Moving target calculation tracking imaging method based on high-frame-frequency image shift superposition

Abstract

The invention discloses a moving target calculation tracking imaging method based on high-frame-frequency image shift superposition. The method comprises the following steps: acquiring and preprocessing high-frame-frequency continuous images; estimating the visual position of the moving target in the image plane, and obtaining the measurement coordinate predicted value of the moving target in the multi-frame image; on the basis of the measurement coordinate predicted value, registration of translation of the image sequence is completed on the moving target at the sub-pixel scale, and an image after superposition enhancement is obtained; performing star image detection processing on the image after superposition enhancement; executing track association processing based on the star image detection result, and screening out a moving target sequence based on a preset navigational speed range; outputting information of the moving target; the problem that a moving target in a view field of a photoelectric telescope in a non-tracking imaging mode cannot be tracked and imaged can be effectively solved, effective accumulation of moving target signals and improvement of the signal-to-noise ratio are achieved on the software algorithm, and finally effective detection of a dark and weak moving target and high-precision positioning and light metering of the moving target are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of astronomical image processing and space moving target observation, and in particular to a moving target calculation tracking imaging method based on high frame rate image shifting and superposition. Background Art

[0002] Ground-based photographic astrometry still has an irreplaceable position in the field of observation of moving targets such as near-Earth small celestial bodies and artificial satellites due to its advantages such as low cost and easy deployment. In recent years, low-cost, ultra-large field-of-view ground-based photoelectric telescopes have been widely used in the above-mentioned moving target observation field with their ultra-high observation efficiency. In the large-field observation image, there are stars and moving targets with inconsistent apparent motion speeds. The telescope cannot obtain good star images of reference stars and many moving targets at the same time. When tracking stars, the image of the reference star is good, while the image of the moving target is elongated, the energy cannot be effectively accumulated, and the signal-to-noise ratio is low; when tracking moving targets, the image of the moving target is good, while the image of the reference star is elongated, and high-precision astrometric information cannot be obtained. When a moving target with inconsistent motion speed appears in the field of view, or the motion speed of the moving target is unknown, the tracking problem of the telescope will be completely unsolvable. Therefore, a moving target computational tracking imaging method based on high frame rate image shift superposition is urgently needed to solve the above problems. Summary of the invention

[0003] The present invention aims to provide a moving target computational tracking imaging method based on high frame rate image shifting and superposition, which can effectively solve the problems existing in the above-mentioned prior art.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solution: a moving target calculation tracking imaging method based on high frame rate image shift superposition, comprising the following steps: S1, performing high frame rate continuous image acquisition at a frequency not less than 10 Hz, and preprocessing the acquired images; S2. Based on the ephemeris of near-Earth small celestial bodies and the catalogued elements of space moving targets, the apparent position of the moving target in the image plane is estimated to obtain the measured coordinate prediction value of the moving target in multiple frames of images; S3, based on the measured coordinate prediction values ​​of the moving target in the multiple frames of images, the translation registration of the image sequence of the moving target is completed at the sub-pixel scale to obtain a stacked enhanced image; S4, performing star detection processing on the superimposed and enhanced image; S5. Based on the star detection results, track association processing is performed, and a moving target sequence is selected based on a preset speed range; S6. Output the information of the moving target.

[0005] Preferably, high frame rate continuous image acquisition is performed at a frequency of 10 Hz.

[0006] Preferably, the acquired images are preprocessed to remove the influence of skylight background, instrument background, and fixed pattern noise, specifically including: Set a median filter window with an odd length L, where L = 2K + 1, K is a positive integer, and K is greater than the diameters of all star images; Let the pixel in the image to be processed be I(x, y), where x and y respectively represent the row and column values of the corresponding pixel in the image; Use the long window to perform filtering calculations on the image to be processed row by row and column by column in the x and y directions at once. The value of the pixel after filtering is I′(x, y); ; Divide the pixel I(x, y) in the original image by the filtered image I′(x, y) to complete the preprocessing. The size of the image to be processed is m×n.

[0007] Preferably, based on the dark field image, the fixed pattern noise caused by dark current non-uniformity is deducted, including the following steps: S1. Cover the lens cap of the device and collect a sequence of dark field images in the absence of light signals; S2. Merge the dark field images to suppress the influence of readout noise; S3. Subtract the dark field image from the measured image sequence to deduct the influence of dark current non-uniformity.

