CN101226639A - Relative radiometric correction method for star-load TDICCD camera - Google Patents

Abstract

Disclosed is a relative radiometric correction method for the satellite-borne TDICCD camera. The steps comprises (1) analyzing the response output of a tapping under different light intensities and acquiring a scatter diagram of tapping light intensity response, (2) carrying out linear fitting to the scatter diagram of tapping light intensity response, then carrying out linear interpolation to the scatter diagram to obtain an interpolation curve, subtracting the interpolation curve of each tapping from respective fitting line to obtain a non-linear modified function of the tapping to light intensity, (3) using the non-linear modified function to carry out non-linear modification one after another to tapping of radiometric calibration data, and obtaining modified radiometric calibration data, (4) processing the modified radiometric calibration data and obtaining a modified inter- calibration coefficient, (5) carrying out radiation homogeneousness correction to arbitrary image produced by TDICCD via using the non-linear modified function and the modified inter- calibration coefficient. The invention resolves the problem that non-linear response to light intensity of the tapping can not be corrected in existing radiometric calibration method, and can increase relative radiometric calibration accuracy.

Description

A kind of relative radiation correction method of star-load TDICCD camera

Technical field

The present invention relates to a kind of relative radiation correction method of remote sensing satellite TDICCD camera, particularly a kind ofly fit and the relative uniformity of radiation bearing calibration of analyzing based on tap.

Background technology

General remote sensing satellite will carry out repeatedly the radiation calibration test to remote sensor (referring to the TDICCD camera here) on ground.For improving treatment effeciency and the uniformity of radiation calibration result of satellite to a large amount of calibration data, the relative calibration precision of evaluation remote sensor proposes a kind of new method of putting in order camera radiation calibration data processing under the starlike attitude, image non-uniform correction and relative calibration precision analysis.

Along with the raising of performances such as the field angle of space camera, spatial resolution, fabric width,, need configuration multi-disc TDICCD device be spliced into long-line array in order to satisfy big visual field, the wide demand of wide cut.Because the influence of composite factors such as optical system, detector splicing, TDICCD photosensitive unit response inconsistency and sensing circuit, make remote sensing camera under same even light initial conditions, the output signal amplitude difference of different pixels, be the system responses heterogeneity, on image, show as uneven row to striped.Every CCD is one group of output with 256 pixels usually, and one group of such physical location is referred to as a tap.The system responses unevenness mainly shows as inhomogeneous between each tap and between each pixel.The fundamental purpose of therefore whole star radiation calibration is to determine the relation of input spoke brightness and system output signal on the one hand, is on the other hand to carry out relative radiant correction to handle, and the heterogeneity of total system is proofreaied and correct (abbreviation radiant correction).

Utilize the view data of radiation calibration to analyze, the process of trying to achieve the correction factor of each pixel of CCD device is referred to as the relative radiometric calibration data processing, and the heterogeneity of carrying out the radiation calibration image behind the homogeneity correction just is referred to as the relative calibration precision.

Radiation calibration response asymmetric correction method commonly used has: any calibrates correcting algorithm, 2 calibration correcting algorithms and multiple spot calibration segmentation correcting algorithm.As Zhang Yufeng etc., " the 02 star CCD camera laboratory radiation calibration Algorithm Analysis of satellite of mini-bus resource ", space flight are returned and remote sensing, and 2005 (06), 41～45; Guo Jianning etc., " the relative radiant correction research of CBERS-01/02 Satellite CCD image ", Chinese science E collects information science 2005,35 (supplementary issue I), 11～25; Li Yang etc., " TDICCD camera system response Nonuniformity Correction ", Chinese journal of scientific instrument, 2006 (6) supplementary issues, 1226～1227; Relative radiometric calibration method in these literary compositions all is based on the CCD device response of light intensity is linear hypothesis.For the response of the light intensity of TDICCD pixel when non-linear, can not play good corrective action, especially at the response linearity of different taps not simultaneously, the radiant correction effect is also bad, be unfavorable for producing the high-quality satellite image, also can cause adverse effect the further application of image.

Chinese patent application number: 200410060986.8, title: the relative radiation correction method of a kind of push-broom type satellite image CCD, this patent emphasis is handled device splicing overlapping region, does not relate to non-linear correction content, does not also relate to the tap correction.

