CN103876772A - Multi-spectrum imaging method and device - Google Patents

Abstract

The invention discloses a multi-spectrum imaging method and device. The multi-spectrum imaging method includes the steps of determining an energy spectrum filtering range and an energy spectrum separating time number of a detected object, carrying out energy-variable imaging on the detected object to obtain a gradient energy projection sequence of energy spectrum separating, carrying out multi-spectrum CT reconstruction according to the gradient energy projection sequence to obtain attenuation images under different energies, and carrying out multi-spectrum fusing on the obtained attenuation images to obtain a required true-color CT image. By means of the scheme, the application range of a CT system can be expanded.

Description

A kind of multispectral formation method and device

Technical field

The present invention relates to computed tomography (CT, Computer Tomography) technology, particularly a kind of multispectral formation method and device.

Background technology

X ray CT imaging technique has been widely used in the fields such as medical science, industry, safety detection at present, and has brought into play the effect can not be substituted.

But along with the development of technology, be applicable to the existing Single energy X ray absorptionmetry CT imaging technique of flaw inspection, structural analysis etc., can not have met the functional imaging demand that in modern industry and modern medicine, component quantification detects.This is because existing CT system is all to realize the structure imaging of detected object based on monoenergetic hypothesis, cannot realize multispectral quantitative analysis and functional imaging, cannot realize multispectral imaging, thereby limit the scope of application of CT system.

Summary of the invention

In view of this, the invention provides a kind of multispectral formation method and device, can expand the scope of application of CT system.

In order to achieve the above object, technical scheme of the present invention is achieved in that

A kind of multispectral formation method, comprising:

The power spectrum filter range of determining detected object separates number of times with power spectrum;

According to definite result, described detected object is become to energy imaging, obtain the alternation energy projection sequence that power spectrum separates;

Carry out multispectral computer tomography CT reconstruction according to described alternation energy projection sequence, obtain the decay pattern picture under different-energy;

The each decay pattern obtaining is looked like to carry out multispectral fusion, obtain required very color CT image.

A kind of multispectral imaging device, comprising:

Ray filtering and separation module, separate number of times for the power spectrum filter range of determining detected object with power spectrum, and will determine that result sends to alternation energy imaging module;

Described alternation energy imaging module, for according to determining result, becomes energy imaging to described detected object, obtains the alternation energy projection sequence that power spectrum separates, and sends to multispectral computer tomography CT to rebuild module described alternation energy projection sequence;

Described multispectral CT rebuilds module, for carrying out multispectral CT reconstruction according to described alternation energy projection sequence, obtains the decay pattern picture under different-energy, and the each decay pattern obtaining is looked like to send to multispectral Fusion Module;

Described multispectral Fusion Module, for the each decay pattern obtaining is looked like to carry out multispectral fusion, obtains required very color CT image.

Visible, adopt scheme of the present invention, can be based on existing CT system, utilize ray filtering and become the technology such as energy imaging, realizing ray emission end power spectrum and separate, on this basis, can further utilize the technology such as multispectral CT reconstruction and multispectral fusion, obtain very color CT image, thereby realized multispectral imaging, and then expanded the scope of application of CT system.

Brief description of the drawings

Fig. 1 is the flow chart of the multispectral formation method embodiment of the present invention.

Fig. 2 is the composition structural representation of the multispectral imaging device embodiment of the present invention.

Detailed description of the invention

For make technical scheme of the present invention clearer, understand, referring to the accompanying drawing embodiment that develops simultaneously, scheme of the present invention is described in further detail.

Fig. 1 is the flow chart of the multispectral formation method embodiment of the present invention.As shown in Figure 1, comprise the following steps 11～14.

Step 11: the power spectrum filter range of determining detected object separates number of times with power spectrum.

Realize very color CT imaging, need to guarantee that the data for projection under each energy is monoenergetic or approximate monoenergetic data, and in existing CT system, due to bremsstrahlung, what obtain is projected as wide range projection, therefore cannot further realize power spectrum and separate.

