CN103175543B - Internal response difference correction method for charge coupled device (CCD) pixel - Google Patents

Abstract

The invention discloses an internal response difference correction method for a charge coupled device (CCD) pixel. The correction is realized by utilizing strong correlation between the measurement flow error of an observational star brought by the internal response difference in the CCD pixel and the normalized centric coordinates of the observational star. The two-dimensional correction map of the CCD target surface grouping is established through the actual timing sequence photometric observations, the measured data can find the corresponding two-dimensional correction map according to the observed star image central coordinate and the normalized coordinate, and the corresponding flow correction amount is obtained from the map; and finally, the correction of the photometric error brought by the internal response difference of the CCD pixel is realized, and the photometric precision is improved.

Description

The antidote of the inner response difference of a kind of CCD pixel

Technical field

The present invention relates to the antidote of the inner response difference of a kind of CCD pixel, be applicable to the chronometer data process needing the light-metering flow error carrying out bringing because of the inner response difference of CCD pixel to correct in astronomical sight.Be specially adapted to wide visual field, low sampling rate and the front scope according to type CCD, the light-metering error because the inner response difference of CCD pixel brings can be reduced significantly.

Background technology

Ccd image sensor (Charge Coupled Device, be called for short CCD), it uses a kind of semiconductor material of ISO to make, light signal can be transformed into charge signal, convert digital signal to by analog to digital converter chip, digital signal is preserved by the flash memory of camera internal or built-in hard disk card after overcompression, thus easily data can be transferred to computing machine, and by means of the process means of computing machine, process image as required.

In Modern Astronomical observation, CCD is as application image acquisition terminal widely.Its major constituents and principle are: CCD is the face battle array rearranged according to certain rules by many photosensitive pixels.Each pixel in the battle array of face is exactly a MOS capacitor (being mostly photodiode), when having in 1 bundle ray cast to MOS capacitor, photon, through transparency electrode and oxide layer, enters MOS capacitor substrate, last photon is converted to electronics, exports with electric signal.The efficiency that photon is converted to electronics is called QE, due to the restriction of manufacturing process, photoelectric transformation efficiency QE for different pixel has this species diversity of small difference and is called as response difference between pixel, in the data handling procedure of Modern Astronomical, adopt the method for " flat field " to correct, namely utilize the equal light of luminosity to beat on CCD front, just can draw correct amount from the output charge difference of last each pixel.Meanwhile, due to the design feature of mos capacitance, CCD also also exists the response difference of pixel inside, if the light that is, a branch of yardstick being less than CCD pixel beats the diverse location at a pixel, the electron number that the difference due to photoelectric transformation efficiency obtains is different.Because CCD grid cell size is little, and the PSF (point spread function) of general astronomical sight star image all can cover the pixel of a greater number, and namely sampling is good.Like this because the light-metering error that the inner response difference of CCD brings generally can not consider process much smaller than other error.But for the scope as Hubble Telescope, because PSF is very little, sampling can seem relative deficiency, and thus can consider the rectification of inner response difference, the method for employing is " shake " method." shake " method requires very high pointing accuracy, and needs all to carry out repeatedly " shake " observation to same visual field at every turn.

The development of modern astronomy has occurred that the astronomy of some small-bore Large visual angle is toured the heavens project, its visual field can reach 14 degree × 14 degree, the sampling rate of such CCD can reduce greatly, and the light-metering error that the inner response difference of CCD brings can become very remarkable, must carry out the correction in data processing.To telescope, " shake " method refers to that accuracy requirement is high, small-bore telescope is difficult to reach, and simultaneously due to the restriction of observation mode, " shake " method is not suitable for the inner response difference of the telescopical CCD of this type and corrects.

Summary of the invention

Technology of the present invention is dealt with problems and is: the response difference of CCD pixel inside can bring in astronomical sight the measuring error observing starlight degree flow, this error for front according to more remarkable under the observation condition of type CCD and low sampling rate.The subject matter that the present invention solves is corrects this photometric measurement error brought because of the inner response difference of CCD.

