CN104007087A - Counting method for gold nanomaterial on transparent levelled sheet-shaped substrate surface - Google Patents

Abstract

The invention relates to a counting method for a gold nanomaterial on a transparent levelled sheet-shaped substrate surface. The counting method comprises the following steps: adding a solvent used in sample determination to the transparent levelled sheet-shaped substrate surface, then placing under a dark-field microscope, observing, and acquiring to obtain a background scattering image; adding the gold nanomaterial to the transparent levelled sheet-shaped substrate surface, then placing under the dark-field microscope, observing, and acquiring to obtain a characteristic scattering image of the gold nanomaterial; respectively adding to-be-detected samples containing different concentrations of the gold nanomaterial to the transparent levelled sheet-shaped substrate surface, then placing under the dark-field microscope, observing, and acquiring to obtain a plurality of scattering images of the to-be-detected samples; and successively carrying out image analysis on the background scattering image, the characteristic scattering image of the gold nanomaterial and the scattering images of the to-be-detected samples, and thus obtaining the number of the gold nanomaterial in the scattering images of the to-be-detected samples. The counting method can be widely applied in counting of the gold nanomaterial on various transparent levelled substrates and having various structures and sizes.

Description

The gold nano-material method of counting on a kind of transparent smooth sheet form base surface

Technical field

The present invention relates to image processing field, particularly about the gold nano-material method of counting on a kind of transparent smooth sheet form base surface.

Background technology

At present, the gold nano-material such as spherical nanoparticle, nanometer rods is widely used in chemistry and bioanalysis detection.Because gold nano-material has surface plasma body resonant vibration character, its scattering spectrum has strong, regulatable feature scattering peak, and therefore gold nano-material can detect for the light scattering biochemical analysis on transparent smooth sheet form base (as glass, organic glass, plastics etc.) surface.Existing biochemical analysis detection method adopts directly combination or sandwich sandwich method principle, by being modified with the gold nano-material absorption of identification molecule or being bonded to substrate surface, then by the intensity that detects scattered light signal, determinand is carried out to qualitative, quantitative.Be subject to the impact of the weak and substrate surface parasitic light of the scattered light intensity of single gold nano grain, existing method is not high for the response sensitivity of gold nano-material optical signalling, cannot realize the quantitative of single nano material, need the subsequent treatment such as silver dyes to amplify the signal of gold nano-material more, thereby make corresponding analytical approach detectability limited, experimental procedure and complexity increase.

Summary of the invention

For the problems referred to above, the object of this invention is to provide a kind of gold nano-material method of counting that can realize the transparent smooth sheet form base surface that single gold nano-material is automatically identified and counted

For achieving the above object, the present invention takes following technical scheme: the gold nano-material method of counting on a kind of transparent smooth sheet form base surface, it comprises the following steps: during 1) by sample determination, solvent for use adds to transparent smooth sheet form base surface, then be placed under dark field microscope and observe, collect backscatter image; 2) gold nano-material is added to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the feature dispersion image of gold nano-material; 3) testing sample that contains variable concentrations gold nano-material is added to respectively to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the dispersion image of some testing samples; 4) to step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains carry out successively graphical analysis, obtain the quantity of gold nano-material in testing sample dispersion image.

Described step 1), step 2) and step 3) in, transparent smooth sheet form base adopts the one in transparent inorganic material and transparent polymer material, transparent inorganic material adopts the one in simple glass and quartz glass, and transparent polymer material adopts the one in dimethyl silicone polymer, organic glass and plastics.

Described step 2) and step 3) in; gold nano-material adopts the aggregation of gold copper-base alloy particle or gold copper-base alloy particle; gold copper-base alloy particle adopts the one in spherical gold nano grain, gold nanorods and gold nanoshell, and the size of gold copper-base alloy particle and gold copper-base alloy particle aggregate is 10nm～1000nm.

Described step 1), step 2) and step 3) in, the time shutter while gathering details in a play not acted out on stage, but told through dialogues dispersion image is 10ms～1000ms; In the details in a play not acted out on stage, but told through dialogues dispersion image collecting, a size corresponding to pixel is less than or equal to 0.5 μ m × 0.5 μ m.

