CN102054487B - Pixel mismatch compensation method for volume hologram storage system - Google Patents

Abstract

The invention relates to a pixel mismatch compensation method for a volume hologram storage system, which is characterized in that a pixel mismatch compensation module based on complex amplitudes is adopted; the number of adjacent pixels is three; and the strength of a detection signal of an image sensor pixel is the integration of the square of the summation of the complex amplitudes generated by the pixel of a spatial light modulator, therefore, when the width of a pixel spread function exceeds one pixel and is less than or equal to two pixels, the pixel mismatch compensation module can be acquired. Through the method provided by the invention, the signal to noise ratio of a data page read by the volume hologram storage system can be efficiently increased, the positive or negative of pixel mismatch quantity is difficult to sense, and the effect of equalization treatment is achieved. The method provided by the invention can be widely applied to the volume hologram storage system in which the width of the pixel spread function is not more than two times of a pixel interval and the partial mismatch quantity of the spatial light modulator relative to the image sensor pixel can be an arbitrary value.

Description

A kind of body hologram memory system pixel mismatch compensation method

Technical field

The present invention relates to the compensation method in a kind of body holographic data storage technical field, particularly about a kind of body hologram memory system pixel mismatch compensation method.

Background technology

In the body holographic data storage system, receive the influence of factors such as optical distortion, material contracts, system vibration, process and assemble error, the pixel mismatch between the input and output device is difficult to avoid, thereby causes the signal to noise ratio (S/N ratio) of read output signal to descend.Especially, can seriously weaken signal to noise ratio (S/N ratio) near the mismatch of half-pixel, when image sensor with respect to spatial light modulator (x direction) translation in the horizontal direction during 0.4 pixel, the signal to noise ratio (S/N ratio) that receives image is reduced to 1.5 (as shown in Figure 1) by 4.2.

In order to solve the problem of pixel mismatch, linear compensation based on light intensity, over-sampling reduction are arranged and in the prior art based on these three types of compensation methodes of second compensation of complex amplitude.The linear compensation method algorithm is simple, carries out fast, but not accurate enough based on the model of light intensity, compensation effect is general; The over-sampling method of reducing can be gone back original image more exactly, and insensitive to skew, out of focus, rotation, but can sacrifice memory capacity exponentially; Second compensation method based on complex amplitude can compensate the random mismatch page under the situation of known local mismatches amount; The method is proposed by the G.W.Burr of IBM Corporation the earliest; Therefore the second compensation method based on complex amplitude abbreviates " IBM compensation method " as, and its cardinal principle is following:

As shown in Figure 2, three pixel segment of image sensor on the one dimension direction, its signal strength detection is respectively r ₀, r ₁, r ₂, (light intensity is respectively p to three corresponding with it slm pixel ₀, p ₁, p ₂) producing micro-translation σ with respect to image sensor, the fill factor, curve factor of slm pixel is g ₁, the spectrum face square aperture diaphragm length of side is D.The COMPLEX AMPLITUDE that each slm pixel produces on the image sensor face is:

$h\left(x\right)=c_h{\&Integral;}_{-{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}^{{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}\sin c\left[\frac{D}{D_N}(x-x^{\&prime;})\right]{\mathrm{dx}}^{\&prime;},---\left(1\right)$

Sinc (x) ≡ sin (π x)/(π x) wherein, c _hFor satisfying

Constant, x, x ' all are unit with the pixel separation, D _N=λ f/ Γ is Nyquist (Nyquist) aperture, and λ is an optical maser wavelength, and f is the focal length of fourier transform lens, and Γ is a pixel separation.