Do not retrieve the non-linear correction of external tap aspect related data.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiency that prior art is only carried out linearity or piecewise linearity radiant correction, the relative radiation correction method of the non-linear correction of each tap response of a kind of TDICCD of fusion is provided, this method has solved can't proofread and correct tap in the existing radiation calibration method light intensity is responded nonlinear problem, can significantly improve the Nonuniformity Correction effect of relative radiometric calibration precision and remote sensing images.The present invention has only also overcome at a light intensity levels or two luminance brightness and has proofreaied and correct shortcoming into the homogeneity coherent image respectively, can carry out the radiation calibration treatment for correcting at any gray scale.

Technical solution of the present invention is: a kind of relative radiation correction method of star-load TDICCD camera is characterized in that step is as follows:

(1) tap light intensity response analysis: analyze the response output situation of tap under different light intensity levels, try to achieve tap light intensity response scatter diagram;

(2) response curve fits with correction function and determines: the tap light intensity response scatter diagram that step (1) is obtained carries out straight line and fits, point to the corresponding formation with ordinate of tap light intensity response scatter diagram horizontal ordinate carries out linear interpolation again, obtain interpolation curve, with the interpolation curve of each tap with separately fit straight line do poor, obtain the non-linear correction function of tap, i.e. the brightness correction value curve of tap to light intensity;

(3) the non-linear correction of radiation calibration data: the non-linear correction function that uses step (2) to obtain carries out non-linear correction one by one to the tap of radiation calibration data, obtains being used for determining the correction radiation calibration data of relative calibration coefficient;

(4) revising the relative calibration coefficient determines: carry out data processing to carrying out non-linear revised calibration data, obtain revising the relative calibration coefficient;

(5) image that TDICCD is produced carries out radiant correction and handles: image additional corrections value before at first utilizing non-linear correction function that step (2) obtains to radiant correction, the i pixel gray-scale value after the nonlinear response has been eliminated.The correction relative calibration coefficient that utilizes step (4) to obtain then is updated in the original non-uniform image, and each pixel image of any gray scale that TDICCD is produced is handled again, obtains through revised even image.

The original heterogeneity image that TDICCD is produced carries out after radiant correction handles, also can carry out quantitative evaluation to the radiation calibration result: the corresponding down homogeneity correction image of different gray scales that utilizes step (5) to obtain, the response consistance behind the camera relative radiometric calibration, relative radiometric calibration precision, twice calibration repeatable accuracy, dynamic range curve are evaluated.

The tap light intensity response analysis of described step (1) is: the output image of supposing to have p different brightness degrees, p the image averaging gray-scale value that obtains after the gradation of image value of p different brightness is averaged respectively responds the horizontal ordinate of scatter diagram as light intensity, is ordinate with each tap at the average image gray-scale value of each brightness, obtains the light intensity response scatter diagram of whole taps.

Non-linear being modified to of radiation calibration data of described step (3): the non-linear correction function that is obtained by step (2) is tried to achieve the non-linear modified value of tap correspondence, then to each the pixel additional corrections value in the tap, after each tap revised one by one, finally obtain being used for determining revising radiation calibration data image after the correction of relative calibration coefficient.

The correction relative calibration coefficient of described step (4) is defined as: the correction radiation calibration data that step (3) is obtained according to

$X_{i_k}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{X}_{i,j_k},(i=\mathrm{1,2},...,m.k=\mathrm{1,2},...p)$

Calculate the nominal output gray level value X of each pixel under p the different brightness respectively _iX in the formula _IkBe the nominal gray scale column mean of i pixel under the brightness of k level, k=1,2 ... p, p are the high-high brightness grade; X _{I, jk}, be pixel under the brightness of k level (i, output numerical value j); M is the pixel columns picture traverse of TDICCD just; N is the pixel line number, just the image length of TDICCD; Again according to

$Y_k=\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{X}_{i_k},(k=\mathrm{1,2},...p)$

Calculate the average output Y of m pixel under p the different brightness _kY in the formula _kAverage output for m pixel under the brightness of k level; M pixel carried out straight line with the method for least square one by one fit, obtain the correction relative calibration coefficient a of each pixel of TDICCD _i', b _i'.