Described step 4) in, dispersion image is carried out to graphical analysis, it specifically comprises the following steps: I, color space conversion; Respectively by step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains be converted to L*A*B* color space from sRGB color space, correspondingly respectively obtain the image in L*A*B* color space; II, luminance threshold computing; In the L*A*B* color space that step I is obtained, all pixels of image are carried out luminance threshold computing, obtain non-background pixel and background pixel; III, image are cut apart; The non-background pixel that adopts four floodings that are communicated with to obtain step II is cut apart, and obtains all four subregion lists that are communicated with; Give up and comprise non-background pixel quantity and be less than 9 subregion, calculate area and the axial ratio of non-background pixel composition figure in each subregion, chosen axis compares in theoretical span and the subregion of area in the theoretical span of gold nano-material imaging area, obtains subimage list; IV, color character extract; Each width subimage in the subimage list of the gold nano-material feature dispersion image obtaining to each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III and through the processing of step I～step III carries out color character extraction; Gold nano-material counting in V, testing sample dispersion image; For each subimage in each subimage or the list of standard model dispersion image subimage in the testing sample dispersion image subimage list obtaining through the processing of step I～step III, L*, the A* of each the non-background pixel color in subimage and the value of B* are rounded to integer, search corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell; If the numerical value of cell is 1, non-background pixel is gold nano passage pixel; In calculating subimage, gold nano passage pixel accounts for the ratio of non-background pixel, if this ratio is greater than 50%, this subimage is gold nano-material subimage; Calculate the quantity of all gold nano-material subimages, this quantity is the gold nano-material quantity in testing sample dispersion image or standard model dispersion image.

In described step III, adopt the axial ratio of the method calculating pixel composition figure of spatial moment, its detailed process for: the center of gravity of (I) calculating pixel institute composition diagram picture is:

$\left\{\begin{array}{c}\stackrel{\_}{x}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\mathrm{jF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\\ \stackrel{\_}{y}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\mathrm{kF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\end{array}\right.,$

In formula, J and K are respectively width and the height of figure that pixel forms, and F (j, k) is the value that point (j, k) is located.Wherein, for background pixel, the value that point (j, k) is located is 0; For non-background pixel, the value that point (j, k) is located is 1; (II) calculates major axis eigenvalue λ _mwith minor axis eigenvalue λ _n, it specifically comprises: according to the center of gravity of pixel institute composition diagram picture calculating row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks intersects moment of inertia U (1,1) and is respectively:

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{x})}^{2}F(j,k),$

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{y})}^{2}F(j,k),$

$U\left(\mathrm{1,1}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}(j-\stackrel{\_}{x})(k-\stackrel{\_}{y})F(j,k)$

According to row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks intersection moment of inertia U (1,1), calculate two eigenvalue λ ₁and λ ₂:

$\left\{\begin{array}{c}{\mathrm{\λ}}_1=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]+\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\\ {\mathrm{\λ}}_2=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]-\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\end{array}\right.,$

According to eigenvalue λ ₁and λ ₂, obtain major axis eigenvalue λ _mwith minor axis eigenvalue λ _nbe respectively:

$\left\{\begin{array}{c}{\mathrm{\λ}}_M=\max \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\\ {\mathrm{\λ}}_N=\min \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\end{array}\right.;$

The major axis eigenvalue λ that (III) obtains according to step (II) _mwith minor axis eigenvalue λ _n, the axial ratio R that obtains pixel composition figure is:

$R=\frac{{\mathrm{\λ}}_N}{{\mathrm{\λ}}_M};$

(IV) according to formula (4), the span of the axial ratio R of pixel composition figure be (0,1]; When pixel composition figure is circular, axial ratio R value is 1.

In described step III, according to far-field optics imaging resolution formula, the theoretical span of reference axis ratio and the theoretical span of gold nano-material imaging area, its computation process is: gold nano-material is considered as ellipse by (I), the major axis that this is oval and minor axis length are respectively a and b, in conjunction with the numerical aperture NA of light wavelength λ and micro objective, calculate the ideally image axial ratio R' of gold nano-material after microscope imaging and be:

$R^{\′}=\frac{2b+r}{2a+r},$

In formula, r is microscopical optical resolution, deviation between (II) setting axial ratio of subimage and the axial ratio of ideal image causes the uncertainty of a marginal existence m pixel of subimage, it is the uncertainty that the major axis of subimage or minor axis exist 2m pixel, and subimage entirety produces the broadening of n pixel, and the theoretical span of axial ratio is:

$[\frac{2b+r-(2m-n)l}{2a+r+(2m+n)l},\frac{2b+r+(2m+n)l}{2a+r+(2m+n)l}],$

In formula, l is the foursquare actual length of side corresponding to single pixel; (III) gold nano-material imaging area A is:

$A=\mathrm{\π}(a+\frac{r}{2})(b+\frac{r}{2}),$

The uncertainty of the marginal existence m pixel based on subimage and subimage entirety produce the broadening of n pixel, and the theoretical span of gold nano-material imaging area is:

in formula, expression rounds up.