For this reason, in this step, the first prior information such as component and overall dimensions based on detected object, the power spectrum filter range of determining detected object separates number of times with power spectrum, thereby provides energy parameter that foundation is set for follow-up obtaining of alternation energy projection sequence.

In addition, in x-ray imaging process, by the filter plate at the additional different materials of ray emission end and thickness, as aluminum, steel and tantalum etc., can make wider continuous spectrum become the narrow spectrum of approximate monoenergetic, thereby can overcome sclerosis artifact that continuous spectrum imaging occurs etc.; Meanwhile, the attenuation quotient of different materials, along with the rate of change difference of ray energy, under the condition of monoenergetic imaging, can reach the object that detected object component is distinguished according to the rate of change difference of attenuation quotient, thereby meet the demand of multispectral imaging; But, can increase the noise level of projection to ray filtering (energy spectral filter), thereby affect projection quality; , in scheme of the present invention, need to meet under the prerequisite of component discrimination for this reason, reduce as much as possible ray filtering degree, thereby ensure projection quality.

Correspondingly, in scheme of the present invention, for every kind of component i in detected object, can be respectively according to step 1)～3) shown in mode process.

1) according to the attenuation quotient rate of change model of component i, determine the ceiling capacity Emax that component i is corresponding;

The specific implementation of this step is prior art.

2) determine the span of the least energy Emin that this Emax is corresponding according to predetermined relationship;

Described predetermined relationship is: $I_0\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\Σ}}}S\left(E\right)\exp (-\underset{i=1}{\overset{I}{\mathrm{\Σ}}}{m}_i\left(E\right){\mathrm{\ρ}}_i{d}_i)\mathrm{\ΔE}\∈\mathrm{\χ};---\left(1\right)$

Wherein, S (E) is illustrated in after energy spectral filter, and the number of photons that energy is E accounts for ray through the number of photons I before detected object ₀ratio; I represents the component species number that detected object comprises, m _i(E) represent the mass attentuation coefficient of component i under energy E, ρ _irepresent the density of component i, d _ithe thickness that represents component i, χ represents the Limit of J-validity that detector can receive, and the value of above each parameter is all known, and Δ E represents the value interval of E, if value is 1;

In actual applications, need to ensure that in the span [Emin, Emax] of energy, the data that detector receives are effective, there will not be overexposure and under exposed situation after energy spectral filter.

3) determine respectively whether the each Emin that is positioned at span meets the following conditions:

$\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)m_i\left(E\right){\mathrm{\&rho;}}_i{d}_i\mathrm{\&Delta;E}<1;---\left(2\right)$

If so, retain this Emin, otherwise, this Emin abandoned;

In order to ensure projection quality, using retain each Emin in value minimum as Emin corresponding to component i.

By the way, obtain the power spectrum filter range of detected object, afterwards, can add up the different value numbers that occur in Emax corresponding to each component, and statistical result is separated to number of times as power spectrum, i.e. power spectrum separation number of times can be less than or equal to the component species number that detected object comprises.

Illustrate: the value of supposing I is 5, in detected object, comprise altogether 5 component kinds, if Emax corresponding to various component is respectively: 10,15,20,25,30, power spectrum separation number of times is 5 so; If Emax corresponding to various components is respectively: 10,15,20,25,10, power spectrum separation number of times is 4 so.

In addition, in the time that Emax corresponding to N component is identical, N is greater than 1 positive integer, can using value minimum in Emax corresponding this N component as this N component and Emin corresponding to this Emax.

And, for the Emax of each different values, can, according to monoenergetic approximate condition, determine material and the thickness of the filter plate that this Emax is corresponding according to (its corresponding component is corresponding) Emin of its correspondence respectively, specific implementation is prior art.

Step 12: according to definite result, detected object is become to energy imaging, obtain the alternation energy projection sequence that power spectrum separates.

