Space target luminosity micro change calculation amplification algorithm

Abstract

The invention relates to a spatial target luminosity micro-change calculation amplification method, which provides support for inverting spatial target characteristic parameters based on luminosity change. The method comprises the following steps of: performing regression analysis on a luminosity micro-change signal containing noise by adopting a mode of firstly fitting and then extracting to obtain a regression curve of the micro-change signal; and subtracting the fitted regression signal from the noisy luminosity micro-change signal to obtain a mean-removed luminosity micro-change signal, which is called a difference signal for short, and the difference signal comprises both an effective target luminosity micro-change signal and noise. Luminosity micro-change extraction is converted into effective micro-change signal extraction from the difference signal; the method comprises the following steps of: decomposing a difference signal based on an aggregation empirical mode decomposition technology, constructing an effective micro-change signal by estimating a signal-to-noise ratio of different mode component signals after decomposition and a threshold judgment rule of permutation entropy, adding the effective micro-change signal and a fitting signal, setting an amplification factor based on a signal mean deviation value of an addition result, and finally obtaining a differential signal. And outputting a luminosity micro-change calculation amplification result. After the luminosity micro-change signal is amplified, the mean value of the signal is not changed, the signal-to-noise ratio is improved, effective micro-change signal characteristics are highlighted, and support is provided for subsequent luminosity curve analysis.

Description

Technical Field

[0001] The present invention relates to the field of photometric characteristic analysis of space targets, and in particular to a calculation and amplification algorithm for micro-changes in photometric information of space targets. Background Art

[0002] There is a correlation between the luminosity change of space targets and the relative position of the sun-target-observation station (described by the phase angle), target state, posture, shape and surface material. Therefore, inverting the relevant characteristics of non-cooperative space targets by sensing the luminosity change of space targets is one of the important supporting technologies of the space target monitoring system. However, due to uncertain factors such as detection conditions, target posture and target state, the luminosity change signals of non-cooperative space targets captured by the monitoring system are mostly micro-change signals. On the one hand, micro-change signals lead to low estimation accuracy of space target related characteristic parameters, and on the other hand, they may cause the perception system to fail to work effectively and timely, reducing the timeliness of the system. How to effectively perceive the luminosity change information of space targets plays an important role in accurately inverting target related characteristics, and is also a difficult problem in the direction of space target detection.

[0003] In reality, micro-change signals contain certain noise interference, and directly amplifying them will cause the noise signal to be amplified as well. Therefore, it is necessary to design a specific algorithm to extract effective micro-changes. At the same time, the measured light curve is limited by practical factors such as observation cost and observation conditions. It is usually not data sampled at equal intervals with the phase angle of the independent variable, so it is difficult to perform operations such as time-frequency analysis. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the problems existing in the above-mentioned practical applications, the present invention provides an algorithm for calculating and amplifying the luminosity micro-changes of space targets. For an input luminosity micro-change signal containing noise, the algorithm fits the overall trend of the curve, extracts the effective micro-change part and then amplifies it. The luminosity micro-change curve is calculated and amplified while keeping the overall mean value of the luminosity unchanged, and the amplified luminosity curve has a higher signal-to-noise ratio than the original micro-change signal, that is, more effective features are retained.

[0006] (II) Technical solution

[0007] The present invention proposes an algorithm for calculating and amplifying the micro-changes in the luminosity of space targets, which includes six main steps: overall fitting of the luminosity curve trend, equally spaced sampling, ensemble empirical mode decomposition (EEMD) of the difference signal, extraction of effective micro-changes, wavelet denoising, and amplification according to the user's preset magnification. The specific steps are:

[0008] S1. Input the photometric signal, use the appropriate order Fourier series to fit the photometric signal curve, and obtain the overall trend signal of the photometric curve;

[0009] S2. Equally spaced sampling: The input photometric micro-change signal and the fitting signal are sampled at an appropriate phase angle sampling rate, and the fitting signal is subtracted from the sampled micro-change signal to obtain an equally spaced sampling difference signal. At the same time, the difference signal is set to 0 at the phase where the original data does not exist (indicating that the difference estimation is not performed on the non-existent data);

[0010] S3. Decompose the difference signal based on the EEMD method to obtain the intrinsic mode function components (IMF) of the difference signal, which are denoted as IMFs;