In described step IV, the detailed process that color character extracts is: I) for each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III, calculate respectively L*, the A* of all non-background pixels and the color mean value of B* color channel, obtain the average color of subimage, use CIEDE2000 aberration algorithm, the difference in calculating subimage between all non-background pixels and average color, the span of default color difference threshold is 10～30, and the pixel quantity that calculated difference is less than color difference threshold accounts for the ratio of non-background pixel, obtains the coverage rate of average color, if the coverage rate of average color is greater than 50%, the average color of subimage is added in the feature color list of backscatter image, II) for each the width subimage in the subimage list of the gold nano-material feature dispersion image obtaining through the processing of step I～step III, by step I) in method, calculate the coverage rate of average color and average color, if the coverage rate of average color is less than 100%, calculate respectively in all non-background pixel of subimage of gold nano-material feature dispersion image, difference is more than or equal to color difference threshold and brightness is greater than L*, the A* of the pixel of average color brightness and the color mean value of B* color channel, obtains the average color of highlight area, use CIEDE2000 aberration algorithm, the pixel that the arbitrary difference of average color of calculating and average color or highlight area is less than color difference threshold accounts for the ratio of non-background pixel, if this ratio is greater than 50%, the average color of subimage of gold nano-material feature dispersion image and the average color of highlight area are added to the list of gold nano-material dispersion image feature color, III) for the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color, use CIEDE2000 aberration algorithm, difference in the feature color of calculating gold nano-material and the list of backscatter characteristics of image color between each background characteristics color, if there is arbitrary difference to be less than color difference threshold, the feature color of gold nano-material is removed from the list of gold nano-material dispersion image feature color, if all differences are all more than or equal to color difference threshold, by the L* of the feature color of gold nano-material, the numerical value of A* and B* is all rounded to integer, IV) set up gold nano-material feature color look-up table, the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color is put into look-up table, the circumstances in which people get things ready for a trip of going forward side by side enlargement of color photos exhibition.

Described step IV) in, set up gold nano-material feature color look-up table, the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color is put into look-up table, the circumstances in which people get things ready for a trip of going forward side by side enlargement of color photos exhibition, it specifically comprises the following steps: (I) sets up the table that comprises L*, A* and tri-dimensions of B*, and using this table as gold nano-material feature color look-up table, wherein the span of L* is [0,100], the span of A* is [86,98], and the span of B* is [108,94], the change step of L*, A* and B* is 1; The numerical value of all cells in gold nano-material feature color look-up table is all preset as to 0; (II), for each color-values in the list of gold nano-material dispersion image feature color, is set to 1 by the numerical value of corresponding (L*, A*, B*) cell in gold nano-material feature color look-up table; The span that (III) presses L* is [35,100], and the span of A* is [86,98], and the span of B* is [1,94] traversal gold nano-material feature color look-up table; For the each (L* in ergodic process, A*, B*) color-values, is used CIEDE2000 aberration algorithm, calculates each (L*, A*, B*) difference between each the gold nano-material feature color in color-values and the list of gold nano-material dispersion image feature color, if there is arbitrary difference to be less than color difference threshold, by corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell is set to 1.

The present invention is owing to taking above technical scheme, it has the following advantages: 1, the present invention's solvent for use when by sample determination, gold nano-material and the testing sample that contains variable concentrations gold nano-material add to respectively transparent smooth sheet form base surface, and be placed in and under dark field microscope, gather corresponding dispersion image, the dispersion image obtaining is carried out to graphical analysis successively, obtain the quantity of gold nano-material in testing sample dispersion image, therefore the present invention can differentiate the dispersion image of single gold nano-material, compared with the existing method that detects gold nano-material entirety scattered light intensity, sensitivity of the present invention is greatly improved.2, the present invention is because the steps such as the conversion of employing color space, luminance threshold computing, image are cut apart, color character extraction are carried out automatic analysis to dispersion image, get rid of from shape and color two aspects of image the scattered signal that non-gold nano-material produces, therefore the present invention is compared with existing method of counting, can increase substantially the signal to noise ratio (S/N ratio) of image, thereby reduce detectability.3, when the present invention analyzes dispersion image, by the feature dispersion image of backscatter image and gold nano-material as a reference, the color character of gold nano-material scattered light is carried out to automatic calibration, avoid manual intervention, improved the robustness of Automated Image Analysis method.Based on above advantage, the present invention can be widely used in the counting of gold nano-material of multiple transparent smooth substrate and various structures and size.

Brief description of the drawings

Fig. 1 is overall flow schematic diagram of the present invention;

Fig. 2 is the result schematic diagram of the details in a play not acted out on stage, but told through dialogues dispersion image of the present invention's gold nano grain that is 50nm to particle diameter; Wherein, figure (a) is that particle diameter is that the gold nano grain of 50nm is at the details in a play not acted out on stage, but told through dialogues dispersion image on amination cover glass surface, figure (b) is the luminance channel of details in a play not acted out on stage, but told through dialogues dispersion image, figure (c) is high pass filter luminance channel after treatment, and figure (d) is high pass filter result;

Fig. 3 is that particle diameter of the present invention is that the details in a play not acted out on stage, but told through dialogues dispersion image of gold nano grain of 50nm is through high pass filter brightness histogram after treatment; Wherein, figure (a) is high pass filter luminance channel histogram after treatment, and figure (b) is 3 luminance channel histograms after level and smooth;