Technical solution of the present invention is: the antidote of the inner response difference of a kind of CCD pixel, and step is as follows:

(1) according to the measuring flow error of observation star and the correlativity of the centre coordinate observed after star normalization, the inner response difference two dimension of structure pixel is corrected and is schemed;

(2) flow and the center of observed object star is measured, the center of normalization observed object star;

(3) correct figure according to the center after the normalization of observed object star by the inner response difference two dimension of interpolate picture elements, calculate flow correct amount;

(4) according to flow correct amount, error correction is carried out to observed object star measuring flow.

Described structure pixel inner response difference two dimension rectification figure step is as follows:

(1.1) continuous print sequential Photometric observation is carried out to the fixing sky district selected, obtain one group of seasonal effect in time series image;

(1.2) to observing all images obtained carry out the removal of CCD background, flat field rectification and dark current Transformatin in step (1.1);

(1.3) aperture photometry is carried out to all stars in the every width image j after step (1.2) process, obtain the flow flux of every observation star k _kjwith signal to noise ratio (S/N ratio) (S/N) _kj; Measure the center (xc of every observation star simultaneously _kj, yc _kj), filter out all observation stars that signal to noise ratio (S/N ratio) is greater than 50;

(1.4) according to the flow flux of the observation star after step (1.3) screening _kjwith center xc _kj, yc _kj, obtain the flow fluxnorm after the normalization of every observation star in whole light-metering time series _kj, and then obtain this group light-metering seasonal effect in time series flow intermediate value fluxMedian _k, according to the ratio calculation outflow ratio R atioFlux of the two _kj; In addition, normalization center position coordinates (xcPixIn has also been needed _kj, ycPixIn _kj) process, computing formula is

$\left\{\begin{array}{c}{\mathrm{xcPixIn}}_{\mathrm{kj}}={\mathrm{xc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{xc}}_{\mathrm{kj}}\right)\\ {\mathrm{ycPixIn}}_{\mathrm{kj}}={\mathrm{yc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{yc}}_{\mathrm{kj}}\right)\end{array}\right.;$

(1.5) actual CCD target surface is divided into groups, according to the center (xc of observation star _kj, yc _kj), obtain the respective packets of every observation star on actual CCD target surface;

(1.6) respectively following process is carried out to each group in step (1.5):

First, 1 pixel is at least divided into the grid of 10 × 10, according to the normalization center position coordinates value (xcPixIn that step (1.4) obtains _kj, ycPixIn _kj), show that normalization center position coordinates value falls into the observation star of each sub-grid respective image;

Then, according to the observation star of respective image in the above-mentioned each sub-grid determined, the flow rate ratio falling into all observation stars of this sub-grid is determined; And calculate its intermediate value median (RatioFlux), represent the value of place sub-grid by this intermediate value, response difference two dimension rectification figure in the pixel namely obtaining above-mentioned grid division.

Interpolation in described step (3) adopts cubic spline function interpolation to improve the precision of the flow correct amount calculated.

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

(1) treatment scheme operation of the present invention is more direct and convenient with realization, easily realizes.

The present invention has avoided real room and has measured the complicacy of CCD pixel portion response difference and require high difficulty to measuring equipment.Meanwhile, concrete higher requirement be there is no yet for pointing accuracy.From observation mode, as long as by selecting good weather to carry out sequential light-metering, the two dimension rectification figure of the inner response difference of construction CCD pixel can realize.

(2) the two dimension rectification figure divided into groups by CCD target surface of the present invention realizes correcting, respond well.

Centre coordinate after the present invention studies measuring flow (because the inner response difference of the CCD pixel brings) error that finds observation star and the normalization of observation star also exists stronger correlativity, correct figure by the two dimension building the inner response difference of CCD pixel, realize correcting.Meanwhile, the grouping of logical CCD target surface is eliminated because the inconsistency of the distribution of PSF on CCD target surface.Shown by examples of implementation, the invention process is respond well.

In a word, this two dimension rectification figure by CCD target surface piecemeal of the present invention realizes observing in astronomical sight the error correction of the measuring flow of star (because the inner response difference of pixel brings), can facilitate, the raising of the light-metering precision effectively realized in astronomical sight.