[0011] S4. In the series of EIMF components, if the signal-to-noise ratio is higher than the preset signal-to-noise ratio threshold, that is, it contains less noise, it is directly regarded as part of the effective micro-change and added;

[0012] S5. If the signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold, determine whether the permutation entropy of the component signal is greater than the preset permutation entropy threshold, and add the EIMF component signal less than the threshold as an effective micro-change, and add the component with a signal-to-noise ratio higher than the preset signal-to-noise ratio threshold as an effective luminosity micro-change signal. The IMFs component greater than the permutation entropy threshold is further subjected to empirical mode decomposition (EMD) to obtain the empirical intrinsic mode function component, recorded as IMF;

[0013] S6. Determine whether the permutation entropy of the IMF component signal is greater than a preset permutation entropy threshold, and add effective micro-changes to components that are less than the threshold;

[0014] S7. The permutation entropy of the IMF component signal is greater than the preset permutation entropy threshold, and the IMF component is processed by wavelet denoising and added to the effective micro-change signal component sequence after denoising;

[0015] S8. Calculate and amplify, add the fitted trend signal in step S1 to the effective micro-change signal, and obtain the micro-change calculated amplified photometric signal based on the preset amplification factor.

[0016] The above technical solution has the following beneficial effects:

[0017] 1. The algorithm framework of "fitting first, then extracting" is adopted, which not only ensures that the overall trend characteristics of the light curve are retained, but also gradually extracts effective micro-change signals from the difference signal from coarse to fine according to criteria such as signal-to-noise ratio and permutation entropy. It can effectively remove the noise signal part and steadily improve the signal-to-noise ratio of the signal after micro-change amplification.

[0018] 2. The problem that non-uniformly spaced sampling data cannot be used for time-frequency analysis is solved, making modal decomposition possible.

[0019] 3. EEMD decomposition utilizes the characteristic of uniform filling of white noise in the time domain and frequency domain. By adding white noise with a preset signal-to-noise ratio to the input signal and averaging it multiple times, the decomposed IMFs can be made less prone to modal aliasing, that is, each IMF is decomposed into modal components with actual physical meaning.

[0020] 4. The original luminosity slight change signal is calculated and amplified, and the mean remains unchanged. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which constitute a part of this specification, are included to provide a further understanding of the present invention.

[0022] Figure 1 It is an overall algorithm flow chart of the present invention;

[0023] Figure 2 It is a specific flowchart for extracting effective micro-change parts from the overall algorithm flowchart;

[0024] Figure 3 is the input original slightly changed signal legend;

[0025] Figure 4 is the legend of the fitted signal;

[0026] Figure 5 is the difference signal legend;

[0027] Figure 6 This is the legend of IMFs signal after EEMD decomposition;

[0028] Figure 7 It is to use the signal-to-noise ratio to preliminarily extract the effective micro-change signal legend;

[0029] Figure 8 This is the legend of the effective micro-change signal results;

[0030] Fig. 9 This is a signal example after the output slight change is amplified. DETAILED DESCRIPTION

[0031] The following detailed descriptions are all exemplary and are intended to provide further explanation of the present invention and to clearly and completely describe the technical solutions in the embodiments of the present invention. All technical and scientific terms used herein have the same meanings as those generally understood by ordinary technicians in the technical field to which the present invention belongs.

[0032] In this embodiment, if Figure 1 As shown in FIG. 1 , it is an overall flow chart of an algorithm for calculating and amplifying a slight change in the luminosity of a space target proposed by the present invention, which includes the following steps:

[0033] S1. Input the photometric signal, use the appropriate order Fourier series to fit the photometric signal curve, and obtain the overall trend signal of the photometric curve;

[0034] S2. Equally spaced sampling: the input luminosity micro-change signal and the fitting signal are sampled at an appropriate phase angle sampling rate, and the fitting signal is subtracted from the sampled micro-change signal to obtain an equally spaced sampling difference signal;

[0035] S3. Decompose the difference signal based on the EEMD method to obtain the intrinsic mode function components (IMF) of the difference signal, which are denoted as IMFs;

[0036] S4. In the series of EIMF components, if the signal-to-noise ratio is higher than the preset signal-to-noise ratio threshold, that is, it contains less noise, it is directly regarded as part of the effective micro-change and added;