[0008] Preferably, in step S2, the orbital elements of the space moving target and the position of the near-Earth small body are obtained; where: For the near-Earth small body, the ephemeris is directly obtained from the website, and its celestial coordinates are interpolated according to the shooting time; For the space moving target, the ephemeris is predicted based on the SPG4 and SDP4 models, and its celestial coordinates are interpolated according to the shooting time Through the gnomonic projection model, the celestial coordinates of the space moving target and the near-Earth small body are converted to the measurement coordinates in the image plane, and the predicted values of the measurement coordinates of the moving target in multiple frames of images are obtained.

[0009] Preferably, the conversion of the celestial coordinates of the moving target to the measurement coordinates is specifically as follows: Establish a rectangular coordinate system with the image center C as the origin. The vertical axis is the projection of the declination circle, and the positive direction is taken as the direction of increasing declination. The horizontal axis is perpendicular to the vertical axis, and the positive direction is taken as the direction of increasing right ascension to construct the ideal coordinate system C- ; Convert the celestial coordinates of the moving target to the ideal coordinates. The conversion formula is: ; where, is the celestial coordinate corresponding to the image center; Solving the negative film model parameters by the least square method based on the polynomial model , the formula is: ; where (x, y) are the measured coordinates of the moving target in the image, and m represents the order of the polynomial; when and the number of (x, y) exceeds the number of parameters to be estimated , the best estimated values of the negative film model parameters are obtained by the least square method.

[0010] Preferably, in step S3, the registration of the translation of the image sequence is completed at the sub-pixel scale to obtain the new shifted image S(i, j), specifically: Taking the first frame image or the middle frame image as a reference, calculate the displacement amount (dx, dy) of the sequential images relative to the reference image, where dx and dy are in pixels; Let P(i, j) be the pixel value of the image, where i and j represent the row and column numbers of the pixels respectively; Perform integer pixel shifting based on the integer parts int(dx) and int(dy) of the displacement amount (dx, dy), that is: Q(i, j)=P(i + int(dx), j + int(dy)); Perform sub-pixel shifting based on the fractional parts dx’ = dx - int(dx) and dy’ = dy - int(dy) of the displacement amount (dx, dy), including: Perform sub-pixel shifting in the row direction: R(i, j)=(1 - dx’) × Q(i, j)+ dx’ × Q(i, j + 1); Perform sub-pixel shifting in the column direction: S(i, j)=(1 - dy’) × R(i, j)+ dy’×R(i + 1, j).

[0011] Preferably, for n moving targets that are predicted to appear in the field of view, based on the same reference image, n groups of sub-pixel shifted image sequences are processed in parallel; the n groups of image sequences are superimposed in parallel to obtain n frames of superimposed and enhanced images for simultaneous tracking imaging of n moving targets in the field of view.

[0012] Preferably, in step S5, when performing track association processing: establish initial tracks for all target pairs in adjacent frame images according to the visual motion speed range of the moving targets, and extrapolate the tracks based on the rules of uniform linear motion, and screen out the sequences with stable tracks and speeds within the set range as the moving target sequences.

[0013] Preferably, in step S6, the moving target information includes the observation time corresponding to the moving target information, the measured coordinates of the moving target, the signal-to-noise ratio, the gray value, as well as the celestial coordinates and instrument stars of the moving target.

[0014] Beneficial effects: The moving target calculation and tracking imaging method based on high-frame-rate image shift and superposition of the present invention can collect wide-field short-exposure images at high frequency through a small-aperture ultra-wide-field optical telescope and an sCMOS camera. The short exposure time can suppress the influence of the sky background, and at the same time, all moving targets and reference stars are "frozen" (not elongated) in the observation image; according to the motion characteristics of the reference stars and various types of moving targets, through the sub-pixel scale image shift and parallel superposition enhancement method, all types of moving targets in the field of view can be simultaneously tracked and imaged in the software algorithm, effectively solving the problem that multiple types of moving targets in the observation image cannot be simultaneously tracked and imaged. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention.

[0016] In the drawings: Figure 1 is a flowchart of the moving target calculation and tracking imaging method based on high-frame-rate image shift and superposition of the invention; Figure 2 is a schematic diagram of sub-pixel image shift of the invention; Figure 3 is an effect diagram of tracking and imaging of a star in the measured star chart sequence by using the method of the present invention in the embodiment; Figure 4 is a tracking and imaging result diagram of a medium Earth orbit moving target (MEO) in the measured star chart sequence by using the method of the present invention in the embodiment; Figure 5 is a tracking and imaging result diagram of a geosynchronous orbit moving target (GEO) in the measured star chart sequence by using the method of the present invention in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] In order to make the purpose and advantages of the present invention clearer, the present invention will be specifically described below in conjunction with embodiments. It should be understood that the following text is only used to describe a moving target calculation and tracking imaging method based on high-frame-rate image shift and superposition of the present invention or several specific implementation manners, and does not strictly limit the specific protection scope claimed by the present invention.