Accompanying drawing explanation

Fig. 1 is pixel of the present invention inner response difference antidote implementing procedure figure;

Fig. 2 is that sequential light-metering light curve of the present invention measures implementing procedure figure;

Fig. 3 process flow diagram that to be observed image of the present invention carry out with the star of reference picture mates;

Fig. 4 be embodiments of the invention pixel in response difference rectification figure;

Fig. 5 is that the typical case of embodiments of the invention corrects light curve figure;

Fig. 6 be embodiments of the invention rectification before and after variance comparison diagram.

Embodiment

The present invention is found by research: the centre coordinate after the measuring flow of observation star (because the inner response difference of CCD pixel brings) error and the normalization of observation star also exists stronger correlativity.The antidote of the inner response difference of CCD pixel of the present invention utilizes this correlativity to carry out correction process just.First, pixel inner response difference two dimension rectification figure is built by studying this correlativity; Then, measure flow and the center of observed object star, normalization target satellite center, and correct figure according to the center after the normalization of observed object star by the inner response difference two dimension of cubic spline function interpolate picture elements, calculate flow correct amount; Finally, the error correction of observed object star measuring flow (because the inner response difference of pixel brings) is realized.As shown in Figure 1, concrete implementation step is as follows:

(1) the inner response difference two dimension of CCD pixel corrects the structure of figure;

The main observation and the processing procedure that build CCD inner response difference two dimension rectification figure are:

1. need to select sky district that weather condition is good and atmospheric transparency is high light-metering observes stellar field dense degree moderate yesterday to carry out sequential Photometric observation, obtain one group of seasonal effect in time series image; The sequential light-metering sampled point of General Requirements Continuous Observation be about 500 or more than, weather condition is good.

2. observe the raw data obtained need the base conditioning of ccd image, processing procedure comprises: the removal (BIAS) of CCD background, the rectification of flat field and dark current Transformatin.These processing procedures use astronomical conventional software I RAF or general conventional astronomical image data processing software bag to realize.

3. aperture photometry is carried out to all stars in above-mentioned every width image j, obtain the flow flux of every observation star k _kjwith signal to noise ratio (S/N ratio) (S/N) _kj(such as can be obtained by IRAF software).Utilize gravity model appoach to measure the center (xc of every observation star simultaneously _kj, yc _kj), its computing formula is:

$\left\{\begin{array}{c}{\mathrm{xc}}_{\mathrm{kj}}=\frac{\mathrm{\Σ}{x}_i{s}_i}{\mathrm{\Σ}{s}_i}\\ {\mathrm{yc}}_{\mathrm{kj}}=\frac{\mathrm{\Σ}{y}_c{s}_i}{\mathrm{\Σ}{s}_i}\end{array}\right.---\left(1\right)$

Wherein, x _iand y _ifor x and the y coordinate figure of observation star i-th pixel.S _iit is the response of i-th pixel starlight.From the file header (Fits form) of image, read observation time count obstime _j.Note: subscript k represents a kth observation star in visual field, and subscript j represents jth width image.Getting k and j to represent for stating convenience, making a general reference certain star and certain figure.Finally, according to the signal to noise ratio (S/N ratio) measured, filter out the calculating that all observation stars that signal to noise ratio (S/N ratio) is greater than 50 carry out following steps.

Flow (the flux of every the observation star 4. 3. filtered out according to above step _kj), signal to noise ratio (S/N ratio) (S/N) _kj, center (xc _kj, yc _kj) and observation time obstime _j, the light curve completing every observation star measures the detailed process needed as shown in Figure 2.By the light curve recorded, the flow fluxnorm after the normalization of every observation star in light-metering time series can be drawn _kj, calculate its intermediate value and be designated as: fluxMedian _k.Then observe the flow rate ratio of star for being designated as kth in jth width figure: RatioFlux _kj, calculating formula is: in addition, also need the normalized of center, normalize to a pixel intrinsic coordinates (note: make use of the characteristic that the inner response difference of each pixel of CCD is substantially identical here) by all centers.Specific formula for calculation is,