[0037] S5. If the signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold, determine whether the permutation entropy of the component signal is greater than the preset permutation entropy threshold, and add the EIMF component signal less than the threshold as an effective micro-change, and add the component with a signal-to-noise ratio higher than the preset signal-to-noise ratio threshold as an effective luminosity micro-change signal. The IMFs component greater than the permutation entropy threshold is further subjected to empirical mode decomposition (EMD) to obtain the empirical intrinsic mode function component, recorded as IMF;

[0038] S6. Determine whether the permutation entropy of the IMF component signal is greater than a preset permutation entropy threshold, and add effective micro-changes to components that are less than the threshold;

[0039] S7. The permutation entropy of the IMF component signal is greater than the preset permutation entropy threshold, and the IMF component is processed by wavelet denoising and added to the effective micro-change signal component sequence after denoising;

[0040] S8. Calculate and amplify, add the fitted trend signal in step S1 to the effective micro-change signal, and obtain the micro-change calculated amplified photometric signal based on the preset amplification factor.

[0041] In the above scheme, the input of step S1 is the original spatial target luminosity change signal, that is, the luminosity micro-change signal containing noise and non-uniformly spaced sampling. First, the fitting step is performed, and the Fourier series is used to fit the original signal to obtain a fitted luminosity trend signal curve. The order of the Fourier series is usually selected from 4 to 6. Then, the step of extracting the effective luminosity micro-change signal is entered, and the original luminosity signal and the fitted trend luminosity signal are used as input to extract the effective micro-changes therein. The specific process is as shown in the attached figure. Figure 2Finally, in the micro-change calculation and amplification step, the effective micro-change signal and the equally spaced sampling fitting signal are first added, the mean of the added signal is calculated, the offset relative to the mean is amplified, and finally the amplified signal is output as the photometric micro-change amplification result.

[0042] like Figure 2 As shown, it is a specific flowchart for extracting the effective micro-change part in the overall algorithm flowchart.

[0043] In the above scheme, step S2 is the trend signal fitting and non-uniform sampling signal resampling. Taking a certain high-orbit satellite measured luminosity micro-change signal as an example, its luminosity curve is as follows Figure 3 As shown, the magnitude range is less than 1. The specific process includes the following:

[0044] S21. Perform a 4th-order Fourier fitting on the original luminosity slight change signal to obtain the overall trend change signal, such as Figure 4 As shown;

[0045] S22. The original signal is interpolated using the cubic Hermite interpolation method to obtain an equally spaced sampling signal. The fitted trend signal is sampled at a sampling frequency of 20 Hz. The difference signal is obtained by subtracting the resampled original photometric signal from the trend signal. The difference signal takes the value of 0 at the phase where the original data does not exist (indicating that the difference estimation is not performed on the non-existent data). The results are as follows: Figure 5 shown.

[0046] In the above scheme, the step S3 is to decompose the difference signal using the EEMD method, which specifically includes the following process:

[0047] S31. A series of IMFs are obtained by EEMD decomposition of the difference signal, such as Figure 6 As shown, the numbers 1-10 represent the decomposition into 10 IMF components, and the bottom-up order refers to the order from IMF10 to IMF1;

[0048] S32. The effective micro-change signal obtained by superimposing the IMF components in a bottom-up order is as follows: Figure 7 As shown, the first one in the first row is the IMF10 component, the second one in the first row is the IMF10+IMF9 component, the first one in the second row is the IMF10+IMF9+IMF8 component, and so on. The more superimposed components there are, the higher the signal-to-noise ratio relative to the difference signal.

[0049] In the above scheme, the specific process of steps S4 and S5 is: superimpose IMF components on IMFs from bottom to top (component numbers from large to small) to construct an effective slight change signal. If the signal-to-noise ratio of the constructed effective slight change signal relative to the difference signal is greater than a preset threshold, proceed to the next step, otherwise repeat the current step to continue superimposing IMF components.