Fig. 4 is that the color character of backscatter image of the present invention and the particle diameter gold nano grain dispersion image that is 50nm extracts result schematic diagram, wherein, figure (a) is the subimage of backscatter image, figure (b) is the average color of the subimage of backscatter image, figure (c) is that the average color of subimage of backscatter image is for the coverage of backscatter image, figure (d) is the subimage of gold nano-material dispersion image, figure (e) is the average color of the subimage of gold nano-material dispersion image, figure (f) is the coverage of the average color of subimage of the gold nano-material dispersion image subimage to gold nano-material dispersion image, figure (g) is the average color of the highlight area of the subimage of gold nano-material dispersion image, figure (h) is the coverage of the average color of the average color of subimage of gold nano-material dispersion image and the highlight area of the subimage of the gold nano-material dispersion image subimage to gold nano-material dispersion image,

Fig. 5 is the result schematic diagram of the details in a play not acted out on stage, but told through dialogues dispersion image of the gold nano grain of the present invention's finishing thrombin aptamer that is 60nm to particle diameter; Wherein, figure (a) is that particle diameter is that the gold nano grain of finishing thrombin aptamer of 60nm is at the details in a play not acted out on stage, but told through dialogues dispersion image on the cover glass surface of finishing thrombin aptamer, figure (b) is the luminance channel of details in a play not acted out on stage, but told through dialogues dispersion image, figure (c) is high pass filter luminance channel after treatment, and figure (d) is high pass filter result;

Fig. 6 is that particle diameter of the present invention is that the details in a play not acted out on stage, but told through dialogues dispersion image of gold nano grain of the finishing thrombin aptamer of 60nm is through high pass filter brightness histogram after treatment; Wherein, figure (a) is high pass filter luminance channel histogram after treatment, and figure (b) is 3 luminance channel histograms after level and smooth.

Embodiment

Below in conjunction with drawings and Examples, the present invention is described in detail.

As shown in Figure 1, the gold nano-material method of counting on the transparent smooth sheet form base of the present invention surface comprises the following steps:

1) during by sample determination, solvent for use adds to transparent smooth sheet form base surface, is then placed under dark field microscope and observes, and collects backscatter image;

2) gold nano-material is added to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the feature dispersion image of gold nano-material;

3) testing sample that contains variable concentrations gold nano-material is added to respectively to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the dispersion image of some testing samples;

4) to step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains carry out successively graphical analysis, obtain the quantity of gold nano-material in testing sample dispersion image.

Above-mentioned steps 1), step 2) and step 3) in, transparent smooth sheet form base adopts the one in transparent inorganic material and transparent polymer material, transparent inorganic material adopts simple glass or quartz glass etc., transparent polymer material adopts PDMS (polydimethylsiloxane, dimethyl silicone polymer), organic glass and plastics etc.

Above-mentioned steps 2) and step 3) in; gold nano-material adopts the aggregation of gold copper-base alloy particle or gold copper-base alloy particle; gold copper-base alloy particle adopts spherical gold nano grain, gold nanorods or gold nanoshell etc., and gold copper-base alloy particle and gold copper-base alloy particle aggregate are of a size of 10nm～1000nm.

Above-mentioned steps 1), step 2) and step 3) in, the time shutter while gathering details in a play not acted out on stage, but told through dialogues dispersion image is 10ms～1000ms; In the details in a play not acted out on stage, but told through dialogues dispersion image collecting, a size corresponding to pixel is less than or equal to 0.5 μ m × 0.5 μ m.

Above-mentioned steps 4) in, to step 1), step 2) and step 3) in the dispersion image that obtains carry out successively graphical analysis, it comprises the steps:

I, color space conversion

Respectively by step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains be converted to L*A*B* color space from sRGB color space, correspondingly respectively obtain the image in L*A*B* color space.

The white point value that color space is changed L*A*B* color space used is International Commission on Illumination defined in international standard ISO10526:1999/CIE S005/E-1998 " CIE standard illuminant D65 " color-values, its coordinate figure in CIE XYZ color space is (X, Y, Z)=(0.9505,1.0000,1.0890).

II, luminance threshold computing

In the L*A*B* color space that step I is obtained, all pixels of image are carried out luminance threshold computing, obtain non-background pixel and background pixel, and it comprises the steps:

(I) carries out high-pass filtering to the luminance channel of image in L*A*B* color space, obtains the luminance channel after high-pass filtering, and it specifically comprises:

1. the luminance channel of image in L*A*B* color space is carried out to Gaussian Blur, wherein the span of the parameter σ of Gaussian Blur is [1,5], obtains the luminance channel after Fuzzy Processing.