In this step, can separate number of times with power spectrum according to the power spectrum filter range of determining in step 11, carry out CT scan imaging, the alternation energy projection sequence separating to obtain power spectrum.

In the time carrying out CT scan, can adopt traditional CT circle track scanning pattern, angle stepping is 1 degree, and angular range is 0～359 degree, and difference is that energy parameter changes in CT scan process.

Specifically, can be according to order from small to large, Emax to variant value sorts, and according to clooating sequence, carry out the CT scan of alternation energy according to the Emax of variant value, separate number of times according to power spectrum and determine the number of times of alternation energy, and can arrange by the tube voltage parameter of adjusting CT scan, realize the change of energy; In addition, also need to add respectively at ray emission window place the filter plate corresponding with the Emax of variant value.

Step 13: carry out multispectral CT reconstruction according to alternation energy projection sequence, obtain the decay pattern picture under different-energy.

In this step, the two-dimentional alternation energy projection sequence getting is carried out to three-dimensional reconstruction, to obtain the internal data structure of detected object under each energy in step 12.Preferably, can adopt the multispectral CT algorithm for reconstructing of low dosage, alternation energy projection sequence is carried out to multispectral CT reconstruction, to obtain the image that under each energy, quality better, component discrimination is higher.

In CT system, it is a stochastic process that the factors such as the scattering of X ray, the fluctuation noise of detector make the acquisition process of data for projection, the research of nuclear physics simultaneously also shows already, the radiation of X-ray meets Poisson random distribution, therefore, can utilize stochastic signal processing method, set up the statistical model of CT imaging process, and the theory that application parameter is estimated is carried out image reconstruction.

Visit first reading d (y) for y for detector under each energy, suppose that it is the Poisson random distribution of g (y, μ) that its obedience is expected:

$g(y,\mathrm{\&mu;})=\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}{I}_0(y,E)\exp [-\underset{x}{\mathrm{\&Sigma;}}{a}_{\mathrm{xy}}\mathrm{\&mu;}(x,E)];---\left(3\right)$

Wherein, I ₀(y, E) is illustrated in while thering is no detected object, visits the number of photons that first energy receiving is E, a for y _xyrepresent that y is visited the length of ray corresponding to unit through detected object x position, μ (x, E) represents line attenuation coefficient, relevant with energy and position; X represents to rebuild x pixel position in the decay pattern picture that obtains, in scheme of the present invention, in the time of reconstruction by discrete the cross section of detected object be digital picture;

Owing to there being the prior information of component, μ (x, E) can be expressed as the linear combination of I kind component, that is:

$(x,E)=\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_i\left(E\right)c_i\left(x\right);$

Wherein, μ _i(E) represent the line attenuation coefficient of component i under energy E, c _i(x) represent that component i is at the shared proportion in x position;

According to the character of Poisson random distribution, the logarithm maximum likelihood function of the first reading of all spies is:

$l(d,\mathrm{\&mu;})=\underset{y}{\mathrm{\&Sigma;}}[d\left(y\right)\ln \left(g(y,\mathrm{\&mu;})\right)-g(y,\mathrm{\&mu;})-\ln (d\left(y\right)!)];---\left(5\right)$

Correspondingly, what in scheme of the present invention, will do is exactly, and is meeting under formula (3) and (4) and non-negative constraint, obtains the c that makes formula (5) maximum _i(x).