[0018] Embodiment: As Figure 1As shown in the figure, a moving target calculation and tracking imaging method based on high-frame-rate image shift superposition includes high-frame star map acquisition and preprocessing; moving target apparent motion speed estimation; image sub-pixel scale shift; sequential image parallel superposition; star detection; moving target association; and outputting the measurement coordinates and gray values of the moving target, finally realizing the tracking imaging of the moving target star image in the field of view. It can effectively solve the problem that the moving target in the field of view cannot be tracked and imaged in the non-tracking imaging mode of the optoelectronic telescope, realize the effective accumulation of the moving target signal and the improvement of the signal-to-noise ratio in the software algorithm, and finally realize the effective detection of dim moving targets and the high-precision positioning and photometry of moving targets. In a specific embodiment, the telescope used has an aperture of 150 mm and a focal length of 160 mm; the detector is a Dhyana4040 camera, and the target surface size is 4096×4096×9 um. (1) Select the celestial area where dim space moving targets or near-Earth small celestial targets may appear as the object, and collect wide-field short-exposure images at high frequency based on a small-aperture ultra-wide-field optical telescope and an sCMOS camera. The short exposure time can suppress the influence of the sky background. Adopt a high-frame-rate (10 Hz) imaging mode and continuously collect 100 frames of star maps. (2) Preprocess the collected sequential images to remove the influence of sky background, instrument background and fixed pattern noise. Specifically, it includes: (21) Assume that the image to be processed is of size mpixel×npixel, and use two variables x and y to represent the row and column values of a certain pixel in the whole image respectively. I(x, y) represents the intensity value matrix of each pixel when reading the image (the actual size is m×n). (22) Set a median filter window with an odd length L, L = 2K + 1, where K is a positive integer, and K needs to be greater than the radius of all star images to ensure that there are no unfiltered star images in the filtered image.

[0019] (23) Perform filtering calculations row by row and column by column in the x and y directions one by one. The value of each pixel after filtering is represented by I′(x, y): ; I′(x, y) is the low-frequency signal in the observed image. Dividing the original image I(x, y) by I′(x, y) can subtract its non-uniform background and complete the preprocessing. (24) Based on the dark field image, subtract the fixed pattern noise caused by the non-uniformity of dark current, including the following steps: S1. Cover the lens cap of the device and collect a sequence of dark field images in the absence of light signals. S2. Merge the dark field images to suppress the influence of readout noise. S3. Subtract the dark field image from the measured image sequence to subtract the influence of the non-uniformity of dark current. (3) Based on the ephemeris of near-Earth small objects and the catalogued elements of space moving targets, the apparent position of the moving target in the image plane is estimated, and its apparent motion speed and direction are obtained based on the frame rate.

[0020] (31) The orbital elements (two-line elements, TLE) of space moving targets and the positions of near-Earth small objects are obtained through the SpaceTrack website and the IAU MPC website.

[0021] (32) For small near-Earth objects, their ephemeris can be obtained directly from the website, and their celestial coordinates can be obtained by interpolation based on the shooting time; (33) For moving targets in space, the ephemeris is obtained based on the SPG4 and SDP4 model predictions, and then the celestial coordinates are obtained by interpolation according to the shooting time; (34) Through the centroidal projection model, the celestial coordinates of moving targets and near-Earth small celestial bodies are converted into the metric coordinates of the image plane.

[0022] (35) In the star map, a rectangular coordinate system is established with the image center C as the origin, and the vertical axis is the projection of the declination circle, with the direction of increasing declination as the positive direction and the horizontal axis as Perpendicular to axis, taking the direction of increasing right ascension as the positive direction, thus constructing the ideal coordinate system C- ; The celestial (equatorial) coordinates of the moving target Convert to ideal coordinates, the conversion formula is: ; in, is the celestial coordinate corresponding to the center of the image; (36) The least squares method is used to solve the film model parameters based on the polynomial model. , the formula is: ; Among them, (x, y) is the measurement coordinate of the moving target in the image, and the conversion of the moving target's celestial coordinates to measurement coordinates is completed; m represents the order of the polynomial; The number of (x, y) exceeds the parameters to be estimated In the case of quantity, the film model parameters can be obtained by the least squares method The best estimate of (4) Based on the predicted values ​​of the measured coordinates of the moving target in multiple frames, the translational registration of the image sequence is completed at the sub-pixel scale.

