CN103377464A - Image processing method and system for removing ghost shadows - Google Patents

Abstract

The invention discloses an image processing method and system for removing ghost shadows. According to the method, firstly, target images and reference images are transformed to a gradient field, then a final image gradient field is obtained through differential treatment, and finally, corrected images are obtained through a least square method in a matching mode. According to the image processing method and system for removing the ghost shadows, the flat panel ghost shadows can be reduced or removed, so that the images are clearer, observation of a doctor is better facilitated, and the times of misdiagnosing are reduced. Meanwhile, compared with a hardware solution in the prior art, the image processing method and system are lower in cost and more flexible in deployment and have broader market application prospects.

Description

A kind of image processing method and system of eliminating ghost

Technical field

The present invention relates to technical field of image processing, particularly a kind of method and system of eliminating the ghost of target image.

Background technology

Along with the development of modern science and technology, the technological means that some are advanced and computer science and technology constantly are being applied in medical domain, and particularly in radiodiagnosis and treatment subject, computer image processing technology is just being brought into play more and more important effect.

Modern increasing x-ray imaging system adopts flat panel detector, mainly contain two large technology schools, amorphous silicon flat panel and amorphous selenium flat-bed, a common problem that exists is exactly ghost, especially amorphous selenium flat-bed, its ghost may exist a couple of days just to understand complete obiteration.Many dull and stereotyped production firms are by hardware design or improve technique and solve this problem, but technology is expensive, complicated, and still can't eliminate ghost fully.

In view of this, prior art is still waiting to improve and improve.

Summary of the invention

The object of the present invention is to provide a kind of image processing method and system of eliminating ghost, have the problem of ghost with the image on the solution prior art midplane detector.

In order to achieve the above object, the present invention has taked following technical scheme:

A kind of image processing method of eliminating ghost wherein, said method comprising the steps of:

A1, predetermined time shutter of basis gather reference picture;

A2, reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

A3, predetermined time shutter of basis gather target image;

A4, target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

A5, reference picture gradient field and target image gradient field are carried out calculus of differences, obtain the final image gradient field;

A6, obtain final image according to described final image gradient field.

The image processing method of described elimination ghost, wherein, when in the described steps A 2 reference picture being carried out noise reduction process, its concrete grammar is as follows:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}|I^{\′}(i,j)-I^{\′}(m,n)|}|I^{\′}(i,j)-I^{\′}(m,n)|I^{\′}(i,j);$

Wherein, I ' (m, n) is pixel pending on the reference picture, and I (i, j) is I ' (m, n) the interior pixel of 5 * 5 neighborhoods on every side, and m, n, i and j are natural number.

The image processing method of described elimination ghost wherein, transforms to the reference picture gradient field with the reference picture behind the noise reduction in the described steps A 2, adopts following algorithm:

$G(m,n)=\sqrt{{(I(m,n)-I(m-1,n))}^2+{(I(m,n)-I(m,n-1))}^2};$

Described G (m, n) is the Grad of (m, n) pixel in the reference picture gradient field, and the position that the Grad of all pixels is inserted separately successively just consists of the reference picture gradient field.

The method of the ghost of described elimination target image wherein, transforms to the reference picture gradient field with the reference picture behind the noise reduction in the described steps A 2, adopts following algorithm:

G(m，n)＝|I(m，n)-I(m-1，n)|+|I(m，n)-I(m，n-1)|；

Described G (m, n) is the Grad of (m, n) pixel in the reference picture gradient field, and the position that the Grad of all pixels is inserted separately successively just consists of the reference picture gradient field.

The image processing method of described elimination ghost, wherein, when in the described steps A 4 target image being carried out noise reduction process, adopt following wave filter:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}{C}_s{C}_d}{C}_s{C}_{d}I(i,j);$

Wherein, $C_s=e^{-\frac{{(I^{\′}(i,j)-I^{\′}(m,n))}^2}{{2\mathrm{\σ}}_s^2}};$ $C_d=e^{-\frac{{(i-m)}^2+{(j-n)}^2}{{2\mathrm{\σ}}_d^2}};$

σ _sThe similarity distribution standard deviation, σ _dIt is the space distribution standard deviation; I ' (m, n) is pending pixel, and I ' (i, j) is the pixel in 5 * 5 neighborhoods around it; M, n, i and j are natural number.

The image processing method of described elimination ghost wherein, carries out reference picture being carried out linear transformation before the calculus of differences to reference picture gradient field and target image gradient field in the described steps A 5, and its formula is as follows:

Linear transformation reference picture gradient field=A * reference picture gradient field+B;

Wherein, the interval of A is [0.5,1.5], and the interval of B is [0.2 reference picture gradient field, 0.2 reference picture gradient field].