Solve this restricted problem, can obtain reconstruction formula as follows:

$c_i^{(k+1)}\left(x\right)=c_i^{\left(k\right)}\left(x\right)-\frac{1}{Z_i\left(x\right)}\ln \left(\frac{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{p}^{\left(k\right)}(y,E)}{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{q}^{\left(k\right)}(y,E)}\right);---\left(6\right)$

Wherein, $q^{\left(k\right)}(y,E)=I_0(y,E)\&times;\exp (-\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{c_i}^k\left(x\right));---\left(7\right)$

$p^{\left(k\right)}(y,E)=q^{\left(k\right)}(y,E)\frac{d\left(y\right)}{{\mathrm{\&Sigma;}}_E{q}^{\left(k\right)}(y,E)};---\left(8\right)$

$Z_i\left(x\right)=\underset{(y,E)}{\max }\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}};---\left(9\right)$

K represents iterations, as 1,2,3 etc.; The span of i is from 1～I, and the span of E is from Emin～Emax, and Emin and Emax are Emin and the Emax that component i is corresponding, and the span of y is 1～M, and M represents the spy unit number that detector comprises; In addition, above-mentioned Z _i(x) be only preferably definition mode of one, also can adopt according to actual needs other definition mode, be not restricted; Above-mentioned μ _i(E), a _xy, I ₀the isoparametric value of (y, E), d (y) is all known.

Correspondingly, in scheme of the present invention, can, for each energy, determine respectively decay pattern picture corresponding to every kind of component i under this energy;

Wherein, the mode of determining the decay pattern picture that every kind of component i under this energy is corresponding can be:

Carry out following calculating according to the mode of iteration:

$c_i^{(k+1)}\left(x\right)=c_i^{\left(k\right)}\left(x\right)-\frac{1}{Z_i\left(x\right)}\ln \left(\frac{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{p}^{\left(k\right)}(y,E)}{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{q}^{\left(k\right)}(y,E)}\right);---\left(6\right)$

$q^{\left(k\right)}(y,E)=I_0(y,E)\&times;\exp (-\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{c_i}^k\left(x\right));---\left(7\right)$

$p^{\left(k\right)}(y,E)=q^{\left(k\right)}(y,E)\frac{d\left(y\right)}{{\mathrm{\&Sigma;}}_E{q}^{\left(k\right)}(y,E)};---\left(8\right)$

$Z_i\left(x\right)=\underset{(y,E)}{\max }\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}};---\left(9\right)$

When often obtaining a c _i ^k+1(x) afterwards, determine respectively c _i ^k+1and c (x) _i ^k(x) whether difference is less than predetermined threshold, if so, and by c _i ^k+1(x) as final result corresponding to x; The concrete value of described threshold value can be decided according to the actual requirements; Wherein, c _i ^k(x) initial value, value when k=1 can preset;

According to final result corresponding to variant x difference obtaining, generate decay pattern picture corresponding to component i, how to be generated as prior art.

Step 14: the each decay pattern obtaining is looked like to carry out multispectral fusion, obtain required very color CT image.

In this step, the each decay pattern getting in step 13 is looked like to carry out fusion treatment, to obtain can specification configuration distinguishing again the very color CT image of material.

Specifically, first, can utilize the multispectral blending algorithm based on principal component analysis, extract respectively relatively independent main constituent image in every width decay pattern picture, afterwards, can be according to extracting result, by adaptive weighted fusion, obtain fusion image, last, can be painted by fusion image is carried out, obtain required very color CT image; Specific implementation is prior art.

Based on above-mentioned introduction, Fig. 2 is the composition structural representation of the multispectral imaging device embodiment of the present invention.As shown in Figure 2, comprising:

Ray filtering and separation module, separate number of times for the power spectrum filter range of determining detected object with power spectrum, and will determine that result sends to alternation energy imaging module;

Alternation energy imaging module, for according to determining result, becomes energy imaging to detected object, obtains the alternation energy projection sequence that power spectrum separates, and sends to multispectral computer tomography CT to rebuild module alternation energy projection sequence;

Multispectral CT rebuilds module, for carrying out multispectral CT reconstruction according to alternation energy projection sequence, obtains the decay pattern picture under different-energy, and the each decay pattern obtaining is looked like to send to multispectral Fusion Module;

Multispectral Fusion Module, for the each decay pattern obtaining is looked like to carry out multispectral fusion, obtains required very color CT image.