As shown in Figure 3, among the figure fill factor, curve factor g of slm pixel ₁Be 1, spectrum face square aperture diaphragm length of side D is 1.5D _NThe time the pixel spread function.Be different from point spread function, h (x) be defined as the pixel spread function of system here, its width is defined as the distance between two zero points of main lobe.If pixel spread function width approximates pixel separation, pixel r on the image sensor so ₂The field intensity value be:

${\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">r</figure-callout>}}_2={\&Integral;}_{-{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}^{{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}{[\sqrt{p_2}h(x-{\mathrm{\&sigma;}}_2)+\sqrt{p_1}h(x-{\mathrm{\&sigma;}}_1)]}^2\mathrm{dx},---\left(2\right)$

G wherein ₂Fill factor, curve factor for the image sensor pixel.Following formula can be put in order and be:

${\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">r</figure-callout>}}_2=p_2{H}_{22}\left(\mathrm{\&sigma;}\right)+2\sqrt{p_1{p}_2}{H}_{12}\left(\mathrm{\&sigma;}\right)+p_1{H}_{11}\left(\mathrm{\&sigma;}\right),---\left(3\right)$

H wherein ₁₁(σ), H ₂₂(σ), H ₁₂(σ) represent that respectively their numerical value can calculate in advance or experiment test obtains by respective pixel, neighbor, normalized signal intensity that both interfere stack to produce.

Because pixel r on the image sensor ₀, r ₁, r ₂, r ₃... Be known, if known p ₁, then can solve p by formula (3) ₂, and then can solve p ₃, p ₄, promptly can on a certain direction, restore the actual value of slm pixel, can make p during record ₀=0, p then ₁=r ₁/ H ₁₁(σ).Because the pixel spread function of system can separate in x, y direction,, recover to obtain the actual value of the whole two-dimentional page pixel of spatial light modulator so can use similar algorithms on other direction, to carry out the recursion solution procedure again one time.In above-mentioned compensation method, work as σ ₂＜0 o'clock, the p in the formula (2) ₁Should be by p ₃Replace.But this moment p ₃Be unknown, recursion is found the solution can not be from p ₀Beginning, and must count from the other end.In real system, the symbol of the amount of mismatch σ of pixel possibly change on a certain direction, for example has magnification error or distortion, and this method is no longer suitable.In addition,, the pixel spread function width of system is not more than twice pixel separation (size of spectrum face footpath diaphragm is less) if having surpassed pixel separation, then image sensor pixel r ₂Value not only receive p ₁Influence, also receive p ₃Influence, this moment formula (2) also no longer suitable.

Summary of the invention

To the problems referred to above, the purpose of this invention is to provide a kind of signal to noise ratio (S/N ratio) that can improve body hologram memory system sense data page or leaf effectively, positive and negative insensitive to the pixel amount of mismatch, and have the body hologram memory system pixel mismatch compensation method of equalization treatment effect.

For realizing above-mentioned purpose; The present invention takes following technical method: a kind of body hologram memory system pixel mismatch compensation method; It is characterized in that: it adopts the pixel mismatch compensation model based on complex amplitude; The number of considering neighbor is three; Because the signal strength detection of image sensor pixel is the integration square on the light-sensitive surface of image sensor pixel of complex amplitude sum that its influential slm pixel is produced, when the width of pixel spread function surpasses a pixel and smaller or equal in two pixels the time, obtain pixel mismatch compensation model to be:

${\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">r</figure-callout>}}_2={\&Integral;}_{-{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}^{{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}{[\sqrt{p_3}h(x-{\mathrm{\&sigma;}}_3)+\sqrt{p_2}h(x-{\mathrm{\&sigma;}}_2)+\sqrt{p_1}h(x-{\mathrm{\&sigma;}}_1)]}^2\mathrm{dx},$

Wherein, σ ₁, σ ₂And σ ₃Be respectively slm pixel p ₁, p ₂And p ₃With respect to image sensor pixel r ₂The local mismatches amount; g ₂Fill factor, curve factor for the image sensor pixel; H (x) is the pixel spread function of system, i.e. the COMPLEX AMPLITUDE that on the image sensor face, produces of slm pixel.

Said pixel spread function h (x) is:

$h\left(x\right)=c_h{\&Integral;}_{-{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}^{{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}\sin c\left[\frac{D}{D_N}(x-x^{\&prime;})\right]{\mathrm{dx}}^{\&prime;},$

Sinc (x) ≡ sin (π x)/(π x) wherein, c _hFor satisfying

Constant, g ₁Be the fill factor, curve factor of image sensor pixel, x and x ' all are unit with the pixel separation, D _N=λ f/ Γ is the Nyquist aperture, and λ is an optical maser wavelength, and f is the focal length of fourier transform lens, and Γ is a pixel separation.