The image processing method of described elimination ghost, wherein, it is to obtain final image by the least square method coupling by the final image gradient field that described steps A 6 is obtained final image according to described final image gradient field.

A kind of image processing system of eliminating ghost, wherein, described system comprises:

Reference picture acquisition module, the time shutter of being scheduled to for basis gather a sub-picture as the reference image;

The reference picture processing module, be used for reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

Target image acquisition module, the time shutter of being scheduled to for basis gather target image;

The target image processing module, be used for target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

Difference block, be used for reference picture gradient field and target image gradient field are done poor, obtain the final image gradient field;

The final image acquisition module, be used for obtaining final image according to described final image gradient field.

Beneficial effect:

Reduce or eliminate dull and stereotyped ghost, make image more clear, more be beneficial to the doctor and observe, reduce mistaken diagnosis, simultaneously, compare with the hardware solution of prior art, cost is lower, disposes more flexibly, has more wide market application foreground.

Description of drawings

Fig. 1 is the process flow diagram of the image processing method of elimination ghost of the present invention.

Fig. 2 is the structured flowchart of the image processing system of elimination ghost of the present invention.

Embodiment

For making purpose of the present invention, technical scheme and effect clearer, clear and definite, the present invention is described in more detail referring to the accompanying drawing examples.

See also Fig. 1, it is the process flow diagram of the image processing method of elimination ghost of the present invention, as shown in the figure, said method comprising the steps of:

S1, predetermined time shutter of basis gather reference picture;

S2, reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

S3, predetermined time shutter of basis gather target image;

S4, target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

S5, reference picture gradient field and target image gradient field are carried out calculus of differences, obtain the final image gradient field;

S6, obtain final image according to described final image gradient field.

The below is described in detail for above-mentioned steps respectively:

Described step S1 gathers a sub-picture as the reference image according to the predetermined time shutter, particularly, such as when the medical diagnosis, can be according to the diagnosis requirement, determine a time shutter, it is made as the predetermined time shutter, according to the above-mentioned predetermined time shutter, in the situation without exposure, gather a sub-picture as the reference image.

Described step S2 transforms to gradient field for reference picture is carried out noise reduction process with the reference picture behind the noise reduction, obtains the reference picture gradient field.Because in black white image, the profile of object is main information, therefore the image conversion after the denoising is arrived gradient field, the body feature that can keep image significantly reduces calculated amount simultaneously.In the present embodiment, when among the described step S2 reference picture being carried out noise reduction process, adopt the holding edge filter device, its concrete grammar is as follows:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}|I^{\′}(i,j)-I^{\′}(m,n)|}|I^{\′}(i,j)-I^{\′}(m,n)|I^{\′}(i,j);$

Wherein, I ' (m, n) is pixel pending on the reference picture, and I ' (i, j) is I ' (m, n) the interior pixel of 5 * 5 neighborhoods on every side, and I (m, n) is the pixel value after processing; M, n, i and j are natural number.Certainly, the included neighborhood of I ' (i, j) also can enlarge as required.At this moment, we are defined as G1 with the reference picture gradient field that obtains.

Carry out the mode of noise reduction process according to above-mentioned employing holding edge filter device, the algorithm of reference picture gradient field G1 is as follows:

$G(m,n)=\sqrt{{(I(m,n)-I(m-1,n))}^2+{(I(m,n)-I(m,n-1))}^2};$

Wherein, G (m, n) is (m among the reference picture gradient field G1, n) Grad of pixel, the Grad (G (m, n) that namely all pixels are corresponding) of each pixel is inserted corresponding position (i.e. (m, n)) successively just consist of reference picture gradient field G1.

In order to reduce operand, can further be reduced to:

G(m，n)＝|I(m，n)-I(m-1，n)|+|I(m，n)-I(m，n-1)|；

Described step S3 gathers target image according to the predetermined time shutter.Target image is will eliminate the image that ghost is processed, and the above-mentioned predetermined time shutter refers to require definite time according to diagnosis in step S1, and step S1 is identical with the time that step S3 uses.

S2 is basic identical with step, and described step S4 is for to carry out noise reduction process to target image.There is larger impact in the quality of target image quality to final image, and the method for noise reduction is more, in order to keep the details that contains in the target image as far as possible, in the present embodiment, has adopted one details kept preferably wave filter, and its concrete grammar is as follows:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}{C}_s{C}_d}{C}_s{C}_{d}I(i,j);$

Wherein, $C_s=e^{-\frac{{(I^{\′}(i,j)-I^{\′}(m,n))}^2}{{2\mathrm{\σ}}_s^2}}$ 、 $C_d=e^{-\frac{{(i-m)}^2+{(j-n)}^2}{{2\mathrm{\σ}}_d^2}};$

σ _sThe similarity distribution standard deviation, σ _dIt is the space distribution standard deviation; I ' (m, n) is pixel pending on the target image, and I ' (i, j) is I ' (m, n) the interior pixel of 5 * 5 neighborhoods on every side, and I (m, n) is the pixel value after processing; M, n, i and j are natural number.