Preferably,

Ray filtering and separation module can, for every kind of component i in detected object, according to the attenuation quotient rate of change model of component i, be determined the ceiling capacity Emax that component i is corresponding respectively; Determine the span of the least energy Emin that this Emax is corresponding according to predetermined relationship; Determine respectively whether the each Emin that is positioned at span meets predetermined condition; If so, retain this Emin, otherwise, this Emin abandoned; Using retain each Emin in value minimum as Emin corresponding to component i;

Wherein, described predetermined relationship is: $I_0\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)\exp (-\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}{m}_i\left(E\right){\mathrm{\&rho;}}_i{d}_i)\mathrm{\&Delta;E}\&Element;\mathrm{\&chi;};---\left(1\right)$

Described predetermined condition is: $\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)m_i\left(E\right){\mathrm{\&rho;}}_i{d}_i\mathrm{\&Delta;E}<1;---\left(2\right)$

S (E) is illustrated in after energy spectral filter, and the number of photons that energy is E accounts for ray through the number of photons I before detected object ₀ratio; I represents the component species number that detected object comprises, m _i(E) represent the mass attentuation coefficient of component i under energy E, ρ _irepresent the density of component i, d _ithe thickness that represents component i, χ represents the Limit of J-validity that detector can receive, Δ E represents the value interval of E;

Ray filtering can add up from separation module the different value numbers that occur in Emax corresponding to each component, and statistical result is separated to number of times as power spectrum.

In addition,

Ray filtering and separation module also can be further used for, and in the time that Emax corresponding to N component is identical, N is greater than 1 positive integer, using value minimum in Emax corresponding this N component as this N component and Emin corresponding to this Emax; For the Emax of each different values, determine respectively material and the thickness of the filter plate that this Emax is corresponding according to its corresponding Emin, and notice is to alternation energy imaging module;

Correspondingly, alternation energy imaging module can, according to order from small to large, sort to the Emax of variant value; According to clooating sequence, carry out the CT scan of alternation energy according to the Emax of variant value; And, add respectively the filter plate corresponding with the Emax of variant value at ray emission window place.

Preferably,

Multispectral CT rebuilds module can, for each energy, determine respectively decay pattern picture corresponding to every kind of component i under this energy; Wherein, for every kind of component i under this energy, carry out following calculating according to the mode of iteration respectively:

$c_i^{(k+1)}\left(x\right)=c_i^{\left(k\right)}\left(x\right)-\frac{1}{Z_i\left(x\right)}\ln \left(\frac{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{p}^{\left(k\right)}(y,E)}{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{q}^{\left(k\right)}(y,E)}\right);---\left(6\right)$

$q^{\left(k\right)}(y,E)=I_0(y,E)\&times;\exp (-\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{c_i}^k\left(x\right));---\left(7\right)$

$p^{\left(k\right)}(y,E)=q^{\left(k\right)}(y,E)\frac{d\left(y\right)}{{\mathrm{\&Sigma;}}_E{q}^{\left(k\right)}(y,E)};---\left(8\right)$

$Z_i\left(x\right)=\underset{(y,E)}{\max }\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}};---\left(9\right)$

When often obtaining a c _i ^k+1(x) afterwards, determine respectively c _i ^k+1and c (x) _i ^k(x) whether difference is less than predetermined threshold, if so, and by c _i ^k+1(x) as final result corresponding to x; According to final result corresponding to variant x difference obtaining, generate decay pattern picture corresponding to component i;

Wherein, μ _i(E) represent the line attenuation coefficient of component i under energy E, a _xyrepresent that y in detector is visited the length of ray corresponding to unit through detected object x position, I ₀(y, E) is illustrated in while thering is no detected object, visits the number of photons that first energy receiving is E, c for y in detector _i(x) represent that component i is at the shared proportion in x position, d (y) represents that y in detector visits first reading, and x represents to rebuild x pixel position in the decay pattern picture obtaining, and k represents iterations; The span of i is from 1～I, and the span of E is from Emin～Emax, and Emin and Emax are Emin and the Emax that component i is corresponding, and the span of y is 1～M, and M represents the spy unit number that detector comprises.