The image that in the described step (5) TDICCD is produced carries out radiant correction and is treated to: the non-linear correction function that utilizes step (2) to obtain, non-linear correction is carried out in each tap, and the correction relative calibration coefficient that utilizes step (4) to obtain then, according to

R _i，j＝a _i′·X _i，j+b _i′(i＝1，2，...m.j＝1，2，...n)

The arbitrary image that TDICCD is produced carries out radiant correction; A in the formula _i', b _i' be the correction relative calibration coefficient of i pixel, X _{I, j}Be pixel (i, output numerical value j), R _{I, j}For proofreading and correct result value.

Relative radiometric calibration precision behind the camera relative radiometric calibration is assessed as:

$\mathrm{\η}=\frac{\mathrm{Std}}{\stackrel{\‾}{R_{i,j}}}$

Wherein

$\mathrm{Std}=\sqrt{\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{[R_{i,j}-\stackrel{\‾}{R_{i,j}}]}^2}\underset{}{,}(i=\mathrm{1,2},...m,j=\mathrm{1,2}...n)$

$\stackrel{\‾}{R_{i,j}}=\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{R}_{i,j}$

In the formula, image has m * n pixel, and Std is the standard variance of image behind the homogenising radiant correction, Be the gray average of the whole pixels of image, η is the relative radiometric calibration precision; Different radiance level data are calculated respectively, obtained the relative radiometric calibration precision curve under the different gray scales.

Twice behind camera relative radiometric calibration calibration repeatable accuracy is assessed as:

$B=\frac{L_A-L_B}{(L_A+L_B)/2}$

In the formula, B is twice radiation calibration result's of identical camera a repeatable accuracy, L _A, L _BBe respectively the gray-scale value mean value of all taps of twice calibration of camera under the same brightness.

Response consistance behind the camera relative radiometric calibration is assessed as:

$C=\frac{L_i-\stackrel{\‾}{L}}{\stackrel{\‾}{L}},(i=\mathrm{1,2,3}...).$

In the formula, C is the consistance of the different tap calibration of camera data,

Be the average gray of all taps under a certain brightness, L _iBe the average gray of a certain tap, i is the tap sequence number; Foundation

And C, brightness and the conforming change curve of different tap response can draw.

Dynamic range curve behind the camera relative radiometric calibration is assessed as: with different gray scales Be linked to be curve, i.e. the dynamic range curve.

The present invention's advantage compared with prior art is:

(1) the non-linear correction of device tap has been merged in the present invention on the basis of existing relative calibration method, solve device light intensity has been responded nonlinear radiation calibration and homogeneity correction problem, also solved the different problem of light intensity response nonlinearity between the different taps of CCD.

(2) the present invention has improved the relative radiometric calibration precision than prior art, especially in the non-linear more sensitive low light level of device zone obvious lifting is arranged, and has overcome radiance low shortcoming of relative calibration precision when low.

(3) the present invention has improved image homogeneity treatment effect by non-linear correction is carried out in tap, has promoted picture quality, helps the application and the interpretation of image.

(4) the present invention is by fitting response curve, overcome only at an even light intensity levels or two different optical radiation to proofread and correct shortcoming into the homogeneity coherent image respectively, can carry out the uniformity of radiation treatment for correcting at any gray scale.

(5) image after the present invention proofreaies and correct uniformity of radiation has carried out multiple parameter evaluation, with parametric line formal description remote sensor radiance, more science practicality.

Description of drawings

Fig. 1 is calibration raw data gray-scale map and each pixel response one dimension figure under a certain gray scale of the present invention, and wherein Fig. 1 a is the calibration raw data gray-scale map (two dimensional image) of TDICCD imaging, and Fig. 1 b is each the pixel response diagram behind the two dimensional image of one-dimensional;

Fig. 2 is the light intensity response synoptic diagram of following 24 the tap TDICCD of a certain gray scale;

Fig. 3 is a relative radiation correction method process flow diagram of the present invention;

Fig. 4 is a tap light intensity response analysis synoptic diagram of the present invention, and wherein Fig. 4 a is the calibration response curve and the straight line of each pixel under certain brightness, and Fig. 4 b is an example with tap 1, to the average scatter diagram of the light intensity of different gray scales response;

Fig. 5 fits and correction function deterministic process synoptic diagram for response curve of the present invention, and wherein Fig. 5 a is that the light intensity response of the tap 1 point straight line that looses fits figure, and Fig. 5 b is the non-linear correction function figure of tap 1;

Fig. 6 is the non-linear correction synoptic diagram of radiation calibration data of the present invention, and wherein Fig. 6 a is primary radiation calibration data one dimension figure, and Fig. 6 b is revised radiation calibration data one dimension figure;

Fig. 7 is the present invention and existing method radiant correction effect contrast figure, wherein Fig. 7 a proofreaies and correct preceding original one dimension image for certain calibration image radiation, Fig. 7 b is for adopting the revised one dimension image of conventional linear relative calibration method, and Fig. 7 c is the one dimension response curve behind the employing radiant correction of the present invention;

Fig. 8 is the present invention and existing method relative calibration precision comparison diagram.

