean to the image saturation gray value is less than the set threshold, increase the integration time by the minimum unit dT, otherwise decrease it by the minimum unit.

[0070] The schematic diagram of multi-band target detection results after the integration time is optimized provided by the embodiment of the present invention is as Figure 4 shown, Figure 4 (a) is the target detection result of band 1. The signal-to-noise ratio of target 1 is 11.105, the signal-to-noise ratio of target 2 is 10.187, and the average background mean is 7980. Figure 4 (b) is the target detection result of band 2. The signal-to-noise ratio of target 1 is 9.89, and the average background mean is 7370.

[0071] In summary, for the multi-band limb detection method of dim space targets of the present invention, the integration time of the detection system is initialized according to the angle between the optical axis of the multi-band infrared detection system and the Earth and the Sun, and the grayscale image output by the multi-band image sensor is obtained; the target is detected and extracted from the image to obtain multi-band candidate target information; the multi-band detection results are subjected to registration transformation, and the registered and transformed targets are matched and associated, and the target feature information is calculated and output by fusing the multi-band target information. Finally, the integration time of the detection system is optimized and adjusted according to the target area and the grayscale feature information of the neighborhood to achieve the optimal observation of dim targets. In summary, this method uses a multi-band infrared detection system to obtain multi-dimensional target information, fuses the observation results of the detection system to enhance the target energy, improves the detection ability of dim targets, and solves the problem of target detection at long working distances. The integration time of the detection system is adaptively adjusted according to the target detection results to optimize the detection ability of the system to the greatest extent. The above are only embodiments of the present invention, and common general technical solutions or characteristics in the solutions are not described in detail here. It should be noted that for those skilled in the art, without departing from the technical solution of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicability of the patent. The protection scope required by this application shall be subject to the content of its claims, and the specific implementation manners described in the specification can be used to interpret the content of the claims.

Claims

1. A multi-band limb detection method for dim space targets, characterized in that: The following steps are involved: S101. Use a multi-band infrared detection system for space target detection to image the limb or deep space area, where the limb area refers to an area 50km-500km above the earth, and the deep space area refers to an area above 500km above the earth; Among them, the multi-band infrared detection system has the following characteristics: the optical system is cryogenically cooled to below 120K, the detection band is subdivided in the 5-16 micron band according to the target characteristics, and at least two bands are used to construct a planar array multi-band infrared detection system. The detection system shares the main optical system for imaging; the multi-band infrared detection system is a planar array image detector; the output frame rate of the detector is not less than 10 frames / second; the number of infrared channels is not less than two; S102, obtaining the angle between the optical axis of the multi-band infrared detection system and the earth, and the angle between the optical axis of the multi-band infrared detection system and the sun, and setting the initial imaging integration time of each band according to the above two angles; S103, using the multi-band infrared detection system whose integration time is initialized in step S102 to observe and image the edge scene, and obtain multi-band infrared image data; S104, performing target detection on the image data acquired in step S103, and detecting candidate targets in the images of each band; S105, matching candidate targets in each band to select valid targets, the target matching matrix between bands is calculated by observing stars through the multi-band infrared detection system, and the valid target is a target that is simultaneously observed in at least two bands among the multiple bands; S106, calculating the neighborhood signal-to-noise ratio and background mean of the effective target in each band, selecting the two bands with the highest signal-to-noise ratio to perform effective target feature fusion, and encapsulating and outputting them together with the location information and angle information of the effective target; S107, using the average neighborhood signal-to-noise ratio and background mean of the target observed in each band to optimize the observation integration time and continuously detect the target.

2. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S102, the initial imaging integration time of each band is set according to the above two angles, including recording the grayscale values ​​of different integration times of each band under different angle combinations in advance, establishing a mapping table, and selecting the nearest neighbor integration time in the mapping table for setting according to the current angle.

3. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S104, target detection is performed on the image data acquired in step S103, including extracting salient point targets in the image by using a local contrast or morphological filtering method.

4. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S105, matching the candidate targets of each band to select a valid target includes uniformly registering the candidate targets of each band to the image coordinate system of band 1, calculating the Euclidean distance between targets of different bands, and determining the target whose Euclidean distance is less than a set threshold as a valid target. The Euclidean distance M between targets of different bands is calculated according to the following formula: Among them, (x1, y1) is the coordinate data of the candidate target in band 1, and (x2, y2) is the coordinate data of the candidate target in band 2 registered to the image coordinate system of band 1.

5. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S106, the neighborhood signal-to-noise ratio (SNR) of the candidate target in each band is L The calculation formula is as follows: Among them, M T is the maximum value of pixels in the center area of ​​the candidate target, M B is the background mean of the target neighborhood, S B is the target neighborhood background standard deviation, M T The calculation formula is as follows: in, Indicates the central block S0; Background mean M B The calculation formula is as follows: Among them, R s represents the neighborhood background block S, and s represents the radius of S; Among them, S B The calculation formula is as follows: The regions of S and S0 are represented as follows: R S ={(p,q)|max(|p-x|,|q-y|≤s)},s=4,7,10,13。 6. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S106, the two bands with the highest signal-to-noise ratio are selected for effective target feature fusion, including separately extracting the radiation intensity and energy concentration of the target, averaging the energy concentration and image plane position features of the two bands, using the intensities of the two bands to fuse and calculate the temperature and equivalent radiation area, and outputting multi-band target fusion features.

7. The multi-band limb detection method for a faint space target according to claim 1, characterized in that: In S107, the observation integration time is optimized using the average neighborhood signal-to-noise ratio and background mean of the observed target in each band, including calculating the average background mean of multiple targets. If the ratio of the background grayscale mean to the image saturation grayscale value is less than a set threshold, the integration time is increased according to the minimum unit dT, otherwise it is reduced according to the minimum unit.

Images