Preferably,

The multispectral blending algorithm of multispectral Fusion Module utilization based on principal component analysis, extracts respectively relatively independent main constituent image in every width decay pattern picture; According to extracting result, by adaptive weighted fusion, obtain fusion image; Painted by fusion image is carried out, obtain very color CT image.

The specific works flow process of Fig. 2 shown device embodiment please refer to the respective description in preceding method embodiment, repeats no more herein.In addition, in actual applications, Fig. 2 shown device can be applicable in CT system.

In sum, these are only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any amendment of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims (10)

1. a multispectral formation method, is characterized in that, comprising:

The power spectrum filter range of determining detected object separates number of times with power spectrum;

According to definite result, described detected object is become to energy imaging, obtain the alternation energy projection sequence that power spectrum separates;

Carry out multispectral computer tomography CT reconstruction according to described alternation energy projection sequence, obtain the decay pattern picture under different-energy;

The each decay pattern obtaining is looked like to carry out multispectral fusion, obtain required very color CT image.

2. method according to claim 1, is characterized in that,

The power spectrum filter range of described definite detected object comprises:

For every kind of component i in described detected object, carry out respectively following processing:

According to the attenuation quotient rate of change model of component i, determine the ceiling capacity Emax that component i is corresponding;

Determine the span of the least energy Emin that this Emax is corresponding according to predetermined relationship;

Described predetermined relationship is: $I_0\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)\exp (-\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}{m}_i\left(E\right){\mathrm{\&rho;}}_i{d}_i)\mathrm{\&Delta;E}\&Element;\mathrm{\&chi;};$

Wherein, S (E) is illustrated in after energy spectral filter, and the number of photons that energy is E accounts for ray through the number of photons I before described detected object ₀ratio; I represents the component species number that described detected object comprises, m _i(E) represent the mass attentuation coefficient of component i under energy E, ρ _irepresent the density of component i, d _ithe thickness that represents component i, χ represents the Limit of J-validity that detector can receive, Δ E represents the value interval of E;

Determine respectively whether the each Emin that is positioned at span meets the following conditions:

$\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)m_i\left(E\right){\mathrm{\&rho;}}_i{d}_i\mathrm{\&Delta;E}<1;$

If so, retain this Emin, otherwise, this Emin abandoned;

Using retain each Emin in value minimum as Emin corresponding to component i;

The power spectrum of described definite detected object separates number of times and comprises: add up the different value numbers that occur in Emax corresponding to each component, statistical result is separated to number of times as described power spectrum.

3. method according to claim 2, is characterized in that,

The method further comprises: in the time that Emax corresponding to N component is identical, N is greater than 1 positive integer, using value minimum in Emax corresponding this N component as this N component and Emin corresponding to this Emax;

For the Emax of each different values, determine respectively material and the thickness of the filter plate that this Emax is corresponding according to its corresponding Emin;

Described described detected object is become to energy imaging, obtains the alternation energy projection sequence that power spectrum separates and comprise:

According to order from small to large, the Emax of variant value is sorted;

According to clooating sequence, carry out the CT scan of alternation energy according to the Emax of variant value; And, add respectively the filter plate corresponding with the Emax of variant value at ray emission window place.