Embodiment

The TDICCD of present embodiment has 24 taps, and each tap has 512 pixels, has 12288 pixels, supposes to have the image of camera under p different brightness (this example is 16 ranks), and image length is n.

According to the push-scanning image principle of TDICCD the radiation calibration data being carried out of one-dimensional handles, Fig. 1 is calibration raw data gray-scale map and each pixel response one dimension figure under a certain gray scale, wherein Fig. 1 a is the calibration raw data gray-scale map (two dimensional image) of TDICCD imaging, Fig. 1 b is each the pixel response diagram behind the two dimensional image of one-dimensional, and data of one-dimensional process is:

$X_{i_k}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{X}_{i,j_k},(i=\mathrm{1,2},...,12288.k=\mathrm{1,2},...p)---\left(1\right)$

X _iNominal gray scale column mean for i pixel under the brightness of k level.

With the pixel number is horizontal ordinate, is ordinate with the gray scale column mean of corresponding pixel, and the response curve of the radiation calibration data behind the of one-dimensional that draws is shown in Fig. 1 b.

Fig. 2 is the light intensity response synoptic diagram of following 24 the tap TDICCD of a certain gray scale.Dotted line indicates is the boundary line between tap, and horizontal ordinate is the pixel number among the figure, and ordinate is the gray scale column mean of corresponding pixel, i.e. { X ₁, X ₂... X ₅₁₂Be tap 1 image, { X ₅₁₃, X ₅₁₄... X ₁₀₂₄Be tap 2 images ... { X ₁₁₇₇₇, X ₁₁₇₇₈... X ₁₂₂₈₈Be tap 24 images.

Fig. 3 is a relative radiation correction method process flow diagram of the present invention.Mainly handling (seeing the latter half of process flow diagram) two large divisions by radiation calibration data processing (seeing the first half of process flow diagram) and radiant correction forms.

One, radiation calibration data processing: be divided into the non-linear correction of tap and revise the relative calibration coefficient and determine two parts.

(1) the non-linear correction of tap: be divided into the response analysis of tap light intensity again, response curve fits determines with correction function, three steps of the non-linear correction of radiation calibration data.

1. tap light intensity response analysis

Handle respectively for the image under p the different brightness, Fig. 4 a is calibration response curve under certain brightness and image average, and the horizontal line among the figure is the gradation of image mean value under this brightness.

At first obtain the average gray of all pixels of camera under each brightness, as shown in Equation 2:

$Y_k=\frac{1}{12288}\underset{i=1}{\overset{12288}{\mathrm{\Σ}}}{X}_{i_k},(k=\mathrm{1,2},...p)---\left(2\right)$

This Y _kBe the average response output of camera under the brightness of k level, 12288 is this routine pixel sum.

Next is to the average response Y of each tap _kAnalyze, for t tap, under the brightness of k level, its average response is exported as shown in Equation 3:

$S_{t_k}=\frac{1}{512}\underset{i=512\×t-511}{\overset{512\×t}{\mathrm{\Σ}}}{X}_{i_k},(t=\mathrm{1,2},...,24.k=\mathrm{1,2},...p)---\left(3\right)$

With { Y ₁, Y ₂... Y _pFor input, with { S _T1, S _T2... S _TpBe output, and just having obtained the light intensity response scatter diagram of t tap, Fig. 4 b is for tap 1 being the light intensity response scatter diagram of example.