$\left\{\begin{array}{c}{\mathrm{xcPicIn}}_{\mathrm{kj}}={\mathrm{xc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{xc}}_{\mathrm{kj}}\right)\\ {\mathrm{ycPixIn}}_{\mathrm{kj}}={\mathrm{yc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{yc}}_{\mathrm{kj}}\right)\end{array}\right.---\left(2\right)$

Finally, according to centre coordinate position, every observation star is divided into groups, grouping is main according to the distribution situation of PSF on CCD target surface, namely the otherness for the full width at half maximum of the point spread function (PSF) of bright star (signal to noise ratio (S/N ratio) is greater than 50) in grouping same on CCD target surface is less than 10%, generally adopts 4 × 4 totally 16 groups.

5. complete the two dimension rectification figure (it is identical that each group completes treatment step) of response difference in pixel, wherein one group of processing procedure is: the grid 1 pixel being divided into 10 × 10, and each sub-grid represents 0.1 picture element unit.According to above measurement result, observation stars all on all light-metering time-series images is had to the RatioFlux of measurement _kjwith the centre coordinate (xcPixIn after the normalization of correspondence _kj, ycPixIn _kj), divide into groups by each unit sub-grid coordinate figure according to the centre coordinate after normalization, the one group of flow rate ratio falling into each sub-grid can be obtained.Intermediate value is got to this group flow rate ratio, i.e. median (RatioFlux), represent the value of sublattice, median (RatioFlux) value (note: if corresponding sublattice does not have corresponding star, then the value of getting all adjacent lattice is averaged the value representing this sublattice afterwards) of 10 × 10 sublattices then can be obtained by above calculating.So just obtain with the flow rate ratio figure of 0.1 pixel, 10 × 10 grid that are sublattice in 1 pixel, namely in pixel, the two dimension of response difference corrects figure.

(2) flow flux and center (xc, the yc) of observed object star is measured, the center (xcPixIn, ycPixIn) of normalization observed object star; Concrete account form is with above-mentioned observation star.

(3) interpolation is passed through (in order to improve the precision of calculated flow rate correct amount according to the center after the normalization of observed object star, cubic spline function interpolation can be adopted) pixel inner response difference two dimension rectification figure, obtain flow correct amount Ratioflux.

(4) according to flow correct amount, error correction is carried out to observed object star measuring flow, obtain the flow (flux after correcting _corrected) be:

As shown in Figure 2, be light curve measuring process of the present invention.Concrete steps are: 1. first find out the good width of picture quality and be designated as reference picture.The air quality of image is main basis for estimation.2. the star of observed image is mated according to position with the star of reference picture, namely find out same corresponding star accordingly.Record flow flux _{k, ref}, center (xc _{k, ref}, yc _{k, ref}) and observation time obstime _{k, ref}.Note: subscript k is a kth star, and following table ref represents reference picture.The step that concrete matching process is shown in Figure 3.3. the flow rate ratio between the corresponding star completing coupling calculates, and formula is: (flux _{k, ref}: kth star flow on a reference; Flux _k,jfor the flow of a kth star on jth width image).Then the flow rate ratio intermediate value between all corresponding star between observed image with reference picture can be calculated, namely (k=1...N coupling star), finally can show that the flow of normalization observation star is to reference stellar field, as a kth star, is calculated as: 4. by the process of above step, the light-metering flow that can obtain about every observation star normalizes to reference to the light-metering sequence after stellar field.