Said slm pixel is an arbitrary value with respect to the local mismatches amount of image sensor pixel.

The actual value of said pixel mismatch compensation model spatial light modulator pixel adopts the recursion alternative manner to find the solution; The initial row of the 2-D data page or leaf that order is stored and the pixel value of initial row are 0; Comprise the home block that is used to test the local mismatches amount in the data page; Its recursion iterative step is following: the amount of mismatch that receives the image identification piece is detected in (1), calculates the local mismatches amount of each pixel; (2) utilize the actual value of pixel mismatch compensation model recursion iterative slm pixel in the x direction; (3) utilize the actual value of pixel mismatch compensation model recursion iterative slm pixel in the y direction, its step is identical with said step (2).

In the said step (2); The approximate value that the recursion iterative of said slm pixel actual value comprises the steps: 1. to establish slm pixel is 0; In the substitution pixel mismatch compensation model, recursion is found the solution the approximate value that obtains slm pixel; 2. according to said step 1. recursion find the solution the approximate value of gained slm pixel, in the mismatch compensation of the substitution pixel again model, carry out once more recursion and find the solution; 3. repeating step 2. once more than, the result that last recursion is found the solution is as the actual value of slm pixel.

The present invention is owing to take above technical method; It has the following advantages: the present invention is owing to adopt the pixel mismatch compensation model based on complex amplitude; Its number of considering neighbor is three, and the local mismatches amount of pixel can be arbitrary value, the actual value that adopts the recursion iterative algorithm to find the solution pixel.Therefore the present invention can effectively improve the signal to noise ratio (S/N ratio) of body hologram memory system sense data page or leaf, and positive and negative insensitive to the pixel amount of mismatch has the effect that equalization is handled.The present invention can be widely used in pixel spread function width and be not more than the twice pixel separation, and slm pixel can be in the body hologram memory system of arbitrary value with respect to the local mismatches amount of image sensor pixel.

Description of drawings

Fig. 1 a, Fig. 1 b are that the pixel mismatch is to the synoptic diagram of signal to noise ratio (S/N ratio) influence in the prior art, and wherein Fig. 1 a is that amount of mismatch is approximately zero reception image, and signal to noise ratio (S/N ratio) is 4.2; Fig. 1 b is the reception image of amount of mismatch when being 0.4 pixel, and signal to noise ratio (S/N ratio) is 1.5;

Fig. 2 is the pixel mismatch compensation model synoptic diagram of amount of mismatch σ >=0 o'clock in the prior art;

Fig. 3 is a pixel spread function synoptic diagram in the prior art;

Fig. 4 is amount of mismatch σ of the present invention＜0 an o'clock pixel mismatch compensation model synoptic diagram;

Fig. 5 receives the signal to noise ratio (S/N ratio) of the page along with x direction amount of mismatch changes synoptic diagram in the embodiment of the invention one;

Fig. 6 receives the signal to noise ratio (S/N ratio) of the page along with y direction amount of mismatch changes synoptic diagram in the embodiment of the invention one;

Fig. 7 a～Fig. 7 d is x in the embodiment of the invention one, the y direction page synoptic diagram before and after 1/3 pixel compensation that squints respectively, and wherein, Fig. 7 a is the spatial light modulator load page; Fig. 7 b is that image sensor receives the page, uncompensated signal to noise ratio (S/N ratio) 2.45; Fig. 7 c is for adopting the page after the processing of IBM compensation method in the prior art, signal to noise ratio (S/N ratio) 7.37; Fig. 7 d is for adopting the page after three pixel compensation methods are handled, signal to noise ratio (S/N ratio) 10.20;

Fig. 8 receives the signal to noise ratio (S/N ratio) of the page with magnification error change synoptic diagram in the embodiment of the invention two;

Fig. 9 a～Fig. 9 d is that relative magnification is 1.1 o'clock pages before and after the compensation in the embodiment of the invention two, and wherein, Fig. 9 a is the spatial light modulator load page; Fig. 9 b is that image sensor receives the page, signal to noise ratio (S/N ratio) 2.03; Fig. 9 c is for adopting the page after the processing of IBM compensation method in the prior art, signal to noise ratio (S/N ratio) 4.03; Fig. 9 d is for adopting the page after three pixel compensation methods are handled, signal to noise ratio (S/N ratio) 8.93.