Then the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field.Its method can adopt the algorithm that is same as reference picture, and the target image gradient field that obtains is defined as G2.

Above-mentioned steps S1 and S2 are used for obtaining reference picture gradient field G1, and step S3 and S4 are used for obtaining target image gradient field G2, and these two operating process do not have the requirement of sequencing, can obtain first target image gradient field G2, obtain reference picture gradient field G1 again.

Described step S5 carries out calculus of differences to reference picture gradient field and target image gradient field, obtains the final image gradient field, i.e. final image gradient field G=G2-G1.

Further, in order to reduce error, can do a linear transformation to the reference picture gradient field first, obtain linear transformation reference picture gradient field M=A * G1+B, carry out again calculus of differences, i.e. final image gradient field G=G2-M.

The value of A, B obtains by manual adjustments, determines the best value of A, B according to the image effect that finally calculates.Wherein, the value of A is near 1, and the suggestion interval is [0.5,1.5], and the value of B is near 0, and the suggestion interval is [0.2G1,0.2G1].

At last, described step S6 is the final image that obtains removing ghost by described final image gradient field G.In the present embodiment, obtain final image by the least square coupling by final image gradient field G, its specific algorithm is as follows:

$F(\▿I,G)={|\▿I-G|}^2={(\frac{\∂I}{\∂x}-G_x)}^2+{(\frac{\∂I}{\∂y}-G_y)}^2;$

Make following formula get minimum value, then must satisfy:

$\frac{\∂F}{\∂I}-\frac{d}{\mathrm{dx}}\frac{\∂F}{{\∂I}_x}-\frac{d}{\mathrm{dy}}\frac{\∂F}{{\∂I}_y}=0;$

Substitution F:

$\frac{{\∂}^{2}I}{{\∂x}^2}+\frac{{\∂}^{2}I}{{\∂y}^2}=\frac{{\∂G}_x}{\∂x}+\frac{{\∂G}_y}{\∂y};$

Namely obtain: ${\▿}^{2}I=\mathrm{div}G;$

With above-mentioned Poisson equation discretize, adopt finite-difference approximation again, obtain:

${\▿}^{2}I\≈I(i+1,j)+I(i-1,j)+I(i,j-1)+I(i,j+1)-4I(i,j);$

divG＝G _x(i，j)-G _x(i-1，j)+G _y(i，j)-G _y(i，j-1)；

Because the least square coupling belongs to optimal fitting, therefore the data error of indivedual points can suppose that in the image border luminance difference is divided into 0 to not obviously impact of fitting result, calculates to simplify, that is:

I(-1，y)-I(0，y)＝0；

Adopt at last full multi grid or method of conjugate gradient to find the solution, can obtain removing the final image of ghost.Remove ghost by above-mentioned gradient domain transformation, calculus of differences, obtain correcting rear image by the least square method coupling again, thereby alleviated the ghost problem that flat panel detector brings.

In addition, the present invention also provides a kind of image processing system of eliminating ghost, and as shown in Figure 2, described system comprises:

Reference picture acquisition module 100, the time shutter of being scheduled to for basis gather a sub-picture as the reference image;

Reference picture processing module 200, be used for reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

Target image acquisition module 300, the time shutter of being scheduled to for basis gather target image;

Target image processing module 400, be used for target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

Difference block 500, be used for reference picture gradient field and target image gradient field are done poor, obtain the final image gradient field;

Final image acquisition module 600, be used for obtaining final image according to described final image gradient field.

The reference picture acquisition module 100 of the image processing system of above-mentioned elimination ghost, reference picture processing module 200, target image acquisition module 300, target image processing module 400, difference block 500 and final image acquisition module 600 are corresponding with step S1-S6 in the image processing method of above-mentioned elimination ghost respectively, the realization details that it is concrete and algorithm are all described in detail in the image processing method of eliminating ghost, have just repeated no more here.

In sum, the image processing method of elimination ghost of the present invention and system, at first target image and reference picture are transformed to gradient field after, obtain the final image gradient field through difference processing again, obtain correcting rear image by the least square method coupling at last.The image processing method of elimination ghost of the present invention and system can reduce or eliminate dull and stereotyped ghost, make image more clear, more being beneficial to the doctor observes, reduce mistaken diagnosis, simultaneously, compare with the hardware solution of prior art, cost is lower, dispose more flexibly, have more wide market application foreground.