As shown in Figure 3, for observed image of the present invention and the star of reference picture carry out the process flow diagram that mates.1. for jth width figure, matching process is: from jth width figure and reference picture, respectively select the coupling that 500 the brightest stars carry out star catalogue respectively.The algorithm of coupling adopts widely used triangle map algorithm in astronomy, namely utilizes the triangle of the star structure in two stellar fields, and then carries out the coupling of similar triangles, finally find out corresponding coupling star.Output after Matching of star atlas completes is one and mates the list of broadcasting TV programs by satellite between two: the 1st, and 2 are classified as (xc _kj, yc _kj) the 3rd, 4 are classified as (xc _{k, ref}, yc _{k, ref}).2. 4 rank 2-d polynomials are utilized to carry out matching (xc _{k, ref}, yc _{k, ref}) and (xc _kj, yc _kj) between relation.Fitting formula can be written as:

$\left\{\begin{array}{c}{\mathrm{xc}}_{k,\mathrm{ref}}=a_1+a_2{\mathrm{xc}}_{\mathrm{kj}}+a_3{\mathrm{xc}}_{\mathrm{kj}}^2+a_4{\mathrm{xc}}_{\mathrm{kj}}^3+a_5{\mathrm{yc}}_{\mathrm{kj}}+a_6{\mathrm{yc}}_{\mathrm{kj}}^2+a_7{\mathrm{yc}}_{\mathrm{kj}}^3\\ +a_8{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}+a_9{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}+a_{10}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}\\ +a_{11}{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}^2+a_{12}{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}^2+a_{13}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}^2\\ +a_{14}{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}^3+a_{15}{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}^3+a_{16}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}^3\\ {\mathrm{yc}}_{,\mathrm{ref}}=b_1+b_2{\mathrm{xc}}_{\mathrm{kj}}+b_3{\mathrm{xc}}_{\mathrm{kj}}^2+g_4{\mathrm{xc}}_{\mathrm{kj}}^3+b_5{\mathrm{yc}}_{\mathrm{kj}}+b_6{\mathrm{yc}}_{\mathrm{kj}}^2+b_7{\mathrm{yc}}_{\mathrm{kj}}^3\\ +b_8{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}+b_9{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}+b_{10}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}\\ +b_{11}{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}^2+b_{12}{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}^2+b_{13}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}^2\\ +b_{14}{\mathrm{xc}}_{\mathrm{kj}}{\mathrm{yc}}_{\mathrm{kj}}^3+b_{15}{\mathrm{xc}}_{\mathrm{kj}}^2{\mathrm{yc}}_{\mathrm{kj}}^3+b_{16}{\mathrm{xc}}_{\mathrm{kj}}^3{\mathrm{yc}}_{\mathrm{kj}}^3\end{array}\right.---\left(3\right)$

Wherein, a _i(i=1 ... 16) and b _i(i=1 ... 16) be fitting parameter, these 32 fitting parameters can draw by the matching of the star matched between two in above 500 ludcidas.Then utilize (3) formula (relational expression that this matching obtains) by the coordinate (xc of all observation stars in jth width figure _kj, yc _kj) be converted into coordinate (xc' under reference picture coordinate system _kj, yc' _kj).3. the distance between two between star of calculating observation star and all reference picture culminant stars under reference picture coordinate system, distance (d) computing formula is distance computing formula, i.e. distance under rectangular coordinate system (wherein (xc _ref, yc _ref) represent the coordinate of all stars in reference picture).A reference star the shortest with observing star distance can be selected according to the calculating of above distance (d), and the distance between them is also eligible, namely (error radius generally gets 1.5 pixels to be less than error radius, according to equipment and observation condition, this parameter can make the appropriate adjustments), then think that this reference star is mate star between two with observation star, can remember that the coordinate of this reference star is (xc _{k, ref}, yc _{k, ref}).Note: the processing procedure every width image all being completed to above jth width figure, can realize mating of star in all observed images and reference picture culminant star.

Embodiment

(1) according to type 3056 × 3056 pixels before scope: the bore of telescope lens is 7 centimetres, and CCD model is: ALTAU9000, each pixel corresponds to 12.6 rads.The full width at half maximum (FWHM) of the point spread function (PSF) of star image is 1.2 pixels (corresponding 1.5 rads).