Embodiment

Below in conjunction with accompanying drawing and embodiment the present invention is carried out detailed description.

As shown in Figure 4, the present invention adopts the pixel mismatch compensation model based on complex amplitude, considers that the number of neighbor is three, and then the signal strength detection of four pixel segment of image sensor on the one dimension direction is that the image sensor pixel is respectively r ₀, r ₁, r ₂, r ₃, four slm pixel corresponding with each image sensor pixel are that light intensity is respectively p ₀, p ₁, p ₂, p ₃, and four slm pixel produce micro-translation σ with respect to image sensor.The signal strength detection of each image sensor pixel is the integration square on the light-sensitive surface of image sensor pixel of complex amplitude sum that its influential slm pixel is produced.Because the width of pixel spread function h (x) surpasses a pixel and is not more than two pixels, then the signal strength detection r of image sensor pixel ₂Value not only receives light intensity p ₂And p ₁Influence, also receive light intensity p ₃Influence, therefore can obtain pixel mismatch compensation model and be:

${\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">r</figure-callout>}}_2={\&Integral;}_{-{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}^{{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">g</figure-callout>}}_2/2}{[\sqrt{p_3}h(x-{\mathrm{\&sigma;}}_3)+\sqrt{p_2}h(x-{\mathrm{\&sigma;}}_2)+\sqrt{p_1}h(x-{\mathrm{\&sigma;}}_1)]}^2\mathrm{dx},---\left(4\right)$

Wherein, σ ₁, σ ₂And σ ₃Be respectively p ₁, p ₂And p ₃With respect to image sensor pixel r ₂The local mismatches amount; g ₂Fill factor, curve factor for the image sensor pixel; H (x) is the pixel spread function of system, and its definition is identical with the middle pixel spread function definition of prior art formula (1), is:

$h\left(x\right)=c_h{\&Integral;}_{-{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}^{{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="HSA00000315329800014.png,HSA00000315329800022.png"\; state="\{\{state\}\}">g</figure-callout>}}_1/2}\sin c\left[\frac{D}{D_N}(x-x^{\&prime;})\right]{\mathrm{dx}}^{\&prime;},$

Sinc (x) ≡ sin (π x)/(π x) wherein, c _hFor satisfying

Constant, g ₁Be the fill factor, curve factor of image sensor pixel, x and x ' all are unit with the pixel separation, D _N=λ f/ Γ is Nyquist (Nyquist) aperture, and λ is an optical maser wavelength, and f is the focal length of fourier transform lens, and Γ is a pixel separation.

Thus, can be with formula (4) arrangement:

${\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HSA00000315329800014.png"\; state="\{\{state\}\}">r</figure-callout>}}_2=p_3{H}_{33}\left(\mathrm{\&sigma;}\right)+p_2{H}_{22}\left(\mathrm{\&sigma;}\right)+p_1{H}_{11}\left(\mathrm{\&sigma;}\right)$

$+2\sqrt{p_3{p}_2}{H}_{32}\left(\mathrm{\&sigma;}\right)+2\sqrt{p_3{p}_1}{H}_{31}\left(\mathrm{\&sigma;}\right)+2\sqrt{p_2{p}_1}{H}_{21}\left(\mathrm{\&sigma;}\right),---\left(5\right)$

Wherein, H ₁₁(σ), H ₂₂(σ) and H ₃₃(σ) represent the normalized signal intensity that left neighbor, respective pixel and right neighbor self interference produces respectively; H ₃₂(σ), H ₃₁(σ) and H ₂₁(σ) represent the normalized signal intensity that respective pixel, left and right sides neighbor are interfered generation each other, their numerical value can be through calculating or experiment test obtain in advance.