2. for each brightness value of the luminance channel after Fuzzy Processing, deduct brightness value with 100, obtain the luminance channel after negate.

3. by the added luminance of the correspondence position of the luminance channel of image in each brightness value of the luminance channel after negate and L*A*B* color space and divided by 2, obtain the luminance channel after high-pass filtering.

(II) calculates luminance threshold according to the luminance channel after high-pass filtering, and it specifically comprises:

1. by luminance channel after high-pass filtering (0,100] brightness value in scope is equally divided into 256 grades, draws the brightness histogram of the luminance channel after high-pass filtering.

2. adopt 3 sliding methods to average processing to the count value of brightness histogram, obtain the brightness histogram after level and smooth.

3. in the brightness histogram after adopting smoothly, maximum count value is normalized brightness histogram, obtains the brightness histogram after normalization.

4. taking brightness value maximum in the corresponding brightness value of count value 1 in the brightness histogram after normalization as starting point, taking the difference of adjacent two brightness values in the brightness histogram after normalization as step-length, increase successively brightness value, and adopt the rate of change S of line fitting method zequin to the closed interval inside counting value of current brightness value _i(i=2,3,4 ...); From i=3, increase successively the value of i, the rate of change S of count value ₃to S _imean value A _i, current count value rate of change S _irate of change S with previous count value _i-1difference D _iand mean value A _irate of change S with count value _idifference V _i; In the process increasing at i, select first difference D _i> 0 and difference V _ithe i value of > 0, using the corresponding brightness value of this i value as luminance threshold.

The luminance threshold that the brightness value of each pixel in the luminance channel after the high-pass filtering that (III) obtains step (I) and step (II) obtain compares.

If the pixel brightness value in the luminance channel after high-pass filtering is greater than luminance threshold, using the pixel in L*A*B* color space the image corresponding pixel brightness value in the luminance channel after high-pass filtering as non-background pixel.If the brightness value of non-background pixel is less than 25, this non-background pixel is set to background pixel, and this background pixel is set to ater.

If the pixel brightness value in the luminance channel after high-pass filtering is less than or equal to luminance threshold, by the pixel pixel as a setting in L*A*B* color space the image corresponding pixel brightness value in the luminance channel after high-pass filtering, and this background pixel is set to ater.

III, image are cut apart

The non-background pixel that adopts four floodings that are communicated with to obtain step II is cut apart, and obtains all four subregion lists that are communicated with; Give up and comprise non-background pixel quantity and be less than 9 subregion, calculate area and the axial ratio of non-background pixel composition figure in each subregion, chosen axis compares in theoretical span and the subregion of area in the theoretical span of gold nano-material imaging area, obtains subimage list.

IV, color character extract

Each width subimage in the subimage list of the gold nano-material feature dispersion image obtaining to each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III and through the processing of step I～step III carries out color character extraction, and it specifically comprises:

I) for each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III, calculate respectively L*, the A* of all non-background pixels and the color mean value of B* color channel, obtain the average color of subimage; Use CIEDE2000 aberration algorithm, the difference in calculating subimage between all non-background pixels and average color; The span of default color difference threshold is 10～30, and the pixel quantity that calculated difference is less than color difference threshold accounts for the ratio of non-background pixel, obtains the coverage rate of average color; If the coverage rate of average color is greater than 50%, the average color of subimage is added in the feature color list of backscatter image.

II) for each the width subimage in the subimage list of the gold nano-material feature dispersion image obtaining through the processing of step I～step III, by step I) in method, calculate the coverage rate of average color and average color.

If the coverage rate of average color is less than 100%, calculate respectively in all non-background pixel of subimage of gold nano-material feature dispersion image, difference is more than or equal to color difference threshold and brightness is greater than L*, the A* of the pixel of average color brightness and the color mean value of B* color channel, obtains the average color of highlight area; Use CIEDE2000 aberration algorithm, the pixel that the arbitrary difference of average color of calculating and average color or highlight area is less than color difference threshold accounts for the ratio of non-background pixel, if this ratio is greater than 50%, the average color of subimage of gold nano-material feature dispersion image and the average color of highlight area are added to the list of gold nano-material dispersion image feature color.

III) for the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color, use CIEDE2000 aberration algorithm, difference in the feature color of calculating gold nano-material and the list of backscatter characteristics of image color between each background characteristics color, if there is arbitrary difference to be less than color difference threshold, the feature color of gold nano-material is removed from the list of gold nano-material dispersion image feature color, if all differences are all more than or equal to color difference threshold, by the L* of the feature color of gold nano-material, the numerical value of A* and B* is all rounded to integer.

IV) set up gold nano-material feature color look-up table, the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color is put into look-up table, the circumstances in which people get things ready for a trip of going forward side by side enlargement of color photos exhibition, it specifically comprises the following steps:

(I) sets up the table that comprises L*, A* and tri-dimensions of B*, and using this table as gold nano-material feature color look-up table, wherein the span of L* is [0,100], the span of A* is [86,98], the span of B* is [108,94], and the change step of L*, A* and B* is 1; The numerical value of all cells in gold nano-material feature color look-up table is all preset as to 0.