Be understandable that; for those of ordinary skills; can be equal to replacement or change according to technical scheme of the present invention and inventive concept thereof; such as changing boundary filter; adopt other algorithm to obtain removing the final image etc. of ghost, and all these changes or replace the protection domain that all should belong to the appended claim of the present invention.

Claims (8)

1. an image processing method of eliminating ghost is characterized in that, said method comprising the steps of:

A1, predetermined time shutter of basis gather reference picture;

A2, reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

A3, predetermined time shutter of basis gather target image;

A4, target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

A5, reference picture gradient field and target image gradient field are carried out calculus of differences, obtain the final image gradient field;

A6, obtain final image according to described final image gradient field.

2. the image processing method of elimination ghost according to claim 1 is characterized in that, when in the described steps A 2 reference picture being carried out noise reduction process, its concrete grammar is as follows:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}|I^{\′}(i,j)-I^{\′}(m,n)|}|I^{\′}(i,j)-I^{\′}(m,n)|I^{\′}(i,j);$

Wherein, I ' (m, n) is pixel pending on the reference picture, and I (i, j) is I ' (m, n) the interior pixel of 5 * 5 neighborhoods on every side, and m, n, i and j are natural number.

3. the image processing method of elimination ghost according to claim 2 is characterized in that, in the described steps A 2 reference picture behind the noise reduction is transformed to the reference picture gradient field, adopts following algorithm:

$G(m,n)=\sqrt{{(I(m,n)-I(m-1,n))}^2+{(I(m,n)-I(m,n-1))}^2};$

Described G (m, n) is the Grad of (m, n) pixel in the reference picture gradient field, and the position that the Grad of all pixels is inserted separately successively just consists of the reference picture gradient field.

4. the method for the ghost of elimination target image according to claim 2 is characterized in that, in the described steps A 2 reference picture behind the noise reduction is transformed to the reference picture gradient field, adopts following algorithm:

G(m，n)＝|I(m，n)-I(m-1，n)|+|I(m，n)-I(m，n-1)|；

Described G (m, n) is the Grad of (m, n) pixel in the reference picture gradient field, and the position that the Grad of all pixels is inserted separately successively just consists of the reference picture gradient field.

5. the image processing method of elimination ghost according to claim 1 and 2 is characterized in that, when in the described steps A 4 target image being carried out noise reduction process, adopts following wave filter:

$I(m,n)=\frac{1}{{\mathrm{\Σ}}_{m-2}^{m+2}{\mathrm{\Σ}}_{n-2}^{n+2}{C}_s{C}_d}{C}_s{C}_{d}I(i,j);$

Wherein, $C_s=e^{-\frac{{(I^{\′}(i,j)-I^{\′}(m,n))}^2}{{2\mathrm{\σ}}_s^2}};$ $C_d=e^{-\frac{{(i-m)}^2+{(j-n)}^2}{{2\mathrm{\σ}}_d^2}};$

σ _sThe similarity distribution standard deviation, σ _dIt is the space distribution standard deviation; I ' (m, n) is pending pixel, and I ' (i, j) is the pixel in 5 * 5 neighborhoods around it; M, n, i and j are natural number.

6. the image processing method of elimination ghost according to claim 1 and 2 is characterized in that, in the described steps A 5 reference picture gradient field and target image gradient field is carried out reference picture being carried out linear transformation before the calculus of differences, and its formula is as follows:

Linear transformation reference picture gradient field=A * reference picture gradient field+B;

Wherein, the interval of A is [0.5,1.5], and the interval of B is [0.2 reference picture gradient field, 0.2 reference picture gradient field].

7. the image processing method of elimination ghost according to claim 1 is characterized in that, it is to obtain final image by the least square method coupling by the final image gradient field that described steps A 6 is obtained final image according to described final image gradient field.

8. an image processing system of eliminating ghost is characterized in that, described system comprises:

Reference picture acquisition module, the time shutter of being scheduled to for basis gather a sub-picture as the reference image;

The reference picture processing module, be used for reference picture is carried out noise reduction process, the reference picture behind the noise reduction is transformed to gradient field, obtain the reference picture gradient field;

Target image acquisition module, the time shutter of being scheduled to for basis gather target image;

The target image processing module, be used for target image is carried out noise reduction process, the target image behind the noise reduction is transformed to gradient field, obtain the target image gradient field;

Difference block, be used for reference picture gradient field and target image gradient field are done poor, obtain the final image gradient field;

The final image acquisition module, be used for obtaining final image according to described final image gradient field.

Images