CN104484879A - Method for estimating number of optimal orders - Google Patents

Abstract

The invention discloses a method for estimating the number of optimal orders. The method is suitable for estimating the number of corresponding orders when the quality of a high-order ghost imaging image is the best, and the number of the orders at the moment is the number of the optimal orders; the number of pixel blocks corresponding to the number of pixels in an image information area is called as the number of effective pixels; high-level processing is performed on a reference image and an object image through stimulating a ghost imaging experiment, a reconstructed image is stimulated, and the contrast ratio and the carrier-to-noise ratio of the reconstructed image are calculated; in combination with a high-order ghost imaging image quality assessment method, the number of the optimal orders corresponding to high-order ghost imaging under the condition of the number of the different effective pixels is estimated, and a relation between the number of the optimal orders and the number of the effective pixels is preliminarily speculated. According to the method, based on data information of the reconstructed image, the effect of the contrast ratio and the carrier-to-noise ratio on the image quality is considered. Compared with a traditional method for determining the number of the optimal orders, the method has the benefits that the image quality assessment is more complete, the actual experiment results are more satisfied, and the practicability is higher.

Description

A kind of optimal factor evaluation method

Technical field

The invention belongs to technical field of image quality evaluation, particularly a kind of optimal factor evaluation method.

Background technology

Ghost imaging (ghost imaging), also known as two-photon imaging (two-photon imaging) or relevance imaging (correlated imaging), is a kind of novel imaging technique of one utilizing two-photon to meet detection recovery object under test spatial information.Traditional optical observation is the distribution measuring of the intensity based on light field, association optics is then based on the correlation measurement of the intensity of light field, and existing imaging technique mainly utilizes the single order related information (intensity and position phase) of light field, and the double velocity correlation of the light field that classical terrible imaging utilizes is considered to a kind of statistical correlation of strength fluctuation.Nineteen ninety-five, Pittman first reported and utilizes the generation of SPDC method to tangle light, and completes terrible imaging experiment with this.But poor based on the terrible image quality of thermal light source, have impact on the practicality of terrible imaging.(PittmanT B, Shih Y H, Strekalov D V, Sergienko A V Optical imagingf by means oftwo-photon quantum entanglement Phys.Rev.A 52 R3429 1995) research show, the imaging of high-order ghost can improve image quality, along with the increase of exponent number, the High order correletion of light field is utilized to significantly improve image contrast.2009, the imaging of Kam Wai Clifford Chan group study high-order ghost, found except contrast, and carrier-to-noise ratio is also evaluate the important parameter of high-order ghost image quality, and carrier-to-noise ratio is higher, and picture quality is better.(Chan K W C, O ' Sullivan M N, Boyd R W High-order thermal ghost imagingOpt.Lett.34 3343 – 3,345 2009) usually only adopt contrast or one of them parameter of carrier-to-noise ratio to evaluate the picture quality of high-order ghost imaging at present.Along with the increase of object light exponent number, contrast and carrier-to-noise ratio first rise and decline, occur peak value at certain specific object light exponent number.Object light exponent number corresponding to contrast and carrier-to-noise ratio peak value is separately also different, how to select suitable object light exponent number to be the key issue improving terrible image quality.

Summary of the invention

The object of the present invention is to provide one to determine optimal factor evaluation method, set up image quality evaluation model, by the relation between numerical fitting initial guess optimal factor and valid pixel number.

The technical solution realizing the object of the invention is: a kind of optimal factor evaluation method, comprises the following steps:

(1) imaging of high-order ghost is divided into two bundle relative photo by signal optical source by BS, light beam is directly received by CCD, be called as reference light source, another light beam is by being received by bucket detector after object, be called as object light source, picture quality is improved, namely according to formula by the signal exponent number improving reference light source and object light source

$G_{m,n}\left(x\right)=\frac{1}{N}\underset{s=1}{\overset{N}{\mathrm{\Σ}}}{\left[I_0^{\left(s\right)}\right]}^m{\left[I^{\left(s\right)}\left(x\right)\right]}^n$

Can obtain after calculating rebuilding image; Wherein, G _m,nx () rebuilds image; it is object light source; I ^(s)x () is reference light source, s refers to which sample calculated; N is sample number; M, n are respectively the exponent number of object light source and reference light source;