(II), for each color-values in the list of gold nano-material dispersion image feature color, is set to 1 by the numerical value of corresponding (L*, A*, B*) cell in gold nano-material feature color look-up table.

The span that (III) presses L* is [35,100], and the span of A* is [86,98], and the span of B* is [1,94] traversal gold nano-material feature color look-up table; For the each (L* in ergodic process, A*, B*) color-values, is used CIEDE2000 aberration algorithm, calculates each (L*, A*, B*) difference between each the gold nano-material feature color in color-values and the list of gold nano-material dispersion image feature color, if there is arbitrary difference to be less than color difference threshold, by corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell is set to 1.

Gold nano-material counting in V, testing sample dispersion image

For each subimage in each subimage or the list of standard model dispersion image subimage in the testing sample dispersion image subimage list obtaining through the processing of step I～step III, L*, the A* of each the non-background pixel color in subimage and the value of B* are rounded to integer, search corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell; If the numerical value of cell is 1, non-background pixel is gold nano passage pixel; In calculating subimage, gold nano passage pixel accounts for the ratio of non-background pixel, if this ratio is greater than 50%, this subimage is gold nano-material subimage.Calculate the quantity of all gold nano-material subimages, this quantity is the gold nano-material quantity in testing sample dispersion image or standard model dispersion image.

In above-mentioned steps III, adopt the axial ratio of the method calculating pixel composition figure of spatial moment, its detailed process is:

The center of gravity of (I) calculating pixel institute composition diagram picture is:

$\left\{\begin{array}{c}\stackrel{\_}{x}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\mathrm{jF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\\ \stackrel{\_}{y}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\mathrm{kF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\end{array}\right.---\left(1\right)$

In formula, J and K are respectively width and the height of figure that pixel forms, and F (j, k) is the value that point (j, k) is located.Wherein, for background pixel, the value that point (j, k) is located is 0; For non-background pixel, the value that point (j, k) is located is 1.

(II) calculates major axis eigenvalue λ _mwith minor axis eigenvalue λ _n, it specifically comprises:

According to the center of gravity of pixel institute composition diagram picture calculating row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks intersects moment of inertia U (1,1) and is respectively:

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{x})}^{2}F(j,k),$

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{y})}^{2}F(j,k),$

$U\left(\mathrm{1,1}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}(j-\stackrel{\_}{x})(k-\stackrel{\_}{y})F(j,k)$

According to row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks intersection moment of inertia U (1,1), calculate two eigenvalue λ ₁and λ ₂:

$\left\{\begin{array}{c}{\mathrm{\λ}}_1=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]+\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\\ {\mathrm{\λ}}_2=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]-\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\end{array}\right.---\left(2\right)$

According to eigenvalue λ ₁and λ ₂, obtain major axis eigenvalue λ _mwith minor axis eigenvalue λ _nbe respectively:

$\left\{\begin{array}{c}{\mathrm{\λ}}_M=\max \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\\ {\mathrm{\λ}}_N=\min \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\end{array}\right.---\left(3\right)$

The major axis eigenvalue λ that (III) obtains according to step (II) _mwith minor axis eigenvalue λ _n, the axial ratio R that obtains pixel composition figure is:

$R=\frac{{\mathrm{\λ}}_N}{{\mathrm{\λ}}_M}---\left(4\right)$

(IV) according to formula (4), the span of the axial ratio R of pixel composition figure be (0,1].Wherein, when pixel composition figure is for circle, axial ratio R value is 1.

In above-mentioned steps III, according to far-field optics imaging resolution formula, the theoretical span of reference axis ratio and the theoretical span of gold nano-material imaging area, its computation process is:

Gold nano-material is considered as ellipse by (I), the major axis that this is oval and minor axis length are respectively a and b, in conjunction with the numerical aperture NA of light wavelength λ and micro objective, calculate the ideally image axial ratio R' of gold nano-material after microscope imaging and be:

$R^{\′}=\frac{2b+r}{2a+r}---\left(5\right)$

In formula, r is microscopical optical resolution,

(II) due to the high-pass filtering of the rasterisation in CCD imaging process and image processing process with in step III, may cause subimage different compared with image ideally, especially high-pass filtering can cause subimage scope broadening to some extent, therefore have deviation compared with the axial ratio of subimage and the axial ratio of ideal image.