4. method according to claim 3, is characterized in that,

Describedly carry out multispectral CT reconstruction according to described alternation energy projection sequence, the decay pattern obtaining under different-energy looks like to comprise:

For each energy, determine respectively decay pattern picture corresponding to every kind of component i under this energy;

Wherein, every kind of decay pattern corresponding to component i under described definite this energy looks like to comprise:

Carry out following calculating according to the mode of iteration:

$c_i^{(k+1)}\left(x\right)=c_i^{\left(k\right)}\left(x\right)-\frac{1}{Z_i\left(x\right)}\ln \left(\frac{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{p}^{\left(k\right)}(y,E)}{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{q}^{\left(k\right)}(y,E)}\right);$

$q^{\left(k\right)}(y,E)=I_0(y,E)\&times;\exp (-\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{c_i}^k\left(x\right));$

$p^{\left(k\right)}(y,E)=q^{\left(k\right)}(y,E)\frac{d\left(y\right)}{{\mathrm{\&Sigma;}}_E{q}^{\left(k\right)}(y,E)};$

$Z_i\left(x\right)=\underset{(y,E)}{\max }\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}};$

Wherein, μ _i(E) represent the line attenuation coefficient of component i under energy E, a _xyrepresent that y in detector is visited the length of ray corresponding to unit through described detected object x position, I ₀(y, E) is illustrated in while thering is no described detected object, visits the number of photons that first energy receiving is E, c for y in detector _i(x) represent that component i is at the shared proportion in x position, d (y) represents that y in detector visits first reading, and x represents to rebuild x pixel position in the decay pattern picture obtaining, and k represents iterations; The span of i is from 1～I, and the span of E is from Emin～Emax, and Emin and Emax are Emin and the Emax that component i is corresponding, and the span of y is 1～M, and M represents the spy unit number that detector comprises;

When often obtaining a c _i ^k+1(x) afterwards, determine respectively c _i ^k+1and c (x) _i ^k(x) whether difference is less than predetermined threshold, if so, and by c _i ^k+1(x) as final result corresponding to x;

According to final result corresponding to variant x difference obtaining, generate decay pattern picture corresponding to component i.

5. according to the method described in claim 1,2,3 or 4, it is characterized in that,

Described the each decay pattern obtaining is looked like to carry out multispectral fusion, obtains required very color CT image and comprise:

Utilize the multispectral blending algorithm based on principal component analysis, extract respectively relatively independent main constituent image in every width decay pattern picture;

According to extracting result, by adaptive weighted fusion, obtain fusion image;

Painted by described fusion image is carried out, obtain described very color CT image.

6. a multispectral imaging device, is characterized in that, comprising:

Ray filtering and separation module, separate number of times for the power spectrum filter range of determining detected object with power spectrum, and will determine that result sends to alternation energy imaging module;

Described alternation energy imaging module, for according to determining result, becomes energy imaging to described detected object, obtains the alternation energy projection sequence that power spectrum separates, and sends to multispectral computer tomography CT to rebuild module described alternation energy projection sequence;

Described multispectral CT rebuilds module, for carrying out multispectral CT reconstruction according to described alternation energy projection sequence, obtains the decay pattern picture under different-energy, and the each decay pattern obtaining is looked like to send to multispectral Fusion Module;

Described multispectral Fusion Module, for the each decay pattern obtaining is looked like to carry out multispectral fusion, obtains required very color CT image.

7. device according to claim 6, is characterized in that,

Described ray filtering and separation module, for every kind of component i in described detected object, according to the attenuation quotient rate of change model of component i, are determined the ceiling capacity Emax that component i is corresponding respectively; Determine the span of the least energy Emin that this Emax is corresponding according to predetermined relationship; Determine respectively whether the each Emin that is positioned at span meets predetermined condition; If so, retain this Emin, otherwise, this Emin abandoned; Using retain each Emin in value minimum as Emin corresponding to component i;

Wherein, described predetermined relationship is: $I_0\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)\exp (-\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}{m}_i\left(E\right){\mathrm{\&rho;}}_i{d}_i)\mathrm{\&Delta;E}\&Element;\mathrm{\&chi;};$

Described predetermined condition is: $\underset{E=E_{\min }}{\overset{E_{\max }}{\mathrm{\&Sigma;}}}S\left(E\right)m_i\left(E\right){\mathrm{\&rho;}}_i{d}_i\mathrm{\&Delta;E}<1;$