Relate to three adjacent slm pixel in the pixel mismatch compensation model of the present invention, therefore pixel mismatch compensation method of the present invention can abbreviate " three pixel compensation methods " as.

The actual value of above-mentioned pixel mismatch compensation model spatial light modulator pixel adopts the recursion alternative manner to find the solution.Because local mismatches amount σ ₁, σ ₂And σ ₃Be the required argument of recursion method for solving, need in data page, interpolation be used to test the home block of local mismatches amount.In addition, recursion is found the solution and is required known initial value, therefore makes the initial row of the 2-D data page or leaf of being stored and the pixel value of initial row be 0.But, owing to contain unknown term light intensity p in the pixel mismatch compensation model formation (5) ₃, therefore can't directly adopt recursion to find the solution.So suppose the light intensity p in the pixel mismatch compensation model formation (5) earlier ₃=0, adopt recursion to find the solution the approximate value that obtains slm pixel, the actual value that utilizes pixel mismatch compensation model formation (5) recursion to find the solution slm pixel again.Its recursion iterative step is following:

1) detects the amount of mismatch that receives the image identification piece, calculate the local mismatches amount of each pixel;

2) utilize the actual value of pixel mismatch compensation model formation (5) recursion iterative slm pixel in the x direction, it comprises the steps:

The approximate value of (1) establishing slm pixel is 0, and in the substitution pixel mismatch compensation model formation (5), recursion is found the solution (p ₀→ p ₁→ p ₂→ p ₃) obtain the approximate value of slm pixel;

(2) find the solution the approximate value of gained slm pixel according to step (1) recursion, in the mismatch compensation of the substitution pixel again model formation (5), carry out once more recursion and find the solution (p ₀→ p ₁→ p ₂→ p ₃);

(3) repeating step (2) one or many, the result that last recursion is found the solution is as the actual value of slm pixel;

Wherein, the number of times of repeating step (2) is taken all factors into consideration the back according to required computational accuracy and computation complexity and is confirmed, rule of thumb, generally repeats once to get final product;

3) utilize the actual value of pixel mismatch compensation model formation (5) recursion iterative slm pixel in the y direction, its step and step 2) in step (1)～step (3) identical.

In order to further specify the present invention, below in conjunction with specific embodiment and two examples of particular instance the present invention is done to further describe, when disclosing exemplary embodiment of the present invention, only be used to the purpose of demonstrating rather than limit scope of the present invention.

Embodiment one: the pixel amount of mismatch of selection is a constant; Choose following data and carry out emulation: the front and back group focal length of fourier transform lens is respectively 44mm and 40mm, and the pixel separation of spatial light modulator is 13.2 μ m, and fill factor, curve factor is 0.90; The pixel separation of image sensor is 12 μ m; Fill factor, curve factor is at x, and the y direction is respectively 0.87 and 0.54, and spectrum face diaphragm is of a size of 1.5 times of Nyquist apertures.The test page size is 64 * 64 pixels, two circle pixel zero setting on every side, and central authorities' 60 * 60 pixels are 1,0 to fill at random.With spatial light modulator page analog imaging to image sensor; And make image sensor at x; The y direction produces a certain amount of translation respectively, calculate respectively the page directly receive signal to noise ratio (S/N ratio), adopt the page after the compensation of IBM compensation method in the prior art signal to noise ratio (S/N ratio), adopt the signal to noise ratio (S/N ratio) of the page after the compensation of three pixel compensation methods.When obtaining the pixel mismatch and existing only in x direction or y direction, receive of the variation (like Fig. 5, shown in Figure 6, in figure solid line be uncompensated method, fine dotted line be three pixel compensation methods, thick dashed line be IBM compensation method) of the signal to noise ratio (S/N ratio) of the page along with amount of mismatch.