Set the uncertainty that above-mentioned deviation causes a marginal existence m pixel of subimage, there is the uncertainty of 2m pixel in the major axis of subimage or minor axis, and the broadening of n pixel of subimage entirety generation, and the theoretical span of axial ratio is:

$[\frac{2b+r-(2m-n)l}{2a+r+(2m+n)l},\frac{2b+r+(2m+n)l}{2a+r+(2m+n)l}]---\left(6\right)$

In formula, l is the foursquare actual length of side corresponding to single pixel.

(III) gold nano-material imaging area A is:

$A=\mathrm{\π}(a+\frac{r}{2})(b+\frac{r}{2})---\left(7\right)$

The uncertainty of a marginal existence m pixel and the broadening of n pixel of subimage entirety generation of considering subimage, the theoretical span of gold nano-material imaging area is:

In formula, l is the foursquare actual length of side corresponding to single pixel, expression rounds up.

Embodiment 1: the imaging time of backscatter image, gold nano-material feature dispersion image and testing sample dispersion image is 620ms, physical size corresponding to each pixel in image is 0.15 μ m × 0.15 μ m, and the size of single image is 4140 × 3096 pixels.Adopt method of counting of the present invention, measure the quantity of the gold nano grain of amination surface of glass slide absorption, it specifically comprises the following steps:

1) 5 μ L deionized waters are dropped to microscope slide surface, cover amidized cover glass, be placed under dark field microscope and observe, collect backscatter image.

2) the spherical gold nano particle colloidal sols that is 50nm by particle diameter drops to amidized cover glass surface, leaves standstill after 5min, dries up with deionized water rinsing cover glass and with nitrogen.5 μ L deionized waters are dropped to microscope slide surface, cover the cover glass after drying up, be placed under dark field microscope and observe, collect gold nano-material feature dispersion image.

3) testing sample is the colloidal sol of the particle diameter that the contains variable concentrations spherical gold nano grain that is 50nm, measure respectively 50～100 μ L testing samples and drop to amidized cover glass surface, leave standstill after 5min, dry up with deionized water rinsing cover glass and with nitrogen.5 μ L deionized waters are dropped to microscope slide surface, cover the cover glass after drying up, be placed under dark field microscope and observe, collect the dispersion image of testing sample.

4) to step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains carry out successively graphical analysis, the quantity that obtains gold nano-material in testing sample dispersion image, it specifically comprises:

I, color space conversion

Adopt CIE standard illuminant D65 white point value, respectively by step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains be converted to L*A*B* color space from sRGB color space, correspondence obtains L*A*B* color space image respectively.

II, luminance threshold computing

The L*A*B* color space image that step I is obtained carries out the luminance threshold computing based on high pass filter, brightness in image is changed to less interval and be set to ater, obtain non-background pixel and background pixel, be convenient to successive image analysis, it specifically comprises the following steps:

(I) as shown in Figure 2, it is 3 Gaussian Blur that the luminance channel (as shown in figure (b)) of L*A*B* color space image that step I is obtained (as shown in figure (a)) is carried out parameter σ; For each brightness value of the luminance channel after Fuzzy Processing, deduct this brightness value with 100, obtain the brightness value after negate; After the added luminance of the correspondence position pixel of the L*A*B* color space image that the brightness value after negate and step I are obtained divided by 2, that is ask for brightness value after anti-and the average brightness value of former brightness value, obtain the luminance channel (as shown in figure (c)) after high-pass filtering.

(II) as shown in Figure 3, by (0 of luminance channel after high-pass filtering, 100] brightness value in scope is equally divided into 256 grades, be that luminance difference between every adjacent two-stage is 0.3906, statistics obtains the brightness histogram (as shown in figure (a)) of the luminance channel after high-pass filtering.Adopt 3 slips to average processing to brightness histogram, obtain the brightness histogram (as shown in figure (b)) after level and smooth.In brightness histogram after adopting smoothly, maximum count value 9534 is normalized brightness histogram, obtains the brightness histogram after normalization.Taking brightness value 49.80 maximum in the corresponding brightness value of count value 1 in the brightness histogram after normalization as starting point, taking the luminance difference 0.3906 between adjacent two-stage as step-length, increase successively brightness value, and adopt the rate of change S of line fitting method zequin to the closed interval inside counting value of current brightness value _i(i=2,3,4 ...); From i=3, increase successively the value of i, the rate of change S of count value ₃to S _imean value A _i, current count value rate of change S _irate of change S with previous count value _i-1difference D _iand mean value A _irate of change S with count value _idifference V _i, result of calculation is as shown in table 1; In the process increasing at i, select first difference D _i> 0 and difference V _ithe i value 5 of > 0, using corresponding this i value 5 brightness value 51.37 as luminance threshold.

