CN102879770B - Target vibration detection method on basis of SAL (Synthetic Aperture Radar) echo data - Google Patents

Abstract

The invention relates to a target vibration detection method on the basis of SAL (synthetic aperture radar) echo data. The method comprises the following sequential steps of: carrying out subaperture division on laser echo data output by a balanced detector in the azimuthal direction; respectively carrying out azimuthal compression on the data subjected to subaperture division to obtain primary subaperture image data and calculating the entropy of each subaperture image; and classifying sub-images according to the entropy values by a K fuzzy C-mean clustering algorithm, determining bad data formed due to the influence of the target vibration, and marking the bad data. According to the invention, the rules and the characteristics of the influence of the vibration on the SAL imaging are sufficiently utilized; the target vibration is directly detected from the SAL echo data; the dependence on other auxiliary equipment and means is avoided; and the target vibration detection method on the basis of the SAL echo data is convenient to implement the SAL imaging coherent compensation under the influence of the vibration and promote the imaging accuracy.

Description

Target method for detecting vibration based on the SAL echo data

Technical field

The present invention relates to laser imaging field, especially a kind of target method for detecting vibration based on the SAL echo data.

Background technology

Along with the progress of laser technology, laser pick-off technology, growing along with each field of society to the detected with high accuracy demand, technique of laser imaging just more and more gets more and more people's extensive concerning.Particularly it,, in the progressively application in the fields such as remote sensing, mapping, atmospheric seeing, resource exploration, has more promoted imaging technique and has constantly advanced in recent years.

Synthetic Aperture Laser Radar (Synthetic Aperture Radar, SAL) is the expansion of synthetic aperture imaging technology in optical band, can make up the low deficiency of radar band synthetic aperture imaging resolution.The SAL imaging mainly utilizes phase of echo to extract the information of target, therefore, detects and compensation phase of echo error, and be the most key link in SAL system and imaging processing technology.The return laser beam phase place is subject to the impact of many factors, as the dispersion of the transmission mediums such as high-order phase error, atmospheric scattering and the turbulent flow of the pattern that transmits, linear frequency modulation light source itself, optical fiber and decay, echoed signal receive the scattering cross-section characteristic of system, transmitting/receiving optical system characteristic, target itself etc., often be difficult to obtain accurate target echo phase information, thereby cause the objective attribute target attribute measuring error.The phase error different from above-mentioned factor, that between target and SAL system, relative motion causes, particularly target vibration error, be difficult to solve by the optimal design of system self, need to be detected by the processing of echo data and process.At present, also by the SAL echo data, do not carry out the detection method of direct-detection target vibration.

Summary of the invention

The object of the present invention is to provide a kind of can be from the SAL echo data vibration of direct-detection target, be convenient to realize the relevant compensation of SAL imaging under vibration effect, promote the target method for detecting vibration based on the SAL echo data of imaging precision.

For achieving the above object, the present invention has adopted following technical scheme: a kind of target method for detecting vibration based on the SAL echo data, and the method comprises the step of following order:

(1) the return laser beam data of balance detection device output are upwards carried out to sub-aperture segmentation in orientation;

(2) data after sub-aperture segmentation are carried out respectively to Azimuth Compression and obtain preliminary sub-subaperture image data, and calculate the entropy of each sub-subaperture image;

(3) use the K means clustering algorithm according to entropy, each subimage to be classified, determine the bad data that formed by target vibration effect, bad data is carried out to mark.

As shown from the above technical solution, the present invention is the rule that affects on the SAL imaging according to the target vibration, echo data is carried out to the decomposition of sub-aperture, differentiate target vibration effect by the method that detects sub-subaperture image entropy, through the K means clustering algorithm, each sub-aperture data are carried out to cluster according to entropy, and then the period of judgement target vibration appearance, after the bad data that the target vibration is caused carries out mark, enter follow-up imaging processing flow process.The present invention takes full advantage of vibration affects rule and characteristic to the SAL imaging, direct-detection target vibration from the SAL echo data, avoid the dependence to other utility appliance and means, be convenient to realize the relevant compensation of SAL imaging under vibration effect, promoted imaging precision.

The accompanying drawing explanation

Fig. 1 is workflow diagram of the present invention.

Embodiment

As shown in Figure 1, a kind of target method for detecting vibration based on the SAL echo data, the method comprises the step of following order:

The first, the return laser beam data of balance detection device output are upwards carried out to sub-aperture segmentation in orientation.For convenience of processing, sub-aperture segmentation is the return laser beam data upwards to be divided into to 2 a Nth power sub-subaperture image data in orientation, if it is many that the number of sub-subaperture image data is got, be the N value get large, may be lacked by the data of mistake mark after carrying out Check processing, yet do like this operand that has increased on the one hand the vibration error compensation, make on the other hand entropy gap corresponding between suffer a loss data and normal data dwindle, be unfavorable for differentiating, therefore, the value of General N is got 3 or 4 and is got final product.