[0050] In the above scheme, the specific process of steps S6 and S7 is as follows: calculate the permutation entropy of IMF1, if the permutation entropy is less than the threshold (the threshold is 2), add it to the previous effective micro-change, otherwise perform EMD decomposition to obtain new IMF components, and again determine whether the permutation entropy of each IMF component is less than the threshold, if so, add it to the effective micro-change, otherwise perform wavelet denoising on the IMF component, and add it to the effective micro-change signal component sequence after denoising. Since IMF1 is the last IMF component obtained by EEMD decomposition, the effective micro-change extraction step is completed (usually all remaining IMFs should be subjected to the above calculation process for IMF1). The extracted effective micro-changes are as follows Figure 8 As shown, it contains less noise than the difference signal.

[0051] In the above scheme, the specific process of step S8 is as follows: the effective micro-change signal is interpolated by three-step Hermite interpolation according to the independent variable range of the original signal, and then added to the value sampled by the fitting curve at the independent variable range of the original signal. The average of the added signal is taken, and the offset is obtained by subtracting the average from the added signal, and the offset is amplified by a preset multiple of 8 times, and then the average is added to finally obtain the micro-change amplified signal, such as Fig. 9 shown.

[0052] Although the specific implementation methods of the present invention are described above in conjunction with the accompanying drawings, this does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work on the basis of the technical solution of the present invention are within the scope of protection of the present invention.

Claims

1. An algorithm for calculating and amplifying micro-changes in the luminosity of a space target, characterized by: The input is a noisy non-uniformly spaced sampling signal (called the original signal) of the space target's photometric value (magnitude) that changes slightly with phase angle or time, and the output is a valuable amplified signal of slight changes in the photometric signal.

2. The algorithm for calculating and amplifying the micro-change in luminosity of a space target according to claim 1, characterized in that: The method of "fitting first, then extracting" is adopted to regress and analyze the noisy photometric micro-change signal to obtain the regression curve of the micro-change signal; the photometric micro-change signal containing noisy light intensity is subtracted from the fitted regression signal to obtain the photometric micro-change signal without the mean, referred to as the difference signal, which contains both the effective target photometric micro-change signal and the noise. Photometric micro-change extraction is transformed into extracting the effective micro-change signal from the difference signal; The difference signal is decomposed based on the aggregated empirical mode decomposition technology. The time domain bandpass filter is constructed by estimating the signal-to-noise ratio of the different modal component signals after decomposition. The target luminosity micro-change signal is calculated and amplified under the framework of the complex steerable pyramid to obtain the target luminosity micro-change signal. After the luminosity micro-change signal is amplified, the mean value of the signal does not change, the signal-to-noise ratio is improved, and the effective micro-change signal characteristics are highlighted, providing support for subsequent luminosity curve analysis.

3. The fitting original luminosity micro-change curve as claimed in claim 2, characterized in that: The fitting curve uses the Fourier function, and the difference signal is sampled at equal intervals at a certain sampling frequency fs.

4. The method for extracting an effective micro-change signal from a difference signal as claimed in claim 2, characterized in that: The difference signal is modally decomposed, and the effective micro-change signal is constructed from a series of intrinsic mode function components using the signal-to-noise ratio and permutation entropy threshold conditions.

5. The method for performing modal decomposition on a difference signal according to claim 2, wherein: Integrated empirical mode decomposition is used in combination with post-processing to obtain a series of intrinsic mode function components; integrated empirical mode decomposition selects the appropriate number of integrations and the signal-to-noise ratio of the Gaussian white noise used in the integration. The signal-to-noise ratio should be the estimated noise signal-to-noise ratio, which is taken as the signal-to-noise ratio of the fitting signal relative to the original signal.

6. The method of constructing an effective micro-change signal from an intrinsic mode function component as claimed in claims 3, 4 and 5, characterized in that: The intrinsic mode function components are superimposed from bottom to top to obtain a newly constructed effective slight change signal, and the signal-to-noise ratio between the signal and the difference signal is calculated until the signal-to-noise ratio is greater than the preset threshold and the cycle is stopped; for the remaining intrinsic mode function component signals, if the permutation entropy is less than the threshold, they are added to the effective slight change signal, otherwise, the permutation entropy of each intrinsic mode function component is calculated after integrated empirical mode decomposition, and if it is less than the threshold, it is added to the effective slight change signal.

7. The method of claim 6 wherein the overall mean value does not change after the calculation and amplification of the slightly changed signal is performed, wherein: The effective micro-change signal and the fitting signal are added and the mean is calculated, and the difference between the mean and the mean is amplified by a preset multiple.

Images