No matter can see thus, always three pixel compensation methods of the present invention are superior to the IBM compensation method in x direction or y direction.When amount of mismatch when negative, the IBM compensation method is little for the improvement of signal to noise ratio (S/N ratio), and three pixel compensation methods positive and negative insensitive for amount of mismatch can significantly be improved signal to noise ratio (S/N ratio).X direction and the signal to noise ratio (S/N ratio) curve shape of y direction are not both because the fill factor, curve factor of image sensor pixel causes in that both direction is different.In the x direction, it is about 50% that the signal to noise ratio (S/N ratio) of IBM compensation method when uncompensated improved, and three pixel compensation methods have improved about 100% with signal to noise ratio (S/N ratio).In the y direction; Because the fill factor, curve factor of image sensor pixel is less; Uncompensated gained signal to noise ratio (S/N ratio) is insensitive for less amount of mismatch (less than 0.15 pixel), and the IBM compensation method keeps higher signal to noise ratio (S/N ratio) in the pixel amount of mismatch scope of [0.2,0.4]; And three pixel compensation methods have all improved about 100% (shown in Fig. 7 a～Fig. 7 d) with signal to noise ratio (S/N ratio) under the situation of other amount of mismatch except uncompensated effect when approaching-0.5 pixel of amount of mismatch.

Three pixel compensation methods of the present invention can also play and be similar to the effect that equalization is handled; For example when the pixel amount of mismatch near 0 the time; Three pixel compensation methods can be approximately a kind of 3 * 3 deconvolution, can eliminate between the page or leaf interior pixel that causes the system finite aperture and crosstalk, and then improve the signal to noise ratio (S/N ratio) of the page.

Embodiment two: select the pixel amount of mismatch to be the variable with the different regular variations in position, the pixel mismatch that the magnification error is brought compensates.This moment, the local mismatches amount satisfied σ=σ ₀+ M (x-x ₀), σ wherein ₀Be the side-play amount of the initial pixel of spatial light modulator with respect to respective pixel on the image sensor, M is relative magnification, and x is the coordinate of slm pixel, x ₀Coordinate for the initial pixel of spatial light modulator.

When being applied to magnification error compensation through numerical simulation three pixel compensation methods more of the present invention and IBM compensation method for the improvement situation of signal to noise ratio (S/N ratio).Simulation parameter is: the focal length of organizing behind the fourier transform lens is 40mm, changes the focal length of organizing before the fourier transform lens, makes the relative magnification of system fade to 1.2 gradually from 1; The pixel separation of spatial light modulator is 13.2 μ m; Fill factor, curve factor is 0.90, and the pixel separation of image sensor is 12 μ m, and fill factor, curve factor is at x; The y direction is respectively 0.87 and 0.54, and spectrum face diaphragm is of a size of 1.5 times of Nyquist apertures.The test page size is 64 * 64 pixels, two circle pixel zero setting on every side, and central authorities' 60 * 60 pixels are 1,0 to fill at random, and the slm pixel initial offset is 1/3 pixel.

As shown in Figure 8, x representes uncompensated method among the figure, and square is represented the IBM compensation method, and roundlet is represented three pixel compensation methods.Owing to can know that three pixel compensation methods are superior to the IBM compensation method.When relative magnification was 1, two kinds of compensation methodes were quite remarkable to the improvement of signal to noise ratio (S/N ratio); When relative magnification error greater than 0 and less than 0.07 the time, two kinds of compensation methodes are little for the raising of signal to noise ratio (S/N ratio); When relative magnification error greater than 0.07 the time; The IBM compensation method is still not obvious for the raising of signal to noise ratio (S/N ratio); And three pixel compensation methods can significantly improve signal to noise ratio (S/N ratio), make signal to noise ratio (S/N ratio) after the compensation deals reach the twice (shown in Fig. 9 a～Fig. 9 d) of the signal to noise ratio (S/N ratio) after the compensation deals of IBM compensation method.

Hence one can see that, and the situation when the pixel amount of mismatch is stochastic variable also can adopt and be similar to aforementioned pixel compensation method and compensate.Therefore, slm pixel of the present invention can be arbitrary value with respect to the local mismatches amount of image sensor pixel.Amount of mismatch can be a forward, also can be negative sense; Amount of mismatch can be a constant, can be with the different and variable of regular variation in position, also can be the variable of random variation within the specific limits.