(2) be multiplied by 255 by after reconstructed image data normalization, make it be compressed in 0 to 255 scopes, according to formula

$V=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{<G_{m,n}\left(x_{\mathrm{in}}\right)>+<G_{m,n}\left(x_{\mathrm{out}}\right)>}$

$\mathrm{CNR}=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{in}}\right)+\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{out}}\right)}$

Calculate the contrast and carrier-to-noise ratio of rebuilding image; Wherein V represents contrast, and CNR represents carrier-to-noise ratio, and angle brackets <> represents average, and Xin is picture signal region; Xout is image background regions, and Δ represents gets standard deviation;

(3) set contrast curves function as V=K (m), exponent number m1 be contrast maximum time corresponding exponent number, if carrier-to-noise ratio curvilinear function is CNR=L (m), exponent number m2 be carrier-to-noise ratio maximum time corresponding exponent number, along with the increase of object light source exponent number, contrast and carrier-to-noise ratio first rise when declining again and occur peak value;

High-order ghost imaging picture element preferably time corresponding exponent number be called optimal factor, if optimal factor is m0, now corresponding reconstruction image picture element is best; Contrast maximal value and carrier-to-noise ratio maximal value be corresponding exponent number m1, a m2 respectively, is shifted onto can obtain optimal factor m0 and m1, m2 coincidence by high-order ghost image Environmental Evaluation Model, or picture quality time between m1 and m2 is best;

(4) the block of pixels number that image information area pixel number is corresponding is called valid pixel number, choose plus sige respectively, the pattern of letter and Chinese character is as simulated object, by step (1), step (2) contrast that under calculating each valid pixel said conditions, different object light sources exponent number is corresponding and carrier-to-noise ratio, again by the high-order ghost image Environmental Evaluation Model evaluate image quality of contrast in step (3) and carrier-to-noise ratio joint effect, thus the optimal factor in-scope obtained under each valid pixel said conditions, optimal factor in-scope is averaged, this average data and valid pixel number are carried out matching, the fitting function obtaining optimal factor and valid pixel number is $m0=1.660*\sqrt{T}+4.119,$ T is valid pixel number.

The present invention compared with prior art, its remarkable advantage is: (1) is considered more comprehensively the factor affecting picture quality, this method propose a kind of new image quality measure method, contain contrast and carrier-to-noise ratio two factors in function model to the impact of picture quality; (2) there is higher practicality, the method based on image quality measure model, by being averaged optimal factor scope, preresearch estimates optimal factor, and propose the fit correlation between optimal factor and valid pixel number.(3) have generality, the method image quality evaluation model does not discuss contrast and carrier-to-noise ratio to the impact of picture quality on concrete formula, just analyzes on Changing Pattern.

Below in conjunction with accompanying drawing, the present invention is described in further detail.

Accompanying drawing explanation

Fig. 1 is process flow diagram in optimal factor evaluation method of the present invention.

Fig. 2 is that in optimal factor evaluation method of the present invention, valid pixel number is the list seam image of 20.

Fig. 3 is contrast curves V-m in optimal factor evaluation method of the present invention.

Fig. 4 is carrier-to-noise ratio curve C NR-m in optimal factor evaluation method of the present invention.

Fig. 5 to be valid pixel number of the present invention be 20 list seam rebuild image.

Fig. 6 is plus sige of the present invention, letter " F " and Chinese character " greatly " image.

Fig. 7 is the matched curve of optimal factor of the present invention and valid pixel number.