S (E) is illustrated in after energy spectral filter, and the number of photons that energy is E accounts for ray through the number of photons I before described detected object ₀ratio; I represents the component species number that described detected object comprises, m _i(E) represent the mass attentuation coefficient of component i under energy E, ρ _irepresent the density of component i, d _ithe thickness that represents component i, χ represents the Limit of J-validity that detector can receive, Δ E represents the value interval of E;

Described ray filtering adds up from separation module the different value numbers that occur in Emax corresponding to each component, and statistical result is separated to number of times as described power spectrum.

8. device according to claim 7, is characterized in that,

Described ray filtering and separation module are further used for, and in the time that Emax corresponding to N component is identical, N is greater than 1 positive integer, using value minimum in Emax corresponding this N component as this N component and Emin corresponding to this Emax; For the Emax of each different values, determine respectively material and the thickness of the filter plate that this Emax is corresponding according to its corresponding Emin, and notice is to described alternation energy imaging module;

Described alternation energy imaging module, according to order from small to large, sorts to the Emax of variant value; According to clooating sequence, carry out the CT scan of alternation energy according to the Emax of variant value; And, add respectively the filter plate corresponding with the Emax of variant value at ray emission window place.

9. device according to claim 8, is characterized in that,

Described multispectral CT rebuilds module for each energy, determines respectively decay pattern picture corresponding to every kind of component i under this energy; Wherein, for every kind of component i under this energy, carry out following calculating according to the mode of iteration respectively:

$c_i^{(k+1)}\left(x\right)=c_i^{\left(k\right)}\left(x\right)-\frac{1}{Z_i\left(x\right)}\ln \left(\frac{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{p}^{\left(k\right)}(y,E)}{\underset{y}{\mathrm{\&Sigma;}}\underset{E}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{q}^{\left(k\right)}(y,E)}\right);$

$q^{\left(k\right)}(y,E)=I_0(y,E)\&times;\exp (-\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}}{c_i}^k\left(x\right));$

$p^{\left(k\right)}(y,E)=q^{\left(k\right)}(y,E)\frac{d\left(y\right)}{{\mathrm{\&Sigma;}}_E{q}^{\left(k\right)}(y,E)};$

$Z_i\left(x\right)=\underset{(y,E)}{\max }\underset{x}{\mathrm{\&Sigma;}}\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&mu;}}_i\left(E\right)a_{\mathrm{xy}};$

When often obtaining a c _i ^k+1(x) afterwards, determine respectively c _i ^k+1and c (x) _i ^k(x) whether difference is less than predetermined threshold, if so, and by c _i ^k+1(x) as final result corresponding to x; According to final result corresponding to variant x difference obtaining, generate decay pattern picture corresponding to component i;

Wherein, μ _i(E) represent the line attenuation coefficient of component i under energy E, a _xyrepresent that y in detector is visited the length of ray corresponding to unit through described detected object x position, I ₀(y, E) is illustrated in while thering is no described detected object, visits the number of photons that first energy receiving is E, c for y in detector _i(x) represent that component i is at the shared proportion in x position, d (y) represents that y in detector visits first reading, and x represents to rebuild x pixel position in the decay pattern picture obtaining, and k represents iterations; The span of i is from 1～I, and the span of E is from Emin～Emax, and Emin and Emax are Emin and the Emax that component i is corresponding, and the span of y is 1～M, and M represents the spy unit number that detector comprises.

10. according to the device described in claim 6,7,8 or 9, it is characterized in that,

The multispectral blending algorithm of described multispectral Fusion Module utilization based on principal component analysis, extracts respectively relatively independent main constituent image in every width decay pattern picture; According to extracting result, by adaptive weighted fusion, obtain fusion image; Painted by described fusion image is carried out, obtain described very color CT image.

Images