2. response curve fits with correction function and determines

For t tap, with data { Y ₁, Y ₂... Y _pAnd { S _T1, S _T2... S _TpAdopt the method for least square to fit to straight line: S _t=c _tY+d _t, Fig. 5 a is that the diffusing point of the light intensity response straight line of tap 1 fits figure, the straight line shown in the figure is curve-fitting results one time.To { Y in the tap light intensity response scatter diagram ₁, Y ₂... Y _pAnd { S _T1, S _T2... S _TpThe corresponding point that forms carries out linear interpolation, obtain with (Y ₁, S _T1) be starting point, with (Y _p, S _Tp) be one section broken line of terminal point, broken line and St fitting a straight line is poor, promptly obtained the non-linear correction function M of tap t _t(S _x) curve, be depicted as the non-linear correction function curve of tap 1 as Fig. 5 b.Function M _t(S _t) physical significance be: for a bit (S on the non-linear correction function of tap t _t, M _t(S _t)), then the average gray value when this tap image is in S _tThe time, need be to the additional M of this tap _t(S _t) correction, the response of revising this tap of back just meets and is linear response in the dynamic range.

3. the non-linear correction of radiation calibration data

Be depicted as radiation calibration data one dimension image before revising as Fig. 6 a, for the part of the tap t in the image, non-linear makeover process as shown in Equation 4:

X _i′＝X _i+M _t(S _t)(i∈[512×t-511，512×t].t＝1，2，...24.) (4)

Promptly utilize formula 3 to try to achieve the average response output S of tap t earlier _t, utilize the non-linear correction function (S of tap t _t, M _x(S _x)) try to achieve the non-linear modified value M of this tap _t(S _t), then to each the some X in this tap _iAdditional M _t(S _t) correction, the X that obtains _i' be and eliminated the gray-scale value that i is ordered after the nonlinear response.Top operation is carried out in each tap in the image, can be obtained the correction radiation calibration data one dimension image shown in Fig. 6 b.

(2) revising the relative calibration coefficient determines

Revise the radiation calibration data according to 2 pairs of formula 1 and formula and calculate, obtain the nominal output X of each pixel under p the different brightness _i' export Y with the average response of camera _kWith data { X _I1', X _x' ... X _Ip' and { Y ₁, Y _p... Y _pAdopt the method for least square to fit to straight line: Y=a _i' X _i'+b _i', thereby obtain the correction relative calibration coefficient of each pixel of TDICCD.(a _i' and b _i' be the correction relative calibration coefficient of i pixel.)

So far finish the radiation calibration data processing, the data that this part obtains are divided into two parts, and a part is the non-linear correction function (S of tap _t, M _t(S _x)), another part is for revising relative calibration coefficient a _i' and b _i'.Two parts are shown in the parallelogram of Fig. 3 bottom.

Two, radiant correction is handled

After having obtained non-linear correction function and having revised the relative calibration coefficient, just can carry out the uniformity of radiation treatment for correcting to the image that this TDICCD produces.Processing procedure is divided into two steps:

1. at first image before the radiant correction is handled, utilized formula 3 to try to achieve the average response output S of tap t _t, utilize the non-linear correction function (S of tap t _t, M _t(S _t)), try to achieve the non-linear modified value M of this tap _t(s _t), then to each the some X in this tap _iAdditional magnitude is M _t(S _t) correction, the i pixel gray-scale value after the nonlinear response has been eliminated.

2. utilize to revise the relative calibration coefficient then image carried out radiant correction, trimming process as shown in Equation 5:

R _i，j＝a _i′·X _i，j+b _i′(i＝1，2，...12288.j＝1，2，...n) (5)

X in the formula _{I, j}Be pixel (i, pixel output gray level value j), R _{I, j}Be the gray-scale value behind the homogeneity correction.

Proofread and correct the result and be image behind the final radiant correction.

Image behind the homogeneity correction is carried out parameter evaluation, specifically comprises:

A, the relative radiometric calibration precision is estimated, computing method are: establish that the standard variance of image is Std behind the homogenising radiant correction, the gray average of whole pixels of image is

Image has m * n pixel, and the relative radiometric calibration precision is η:

$\mathrm{Std}=\sqrt{\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{[R_{i,j}-\stackrel{\‾}{R_{i,j}}]}^2},(i=\mathrm{1,2},...m,j=\mathrm{1,2}...n)$

$\stackrel{\‾}{R_{i,j}}=\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{\text{i=1}}{\overset{m}{\mathrm{\Σ}}}{R}_{i,j}$

$\mathrm{\η}=\frac{\mathrm{Std}}{\stackrel{\‾}{R_{i,j}}}$

Different spoke intensity level data are calculated respectively, obtained the relative radiometric calibration precision curve under the different gray scales.