Embodiment

Composition graphs 1, optimal factor evaluation method of the present invention, comprises the following steps:

(1) imaging of high-order ghost is divided into two bundle relative photo by signal optical source by BS, light beam is directly received by CCD, be called as reference light source, another light beam is by being received by bucket detector after object, be called as object light source, picture quality is improved, namely according to formula by the signal exponent number improving reference light source and object light source

$G_{m,n}\left(x\right)=\frac{1}{N}\underset{s=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left[I_0^{\left(s\right)}\right]}^m{\left[I^{\left(s\right)}\left(x\right)\right]}^n$

Can obtain after calculating rebuilding image; Wherein, G _m,nx () rebuilds image; it is object light source; I ^(s)x () is reference light source, s refers to which sample calculated; N is sample number; M, n are respectively the exponent number of object light source and reference light source;

(2) be multiplied by 255 by after reconstructed image data normalization, make it be compressed in 0 to 255 scopes, according to formula

$V=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{<G_{m,n}\left(x_{\mathrm{in}}\right)>+<G_{m,n}\left(x_{\mathrm{out}}\right)>}$

$\mathrm{CNR}=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{in}}\right)+\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{out}}\right)}$

Calculate the contrast and carrier-to-noise ratio of rebuilding image; Wherein V represents contrast, and CNR represents carrier-to-noise ratio, and angle brackets <> represents average, and Xin is picture signal region; Xout is image background regions, and Δ represents gets standard deviation;

(3) set contrast curves function as V=K (m), exponent number m1 be contrast maximum time corresponding exponent number, if carrier-to-noise ratio curvilinear function is CNR=L (m), exponent number m2 be carrier-to-noise ratio maximum time corresponding exponent number, along with the increase of object light source exponent number, contrast and carrier-to-noise ratio first rise when declining again and occur peak value;

High-order ghost imaging picture element best (contrast and carrier-to-noise ratio maximum time picture element best) time corresponding exponent number be called optimal factor, if optimal factor is m0, now corresponding reconstruction image picture element is best; Contrast maximal value and carrier-to-noise ratio maximal value be corresponding exponent number m1, a m2 respectively, is shifted onto can obtain optimal factor m0 and m1, m2 coincidence by high-order ghost image Environmental Evaluation Model, or picture quality time between m1 and m2 is best;

(4) the block of pixels number that image information area pixel number is corresponding is called valid pixel number, choose plus sige respectively, the pattern of letter and Chinese character is as simulated object, by step (1), step (2) contrast that under calculating each valid pixel said conditions, different object light sources exponent number is corresponding and carrier-to-noise ratio, again by the high-order ghost image Environmental Evaluation Model evaluate image quality of contrast in step (3) and carrier-to-noise ratio joint effect, thus the optimal factor in-scope obtained under each valid pixel said conditions, optimal factor in-scope is averaged, this average data and valid pixel number are carried out matching, the fitting function obtaining optimal factor and valid pixel number is $m0=1.660*\sqrt{T}+4.119,$ T is valid pixel number.

Embodiment

As shown in Figure 1, optimal factor evaluation method of the present invention, comprises the following steps:

(1) the block of pixels number that image information area pixel number is corresponding is called valid pixel number.Get single seam image to emulate, valid pixel number is respectively 20,50,100,140,200,240.Valid pixel number be 20 list seam image as shown in Figure 2, valid pixel number equals signal area pixel count divided by block of pixels size.The imaging of high-order ghost is divided into two bundle relative photo by signal optical source by BS (beam splitter), light beam is directly received by CCD, be called as reference light source, another light beam is by being received by same detector after object, be called as object light source, the corresponding reference picture of reference light source, object light source homologue image.Improve picture quality by the signal exponent number improving reference light source and object light source, two-beam source information is according to formula

$G_{m,n}\left(x\right)=\frac{1}{N}\underset{s=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left[I_0^{\left(s\right)}\right]}^m{\left[I^{\left(s\right)}\left(x\right)\right]}^n$

Can obtain after calculating rebuilding image.Wherein, G _m,nx () rebuilds image; it is object light source; I ^(s)x () is reference light source; N is sample number; M, n are respectively the exponent number of object light source and reference light source.

(2) be multiplied by 255 by after reconstructed image data normalization, make it be compressed in 0 to 255 scopes, to get n be 1, m is 1 to 60, according to formula

$V=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{<G_{m,n}\left(x_{\mathrm{in}}\right)>+<G_{m,n}\left(x_{\mathrm{out}}\right)>}$

$\mathrm{CNR}=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{in}}\right)+\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{out}}\right)}$

Calculate the contrast and carrier-to-noise ratio of rebuilding image.Wherein V represents contrast, and CNR represents carrier-to-noise ratio, and angle brackets <> represents average, and Xin is picture signal region; Xout is image background regions, and Δ represents gets standard deviation.Contrast and carrier-to-noise ratio function curve are respectively as shown in Figure 3,4.