(2) choose the observation that the good observation of weather carries out building pixel inner response difference two dimension rectification figure night, concrete data acquisition and processing (DAP) requirement and process are undertaken by the requirement described in " embodiment " of the present invention.Finally obtain 4 × 4 the inner response difference two dimension of totally 16 groups of pixels correct figure as shown in Figure 4:

(3) choose the Photometric observation in other moment, according to treatment step of the present invention, utilize above rectification figure to correct, obtain typical rectification effect as shown in Figure 5,1. light curve is the data after rectification; 2. be correct amount; 3. be raw data.The figure shows the present invention and can eliminate light-metering style (error) because the inner response difference of CCD pixel brings preferably.Fore-and-aft survey variance (error) Statistical Comparison improving process is carried out to a collection of observation star, as shown in Figure 6.In upper figure a in Fig. 6, dotted line represents the variance before correction process and magnitude relation, and solid line represents the relation of the variance after correction process and magnitude.B represents the reduction of variance and the graph of a relation of magnitude.This figure shows the star for high s/n ratio, and measuring error can reduce about 50%.

The unspecified part of the present invention belongs to general knowledge as well known to those skilled in the art.

Claims (3)

1. an antidote for the inner response difference of CCD pixel, is characterized in that step is as follows:

(1) according to the measuring flow error of observation star and the correlativity of the centre coordinate observed after star normalization, the inner response difference two dimension of structure pixel is corrected and is schemed;

(2) flow and the center of observed object star is measured, the center of normalization observed object star;

(3) correct figure according to the center after the normalization of observed object star by the inner response difference two dimension of interpolate picture elements, calculate flow correct amount;

(4) according to flow correct amount, error correction is carried out to observed object star measuring flow.

2. the antidote of the inner response difference of a kind of CCD pixel according to claim 1, is characterized in that: described structure pixel inner response difference two dimension rectification figure step is as follows:

(1.1) continuous print sequential Photometric observation is carried out to the fixing sky district selected, obtain one group of seasonal effect in time series image;

(1.2) to observing all images obtained carry out the removal of CCD background, flat field rectification and dark current Transformatin in step (1.1);

(1.3) aperture photometry is carried out to all stars in the every width image j after step (1.2) process, obtain the flow flux of every observation star k _kjwith signal to noise ratio (S/N ratio) (S/N) _kj; Measure the center (xc of every observation star simultaneously _kj, yc _kj), filter out all observation stars that signal to noise ratio (S/N ratio) is greater than 50;

(1.4) according to the flow flux of the observation star after step (1.3) screening _kjwith center xc _kj, yck _j, obtain the flow fluxnorm after the normalization of every observation star in whole light-metering time series _kj, and then obtain this group light-metering seasonal effect in time series flow intermediate value fluxMedian _k, according to the ratio calculation outflow ratio R atioFlux of the two _kj; In addition, normalization center position coordinates (xcPixIn has also been needed _kj, ycPixIn _kj) process, computing formula is

$\left\{\begin{array}{c}{\mathrm{xcPixIn}}_{\mathrm{kj}}={\mathrm{xc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{xc}}_{\mathrm{kj}}\right)\\ {\mathrm{ycPixIn}}_{\mathrm{kj}}={\mathrm{yc}}_{\mathrm{kj}}-\mathrm{int}\left({\mathrm{yc}}_{\mathrm{kj}}\right)\end{array}\right.;$

(1.5) actual CCD target surface is divided into groups, according to the center (xc of observation star _kj, yc _kj), obtain the respective packets of every observation star on actual CCD target surface;

(1.6) respectively following process is carried out to each group in step (1.5):

First, 1 pixel is at least divided into the grid of 10 × 10, according to the normalization center position coordinates value (xcPixIn that step (1.4) obtains _kj, ycPixIn _kj), show that normalization center position coordinates value falls into the observation star of each sub-grid respective image;

Then, according to the observation star of respective image in the above-mentioned each sub-grid determined, the flow rate ratio falling into all observation stars of this sub-grid is determined; And calculate its intermediate value median (RatioFlux), represent the value of place sub-grid by this intermediate value, response difference two dimension rectification figure in the pixel namely obtaining above-mentioned grid division.

3. the antidote of the inner response difference of a kind of CCD pixel according to claim 1, is characterized in that: the interpolation in described step (3) adopts cubic spline function interpolation to improve the precision of the flow correct amount calculated.