B, to twice the calibration repeatable accuracy estimate, computing method are: establish the repeatable accuracy that B is twice radiation calibration result of identical camera, L _A, L _BBe respectively the gray average of twice calibration of camera under the same brightness, then have

$B=\frac{L_A-L_B}{(L_A-L_B)/2}$

C, the response consistance is estimated, computing method are: the consistance of establishing the different taps calibration of camera data is C,

Be the average gray of all taps under a certain brightness, L _iBe the average gray of a certain tap, i is the tap sequence number, then has

$C=\frac{L_i-\stackrel{\‾}{L}}{\stackrel{\‾}{L}},(i=\mathrm{1,2,3}...)$

Foundation

And C, brightness and the conforming change curve of different tap response can draw.

D, the dynamic range curve is estimated, method is: with different gray scales

Be linked to be curve, i.e. the dynamic range curve.

Fig. 7 is the present invention and existing method radiant correction effect contrast figure, wherein Fig. 7 a proofreaies and correct preceding original one dimension image for certain calibration image radiation, Fig. 7 b is for adopting the revised one dimension image of conventional linear relative calibration method, Fig. 7 c is the of one-dimensional curve behind the employing radiant correction of the present invention, the response of the unevenness of each pixel response of image and different taps rises and falls and significantly improves as can be seen, and promptly the nonuniformity correction effect gets a promotion.

Fig. 8 is the present invention and existing method relative calibration precision comparison diagram.Average gray value with image among the figure is a horizontal ordinate, is ordinate (unit is %) with the relative radiometric calibration precision.The curve of below is a relative calibration precision of the present invention among the figure, the curve of top is the relative calibration precision of existing radiation calibration method, can find out obviously that this method is significantly improved than existing methods on scheming, the relative radiometric calibration precision in the dynamic range has improved 15%～40% with respect to former method by statistics.Especially it is bigger to improve degree in the lower zone of brightness.

Part not in the detailed description of the invention belongs to present technique field known technology.

Claims (10)

1. the relative radiation correction method of a star-load TDICCD camera is characterized in that step is as follows:

(1) tap light intensity response analysis: analyze the response output situation of tap under different light intensity, try to achieve tap light intensity response scatter diagram;

(2) response curve fits with non-linear correction function and determines: the tap light intensity response scatter diagram that step (1) is obtained carries out straight line and fits, and the point to the corresponding formation with ordinate of tap light intensity response scatter diagram horizontal ordinate carries out linear interpolation again, obtains interpolation curve; Do the interpolation curve of each tap and the straight line that fits separately poor, obtain the non-linear correction function of tap to light intensity, i.e. tap is to the responsive corrections value curve of different brightness;

(3) the non-linear correction of radiation calibration data: the non-linear correction function that uses step (2) to obtain carries out non-linear correction one by one to the tap of radiation calibration data, obtains being used for determining to revise the non-linear revised radiation calibration data of relative calibration coefficient;

(4) revising the relative calibration coefficient determines: carry out the relative radiometric calibration data processing to carrying out non-linear revised calibration data, obtain revising the relative calibration coefficient;

(5) the original heterogeneity image that TDICCD is produced carries out radiant correction and handles: the correction relative calibration coefficient that utilizes non-linear correction function that step (2) obtains and step (4) to obtain, the image of any brightness degree that TDICCD is produced carries out uniformity of radiation and proofreaies and correct.

2. the relative radiation correction method of star-load TDICCD camera according to claim 1, it is characterized in that: the original heterogeneity image that TDICCD is produced carries out after radiant correction handles, also can carry out quantitative evaluation to the radiation calibration result: the corresponding down homogeneity correction image of different gray scales that utilizes step (5) to obtain, the response consistance behind the camera relative radiometric calibration, relative radiometric calibration precision, twice calibration repeatable accuracy, dynamic range curve are evaluated.

3. the relative radiation correction method of star-load TDICCD camera according to claim 1, it is characterized in that: the tap light intensity response analysis of described step (1) is: the output image of supposing to have p different brightness degrees, p the image averaging gray-scale value that obtains after the gradation of image value of p different brightness is averaged respectively responds the horizontal ordinate of scatter diagram as light intensity, is ordinate with each tap at the average image gray-scale value of each brightness, obtains the light intensity response scatter diagram of whole taps.