(3) set contrast curves function as V=K (m), exponent number m1 be contrast maximum time corresponding exponent number, if carrier-to-noise ratio curvilinear function is CNR=L (m), exponent number m2 be carrier-to-noise ratio maximum time corresponding exponent number, along with the increase of object light exponent number, contrast and carrier-to-noise ratio first rise and decline, occur peak value at certain specific object light exponent number.High-order ghost imaging picture element preferably time corresponding exponent number be called optimal factor, if optimal factor is m0, now corresponding reconstruction image picture element is best.

Because contrast is higher, picture quality is better; Carrier-to-noise ratio is higher, and picture quality is better.Consider the picture quality that contrast and carrier-to-noise ratio joint effect factor shift onto when can obtain optimal factor or (comprise m0 and m1, m2 to overlap) between m1 and m2 best, namely picture quality is determined jointly by contrast and carrier-to-noise ratio, a contrast maximal value and carrier-to-noise ratio maximal value corresponding exponent number respectively, the picture quality between two exponent numbers is best.Proof procedure is as follows:

1), during m1=m2, contrast and carrier-to-noise ratio reach maximal value at unified exponent number, so m0=m1=m2, namely optimal factor m0 is between m1 and m2.

2) during m1 ≠ m2, reduction to absurdity: suppose that optimal factor m0 is not between m1 and m2, namely m0 is outside m1 and m2.Now

V(m1)>V(m0),CNR(m1)>CNR(m0)

V(m2)>V(m0),CNR(m2)>CNR(m0)

Because contrast is higher, picture quality Q value is larger; Carrier-to-noise ratio is higher, and picture quality Q value is larger.So Q (m1) >Q (m0)

Q(m2)>Q(m0)

Be optimal factor contradiction with m0, so optimal factor m0 can be obtained between m1 and m2 by reduction to absurdity.

Valid pixel be the reconstruction image corresponding to list seam image optimum exponent number of 20 as shown in Figure 5.

(4) pattern of plus sige, letter " F " and Chinese character " greatly " is chosen respectively as simulated object, as shown in Figure 6.Valid pixel number is respectively 50,100,150,200,250,300,350,400,450 and 500.Be multiplied by 255 by after reconstructed image data normalization, make it be compressed in 0 to 255 scopes, to get n be 1, m is 1 to 60, according to formula

$V=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{<G_{m,n}\left(x_{\mathrm{in}}\right)>+<G_{m,n}\left(x_{\mathrm{out}}\right)>}$

$\mathrm{CNR}=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{in}}\right)+\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{out}}\right)}$

Calculate the contrast and carrier-to-noise ratio of rebuilding image.According to high-order ghost image Environmental Evaluation Model, picture quality is determined jointly by contrast and carrier-to-noise ratio, a contrast maximal value and carrier-to-noise ratio maximal value corresponding exponent number respectively, picture quality between two exponent numbers is best, optimal factor in-scope under each valid pixel said conditions, as shown in table 1:

The optimal factor that the different valid pixel of table 1 is corresponding

Wherein bracket [] represents interval, and T is valid pixel number.

Be described for the optimal factor his-and-hers watches 1 obtained when image plus sige T is respectively 50 and 400 in table 1: when T is 50, the span of optimal factor m0 is 14 to 16, in table 1, [1416] represent the optimal factor scope of the contrast that drawn by high-order ghost image Environmental Evaluation Model and corresponding m1, the m2 of carrier-to-noise ratio and these two values correspondences, wherein in bracket, left data is one less in corresponding exponent number m1, the m2 of contrast and carrier-to-noise ratio, and in bracket, right side data are one larger in corresponding exponent number m1, the m2 of contrast and carrier-to-noise ratio; When T is 400, optimal factor m0 is 37, and in table 1,37 represent that the contrast that drawn by high-order ghost image Environmental Evaluation Model and corresponding m1, the m2 of carrier-to-noise ratio are 37, so optimal factor m0 and m1, m2 coincidence.