The second, the data after sub-aperture segmentation are carried out respectively to Azimuth Compression and obtain preliminary sub-subaperture image data, and calculate the entropy of each sub-subaperture image.After carrying out sub-aperture segmentation, at first the sub-subaperture image after cutting apart is carried out to orientation to compression, then in antithetical phrase subaperture image data, the amplitude of each point is carried out normalization, finally calculates the entropy of sub-subaperture image.Consider the impact of the impact of the factor antithetical phrase image entropies such as speed, acceleration error, range migration far away from the target vibration, here without carrying out range migration correction and phase error estimation and phase error, improved operation efficiency.

The formula of entropy that calculates sub-subaperture image is as follows:

$C_H=-\underset{x=0}{\overset{M-1}{\mathrm{\Σ}}}\underset{y=0}{\overset{N-1}{\mathrm{\Σ}}}\stackrel{\‾}{I}(x,y)1n\left[\stackrel{\‾}{I}(x,y)\right]---\left(1\right)$

Wherein

be the value after picture amplitude normalization, have

$\stackrel{\‾}{I}(x,y)=\frac{{\left|\tilde{g}(x,y)\right|}^2}{\underset{x}{\mathrm{\Σ}}\underset{y}{\mathrm{\Σ}}{\left|\tilde{g}(x,y)\right|}^2}---\left(2\right)$

The definition that formula (1) is image entropy, formula (2) is asked is range value g(x, the y of (x, y) point on image) value after normalization, M, N be the pixel number of corresponding x, y axle respectively.

The 3rd, use the K means clustering algorithm according to entropy, each subimage to be classified, determine the bad data that formed by target vibration effect, bad data is carried out to mark.The method step that uses the K means clustering algorithm according to entropy, each subimage to be classified is as follows:

A) choose initial cluster center;

B) sample point is classified;

C) adjust cluster centre, and interative computation.

After classification completes, judge whether a class of entropy maximum is greater than setting threshold, if judgment result is that, be, be determined with the target vibration error, carry out the bad data mark; Otherwise, do not carry out vibration compensation.After to the bad data mark, carry out subsequent treatment, in the imaging processing of next carrying out, abandon need not, or be different from the error compensation of good data, thereby avoided bad data infringement image quality.

In a word, advantage of the present invention is as follows:

1, by extraction and the classification of SAL image entropy feature, solve target vibration error in the SAL imaging and detected a difficult problem, can realize the target vibration detection based on echo data;

2, take full advantage of target velocity, acceleration error and the vibration error difference on the image entropy impact, the target vibration error that will mix among the Velocity-acceleration error detects, and has solved the difference problem of target vibration error and Velocity-acceleration error;

3, in sub-subaperture image entropy detects, an antithetical phrase aperture data is carried out Azimuth Compression, and, without factors such as consideration range migrations, has greatly reduced the operand of algorithm, has improved detection efficiency;

4,, by reasonable sub-aperture segmentation, the bad data mark, for follow-up imaging processing provides high-quality echo data, reduced the phase error compensation difficulty.

Claims (5)

1. the target method for detecting vibration based on the SAL echo data, the method comprises the step of following order:

(1) the return laser beam data of balance detection device output are upwards carried out to sub-aperture segmentation in orientation;

(2) data after sub-aperture segmentation are carried out respectively to Azimuth Compression and obtain preliminary sub-subaperture image data, and calculate the entropy of each sub-subaperture image; After carrying out sub-aperture segmentation, at first the sub-subaperture image after cutting apart is carried out to orientation to compression, then in antithetical phrase subaperture image data, the amplitude of each point is carried out normalization, finally calculates the entropy of sub-subaperture image;

(3) use the K means clustering algorithm according to entropy, each subimage to be classified, determine the bad data that formed by target vibration effect, bad data is carried out to mark.

2. the target method for detecting vibration based on the SAL echo data according to claim 1 is characterized in that: described sub-aperture segmentation is the return laser beam data upwards to be divided into to 2 a Nth power sub-subaperture image data in orientation.

3. the target method for detecting vibration based on the SAL echo data according to claim 1 is characterized in that: the method step that uses the K means clustering algorithm according to entropy, each subimage to be classified is as follows:

A) choose initial cluster center;

B) sample point is classified;

C) adjust cluster centre, and interative computation.

4. the target method for detecting vibration based on the SAL echo data according to claim 1, it is characterized in that: after classification completes, judge whether a class of entropy maximum is greater than setting threshold, if judgment result is that be, be determined with the target vibration error, carry out the bad data mark; Otherwise, do not carry out vibration compensation.

5. the target method for detecting vibration based on the SAL echo data according to claim 1 is characterized in that: the formula of entropy that calculates sub-subaperture image is as follows:

Wherein

be the value after picture amplitude normalization, have

The definition that formula (1) is image entropy, formula (2) is asked is range value g(x, the y of (x, y) point on image) value after normalization, M, N be the pixel number of corresponding x, y axle respectively.

Images