4. the relative radiation correction method of star-load TDICCD camera according to claim 1, it is characterized in that: non-linear being modified to of radiation calibration data of described step (3): the non-linear correction function that is obtained by step (2) is tried to achieve the non-linear modified value of tap correspondence, then to each the pixel additional corrections value in the tap, after each tap revised one by one, finally obtain being used for determining revising radiation calibration data image after the correction of relative calibration coefficient.

5. the relative radiation correction method of star-load TDICCD camera according to claim 1, it is characterized in that: the correction relative calibration coefficient of described step (4) is defined as: the correction radiation calibration data that step (3) is obtained according to

$X_{i_k}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{X_{i,j}}_k,(i=\mathrm{1,2},...,m.k=\mathrm{1,2},...p)$

Calculate the nominal output gray level value X of each pixel under p the different brightness _iX in the formula _IkBe the nominal gray scale column mean of i pixel under the brightness of k level, k=1,2 ... p, p are the high-high brightness grade; X _i, j _kBe pixel under the brightness of k level (i, output numerical value j); M is the pixel columns picture traverse of TDICCD just; N is the pixel line number, just the image length of TDICCD; Again according to

$Y_k=\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{X}_{i_k},(k=\mathrm{1,2},...p)$

Calculate the average output Y of m pixel under p the different brightness respectively _kY in the formula _kAverage output for m pixel under the brightness of k level; M pixel carried out straight line with the method for least square one by one fit, obtain the correction relative calibration coefficient a of each pixel of TDICCD _i', b _i'.

6. the relative radiation correction method of star-load TDICCD camera according to claim 1, it is characterized in that: the image that in the described step (5) TDICCD is produced carries out radiant correction and is treated to: the non-linear correction function that utilizes step (2) to obtain, non-linear correction is carried out in each tap, the correction relative calibration coefficient that utilizes step (4) to obtain then, according to

R _i，j＝a _i′·X _i，j+b _i′(i＝1，2，...m.j＝1，2，...n)

The image of any brightness degree that TDICCD is produced carries out radiant correction; A in the formula _i', b _i' be the correction relative calibration coefficient of i pixel, X _{I, j}Be pixel (i, output numerical value j), R _{I, j}For proofreading and correct result value.

7. the relative radiation correction method of star-load TDICCD camera according to claim 2 is characterized in that: the relative radiometric calibration precision behind the camera relative radiometric calibration is assessed as:

$\mathrm{\η}=\frac{\mathrm{Std}}{\stackrel{\‾}{R_{i,j}}}$

Wherein

$\mathrm{Std}=\sqrt{\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{[R_{i,j}-\stackrel{\‾}{R_{i,j}}]}^2},(i=\mathrm{1,2},...m,j=\mathrm{1,2}...n)$

$\stackrel{\‾}{R_{i,j}}=\frac{1}{\mathrm{mn}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{R}_{i,j}$

In the formula, image has m * n pixel, and Std is the standard variance of image behind the homogenising radiant correction, Be the gray average of the whole pixels of image, η is the relative radiometric calibration precision; Different radiance level data are calculated respectively, obtained the relative radiometric calibration precision curve under the different gray scales.

8. the relative radiation correction method of star-load TDICCD camera according to claim 2 is characterized in that: twice behind camera relative radiometric calibration calibration repeatable accuracy is assessed as:

$B=\frac{L_A-L_B}{(L_A+L_B)/2}$

In the formula, B is twice radiation calibration result's of identical camera a repeatable accuracy, L _A, L _BBe respectively the average gray of all taps of twice calibration of camera under the same brightness.

9. the relative radiation correction method of star-load TDICCD camera according to claim 2 is characterized in that: the response consistance behind the camera relative radiometric calibration is assessed as:

$C=\frac{L_i-\stackrel{\‾}{L}}{\stackrel{\‾}{L}},(i=\mathrm{1,2,3}...).$

In the formula, C is the consistance of the different tap calibration of camera data,

Be the average gray of all taps under a certain brightness, L _iBe the average gray of a certain tap, i is the tap sequence number; Foundation

And C, brightness and the conforming change curve of different tap response can draw.

10. the relative radiation correction method of star-load TDICCD camera according to claim 2 is characterized in that: the dynamic range curve behind the camera relative radiometric calibration is assessed as: with different gray scales

Be linked to be curve, i.e. the dynamic range curve.

Images