Be averaged as shown in table 2 to optimal factor m0 in-scope:

The optimal factor that the different valid pixel of table 2 is corresponding

Be described for the optimal factor his-and-hers watches 2 obtained when image plus sige T is respectively 50 and 400 in table 2: when T is 50, the span of optimal factor m0 is 14 to 16, getting average to in-scope, to obtain optimal factor be 15, occurs decimal, then round if get in average process; When T is 400, optimal factor m0 is 37, is certain value, then directly get this definite value as optimal factor average.

3 groups of image optimal factors of gained under identical valid pixel number T condition are averaged, obtain final optimal factor as shown in table 3:

The optimal factor that the different valid pixel of table 3 is corresponding

Wherein it is the average of 3 object optimal factors.

(5) optimal factor in table 3 and the matching of valid pixel logarithmic data are depicted as curve, the fitting function that can obtain optimal factor and valid pixel number is shown below: matched curve as shown in Figure 7.

Claims (2)

1. an optimal factor evaluation method, is characterized in that comprising the following steps:

(1) imaging of high-order ghost is divided into two bundle relative photo by signal optical source by BS, light beam is directly received by CCD, be called as reference light source, another light beam is by being received by bucket detector after object, be called as object light source, picture quality is improved, namely according to formula by the signal exponent number improving reference light source and object light source

$G_{m,n}\left(x\right)=\frac{1}{N}\underset{s=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left[I_0^{\left(s\right)}\right]}^m{\left[I^{\left(s\right)}\left(x\right)\right]}^n$

Can obtain after calculating rebuilding image; Wherein, G _m,nx () rebuilds image; it is object light source; I ^(s)x () is reference light source, s refers to which sample calculated; N is sample number; M, n are respectively the exponent number of object light source and reference light source;

(2) be multiplied by 255 by after reconstructed image data normalization, make it be compressed in 0 to 255 scopes, according to formula

$V=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{<G_{m,n}\left(x_{\mathrm{in}}\right)>+<G_{m,n}\left(x_{\mathrm{out}}\right)\text{>}}$

$\mathrm{CNR}=\frac{<G_{m,n}\left(x_{\mathrm{in}}\right)>-<G_{m,n}\left(x_{\mathrm{out}}\right)>}{\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{in}}\right)+\mathrm{\&Delta;}{G}_{m,n}\left(x_{\mathrm{out}}\right)}$

Calculate the contrast and carrier-to-noise ratio of rebuilding image; Wherein V represents contrast, and CNR represents carrier-to-noise ratio, and angle brackets <> represents average, and Xin is picture signal region; Xout is image background regions, and Δ represents gets standard deviation;

(3) set contrast curves function as V=K (m), exponent number m1 be contrast maximum time corresponding exponent number, if carrier-to-noise ratio curvilinear function is CNR=L (m), exponent number m2 be carrier-to-noise ratio maximum time corresponding exponent number, along with the increase of object light source exponent number, contrast and carrier-to-noise ratio first rise when declining again and occur peak value;

High-order ghost imaging picture element preferably time corresponding exponent number be called optimal factor, if optimal factor is m0, now corresponding reconstruction image picture element is best; Contrast maximal value and carrier-to-noise ratio maximal value be corresponding exponent number m1, a m2 respectively, is shifted onto can obtain optimal factor m0 and m1, m2 coincidence by high-order ghost image Environmental Evaluation Model, or picture quality time between m1 and m2 is best;

(4) the block of pixels number that image information area pixel number is corresponding is called valid pixel number, choose plus sige respectively, the pattern of letter and Chinese character is as simulated object, by step (1), step (2) contrast that under calculating each valid pixel said conditions, different object light sources exponent number is corresponding and carrier-to-noise ratio, again by the high-order ghost image Environmental Evaluation Model evaluate image quality of contrast in step (3) and carrier-to-noise ratio joint effect, thus the optimal factor in-scope obtained under each valid pixel said conditions, optimal factor in-scope is averaged, this average data and valid pixel number are carried out matching, the fitting function obtaining optimal factor and valid pixel number is $m0=1.660*\sqrt{T}+4.119,$ T is valid pixel number.

2. optimal factor evaluation method according to claim 1, the n that it is characterized in that in step 2 is 1, m is 1 to 60.