[0023] (41) Taking the first frame image or the intermediate frame image as a reference, calculate the displacement of the time series image relative to the reference image.

[0024] Let P(i, j) be the pixel value of the image, where i and j represent the row and column numbers of the pixel respectively; let the displacement to be calculated in the time-series image be (dx, dy), with dx and dy in pixels; the steps of sub-pixel image displacement are divided into two steps: "integer-pixel displacement" and "sub-pixel displacement". (43)The displacement of "integer-pixel displacement" is the integer parts int(dx) and int(dy) of dx and dy, that is: Q(i, j) = P(i + int(dx), j + int(dy)); (44)The displacement of "sub-pixel displacement" is the fractional parts dx' = dx - int(dx) and dy' = dy - int(dy) of dx and dy.

[0025] First, perform "sub-pixel displacement" in the row direction, that is: R(i, j) = (1 - dx') × Q(i, j) + dx' × Q(i, j + 1); Then perform "sub-pixel displacement" in the column direction, that is: S(i, j) = (1 - dy') × R(i, j) + dy' × R(i + 1, j); After the above operations, the new displaced image S(i, j) is obtained.

[0026] (45)For all moving targets that are predicted to appear in the field of view (assuming there are n moving targets), based on the same reference image, n groups of sub-pixel displacement image sequences are processed in parallel; the n groups of image sequences are superimposed in parallel to obtain n frames of superimposed and enhanced images, corresponding to the simultaneous tracking imaging of n moving targets in the field of view.

[0027] Based on the above tracking imaging results, as follows: As Figure 3 shown, it is the calculated tracking imaging result of a stellar target, where dx and dy are approximately 0.460 pixel / s and -3.160 pixel / s respectively.

[0028] As Figure 4 shown, it is the calculated tracking imaging result of a moving target in a certain MEO orbit, where dx and dy are approximately 3.074 pixel / s and 0.667 pixel / s respectively.

[0029] As Figure 5 shown, it is the calculated tracking imaging result of a moving target in a certain GEO orbit, where dx and dy are approximately -0.010 pixel / s and 0.003 pixel / s respectively.

[0030] (5) Perform star detection processing on the superimposed and enhanced image; calculate the background mean and variance of the image, perform binary segmentation on the image, and count the pixel blocks of the connected regions as stars. (6) Based on the star detection results of the sub-pixel scale superimposed and enhanced multi-group measurement images, perform track association processing. Establish initial tracks for all target pairs in adjacent frame images according to the visual motion speed range of moving targets, and extrapolate the tracks based on the rules of uniform linear motion. Only the sequence with a stable track and a speed within the set range is a moving target sequence, avoiding the false alarm influence introduced by shift superposition and stellar interference in the field of view. (7) Output the moving target information, including the observation time corresponding to the moving target information, the measurement coordinates of the moving target, the signal-to-noise ratio, the gray value, as well as the celestial coordinates and instrumental magnitudes of the moving target.

[0031] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments. For those of ordinary skill in the art in this technical field, after learning the content recorded in the present invention, without departing from the principle of the present invention, several equivalent transformations and substitutions can still be made, and these equivalent transformations and substitutions should also be regarded as belonging to the protection scope of the present invention.

Claims

1. A moving target computational tracking imaging method based on high frame rate image shifting and superposition, characterized in that: The steps include: S1, performing high frame rate continuous image acquisition at a frequency not less than 10 Hz, and preprocessing the acquired images; S2. Based on the ephemeris of near-Earth small celestial bodies and the catalogued elements of space moving targets, the apparent position of the moving target in the image plane is estimated to obtain the measured coordinate prediction value of the moving target in multiple frames of images; S3, based on the measured coordinate prediction values ​​of the moving target in the multiple frames of images, the translation registration of the image sequence of the moving target is completed at the sub-pixel scale to obtain a stacked enhanced image; S4, performing star detection processing on the superimposed and enhanced image; S5. Based on the star detection results, track association processing is performed, and a moving target sequence is selected based on a preset speed range; S6. Output the information of the moving target.

2. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1 is characterized in that: High frame rate continuous image acquisition was performed at a frequency of 10 Hz.

3. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1 is characterized in that: The collected images are preprocessed to remove the influence of sky background, instrument background and fixed pattern noise, including: Set the median filter window with an odd length L, L=2K+1, K is a positive integer, and K is greater than the diameter of all stars; Assume that the pixel in the image to be processed is I(x, y), where x and y represent the row and column values ​​of the corresponding pixel in the image respectively; Use a long window to filter the image to be processed row by row and column by column in the x and y directions at a time, and the value of the pixel after filtering is I′(x, y); ； The pixel I(x, y) in the original image is divided by the filtered image I′(x, y) to complete the preprocessing. The size of the image to be processed is m×n.

4. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1 or 3, characterized in that: Based on the dark field image, the fixed pattern noise caused by dark current inhomogeneity is subtracted, including the following steps: S1. Cover the lens cover of the device and collect dark field image sequences in the absence of light signals; S2, merging dark field images to suppress the influence of readout noise; S3. Subtract the dark field image from the measured image sequence to eliminate the influence of dark current non-uniformity.

5. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1, characterized in that: In step S2, the orbital elements of the space moving target and the position of the near-Earth small celestial body are obtained; wherein: For near-Earth small objects, the ephemeris is obtained directly from the website, and then its celestial coordinates are interpolated according to the shooting time; For moving targets in space, the ephemeris is obtained based on the SPG4 and SDP4 model predictions, and then the celestial coordinates are obtained by interpolation according to the shooting time. Through the centroidal projection model, the celestial coordinates of space moving targets and near-Earth small celestial bodies are converted into the metric coordinates of the image plane, and the metric coordinate prediction values ​​of the moving targets in multiple frames of images are obtained.

6. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 5, characterized in that: The conversion of the celestial coordinates of a moving target to the metric coordinates is as follows: A rectangular coordinate system is established with the image center C as the origin. The vertical axis is the projection of the declination circle, the direction of increasing declination is taken as the positive direction, and the horizontal axis is perpendicular to the axis, the direction of increasing right ascension is taken as the positive direction, and the ideal coordinate system C- ; The celestial coordinates of the moving target are converted to ideal coordinates. The conversion formula is: ； in, is the celestial coordinate corresponding to the center of the image; The least square method is used to solve the film model parameters based on the polynomial model , the formula is: ； Where (x, y) is the measurement coordinate of the moving target in the image, and m represents the order of the polynomial. The number of (x, y) exceeds the parameters to be estimated In the case of quantity, the film model parameters are obtained by the least squares method The best estimate of .

7. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1, characterized in that: In step S3, the translation registration of the image sequence is completed at the sub-pixel scale to obtain the shifted new image S(i, j), specifically: Taking the first frame image or the intermediate frame image as the reference, calculate the displacement (dx, dy) of the time series image relative to the reference image, where dx and dy are in pixels; Let P(i,j) be the pixel value of the image, where i and j represent the row and column number of the pixel respectively; Integer pixel shifting is performed based on the integer parts int(dx) and int(dy) of the displacement (dx, dy), that is: Q(i,j)=P(i+int(dx),j+int(dy)); Sub-pixel shifting is performed based on the fractional part of the displacement (dx, dy) dx'=dx-int(dx) and dy'=dy-int(dy), including: Perform sub-pixel shift in the row direction: R(i,j)=(1- dx') ×Q(i,j)+ dx' ×Q(i,j+1); Perform sub-pixel shift in the column direction: S(i,j)=(1- dy') ×R(i,j)+ dy'×R(i+1,j).

8. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 7, characterized in that: For all n moving targets predicted to appear in the field of view, based on the same reference image, parallel processing is performed to obtain n groups of sub-pixel shifted image sequences; the n groups of image sequences are superimposed in parallel to obtain n frames of superimposed enhanced images, corresponding to the simultaneous tracking imaging of n moving targets in the field of view.

9. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1, characterized in that: In step S5, when performing track association processing: establish initial tracks for all target pairs in adjacent frame images according to the apparent motion speed range of the moving target, and extrapolate the tracks based on the rule of uniform linear motion, and select the sequences that establish stable tracks and whose speeds are within the set range as moving target sequences.

10. The moving target computational tracking imaging method based on high frame rate image shifting and superposition according to claim 1, characterized in that: In step S6, the moving target information includes the observation time corresponding to the moving target information, the measurement coordinates of the moving target, the signal-to-noise ratio, the grayscale value, and the celestial coordinates and instrument star of the moving target.

Images