Above-mentioned each embodiment only is used to explain the present invention; Protection scope of the present invention is not limited thereto; On the basis of technical method of the present invention, all improvement and equivalents of individual steps being carried out according to the principle of the invention all should not got rid of outside protection scope of the present invention.

Claims (5)

1. body hologram memory system pixel mismatch compensation method; It is characterized in that: it adopts the pixel mismatch compensation model based on complex amplitude; The number of considering the neighbor of image sensor is three; Because the signal strength detection of image sensor pixel is the integration square on the light-sensitive surface of image sensor pixel of complex amplitude sum that its influential slm pixel is produced; When the width of pixel spread function surpasses a pixel and during smaller or equal to two pixels, obtain pixel mismatch compensation model to be:

$r_2={\&Integral;}_{-g_2/2}^{g_2/2}{[\sqrt{p_3}h(x-{\mathrm{\&sigma;}}_3)+\sqrt{p_2}h(x-{\mathrm{\&sigma;}}_2)+\sqrt{p_1}h(x-{\mathrm{\&sigma;}}_1)]}^2\mathrm{dx},$

Wherein, σ ₁, σ ₂And σ ₃Be respectively slm pixel p ₁, p ₂And p ₃With respect to image sensor pixel r ₂The local mismatches amount; g ₂Fill factor, curve factor for the image sensor pixel; X is unit with the pixel separation; H (x) is the pixel spread function of system, i.e. the COMPLEX AMPLITUDE that on the image sensor face, produces of slm pixel.

2. a kind of body hologram memory system pixel mismatch compensation method as claimed in claim 1, it is characterized in that: said pixel spread function h (x) is:

$h\left(x\right)=c_h{\&Integral;}_{-g_1/2}^{g_1/2}\sin c\left[\frac{D}{D_N}(x-x^{\&prime;})\right]d{x}^{\&prime;},$

Sinc (x) ≡ sin (π x)/(π x) wherein, c _hFor satisfying Constant, g ₁Be the fill factor, curve factor of slm pixel, x and x ' all are unit with the pixel separation, and D is the spectrum face square aperture diaphragm length of side, D _N=λ f/ Γ is the Nyquist aperture, and λ is an optical maser wavelength, and f is the focal length of fourier transform lens, and Γ is a pixel separation.

3. a kind of body hologram memory system pixel mismatch compensation method as claimed in claim 1, it is characterized in that: said slm pixel is an arbitrary value with respect to the local mismatches amount of image sensor pixel.

4. a kind of body hologram memory system pixel mismatch compensation method as claimed in claim 2, it is characterized in that: said slm pixel is an arbitrary value with respect to the local mismatches amount of image sensor pixel.

5. like claim 1 or 2 or 3 or 4 described a kind of body hologram memory system pixel mismatch compensation methods; It is characterized in that: the actual value of said pixel mismatch compensation model spatial light modulator pixel adopts the recursion alternative manner to find the solution; The initial row of the 2-D data page or leaf that order is stored and the pixel value of initial row are 0; Comprise the home block that is used to test the local mismatches amount in the data page, its recursion iterative step is following:

(1) detects the amount of mismatch that receives the image identification piece, calculate the local mismatches amount of each pixel;

(2) utilize the actual value of pixel mismatch compensation model recursion iterative slm pixel in the x direction, it comprises the steps:

The approximate value of 1. establishing slm pixel is 0, and in the substitution pixel mismatch compensation model, recursion is found the solution the approximate value that obtains slm pixel;

2. according to said step 1. recursion find the solution the approximate value of gained slm pixel, in the mismatch compensation of the substitution pixel again model, carry out once more recursion and find the solution;

3. repeating step 2. once more than, the result that last recursion is found the solution is as the actual value of slm pixel;

(3) utilize the actual value of pixel mismatch compensation model recursion iterative slm pixel in the y direction, its step is identical with said step (2).

Images