The selection reckoner of luminance threshold in the operation of table 1 high pass filter

i	L*	S i	A i	D i	V i
						1	49.80	?	?	?	?
2	50.20	-0.07601	?	?	?
						3	50.59	-0.3146	-0.1953	-0.2386	0.1193
4	50.98	-0.3462	-0.2456	-0.03161	0.1006
						5	51.37	-0.2863	-0.2559	0.05995	0.03049
6	51.76	-0.2289	-0.2504	0.05737	-0.02151
						7	52.16	-0.1841	-0.2393	0.04478	-0.05524

Each brightness value and luminance threshold in luminance channel after high-pass filtering are compared, if brightness value is greater than luminance threshold, pixel in the L*A*B* color space image that brightness value is corresponding is non-background pixel, and all brightness values in non-background pixel are less than to 25 pixel are made as ater; If brightness value is less than or equal to luminance threshold, the pixel in the L*A*B* color space image that brightness value is corresponding is background pixel, and background pixel is set to ater (as shown in Fig. 2 (d)).

III, image are cut apart

Cut apart by image, the original image that comprises multiple continuous shapes can be converted to a series of subimages that only comprise single continuous shape, thereby be convenient to shape recognition.The non-background pixel that adopts four floodings that are communicated with to obtain step II is cut apart, and obtains all four subregion lists that are communicated with; Give up the impact with the noise getting rid of imaging process and may introduce of subimage that all non-background pixel quantity is less than 9 (3 × 3); For each subimage, area and the axial ratio of calculating pixel composition figure, and according to used gold nano-material pattern and size, screening meets the subimage of gold nano-material shape characteristic.

Adopt the axial ratio of the method calculating pixel composition figure of spatial moment.Adopt the theoretical span of far-field optics imaging resolution formula reference axis ratio and the theoretical span of gold nano-material imaging area.In the present embodiment, microscopical numerical aperture used is: NA=0.47, the color of image of gold nano-material is green to red, corresponding optical wavelength is 500nm～700nm, the uncertainty of image border is 1 pixel, and image broadening is 1 pixel, the theoretical span of the axial ratio of the image that spherical gold nano grain becomes that particle diameter is 50nm is: [0.5,1], the theoretical span of gold nano-material imaging area is [13,75].

IV, color character extract

Gold nano-material has the scattering properties that concentrates on specific band because of its peculiar surface plasmon resonance effect, thereby is scattering in picture and is shown as feature color in details in a play not acted out on stage, but told through dialogues, and this feature color changes with the pattern of gold nano-material and the variation of particle diameter.

The gold nano-material that the particle diameter that the present embodiment uses is 10nm～1000nm, its feature color comprise as particle diameter be less than the green of the spherical gold nano grain performance of 70nm, as particle diameter in the yellow of the spherical gold nano grain performance of 70nm～100nm and as length-diameter ratio be greater than 2 gold nanorods, the redness that spherical gold nano grain and the gold nanoshell of composite structure etc. of particle diameter more than 100nm shows.And other impurity particle in flaw or detection system on slide has comparatively average scattering properties in all band of visible ray conventionally, thereby be scattering into and in picture, be shown as white or light blue in details in a play not acted out on stage, but told through dialogues.Therefore can be according to gold nano-material and impurity the feature on color, gold nano-material and impurity are distinguished.

For simplifying the color description to subimage, adopt the average color of subimage as the color character of subimage, L*, the A* that the average color of subimage is all non-background pixels and the color mean value of B* color channel.As shown in Figure 4, because the subimage (as shown in figure (a)) of backscatter image is generally white or light blue, and Luminance Distribution is comparatively concentrated, therefore use average color (as shown in figure (b)) can cover preferably the most of pixel (as shown in figure (c)) in subimage.And the subimage of gold nano-material dispersion image (as shown in figure (d)) can have larger Luminance Distribution, now average color (as shown in figure (e)) only can cover the partial pixel (as shown in figure (f)) of low brightness area.The calculating (as shown in figure (g)) that therefore need to again average color for the highlight area of gold nano-material dispersion image is to embody the color character (as shown in figure (h)) of highlight area in gold nano-material dispersion image.

For calculating the representative degree of average color for subimage, adopt CIEDE2000 aberration algorithm, difference in calculating subimage between each non-background pixel and average color, if difference is less than color difference threshold 15, thinks that this non-background pixel is consistent with average color.The pixel quantity that calculated difference is less than color difference threshold accounts for the ratio of non-background pixel, obtains the coverage rate of average color.

For the subimage of backscatter image, if the coverage rate of average color is greater than 50%, average color is added to the list of backscatter characteristics of image color.For gold nano-material feature dispersion image, if total coverage rate of average color and high light average color is greater than 50%, average color and high light average color are added to the list of gold nano-material dispersion image feature color.Owing to also may there being the background image causing because of other impurity particle in slide flaw or detection system in gold nano-material feature dispersion image, therefore the color list of gold nano-material dispersion image feature and the list of backscatter characteristics of image color need to be compared, to remove the actual feature color that belongs to background image in the list of gold nano-material dispersion image feature color.

For each the feature color in the list of gold nano-material dispersion image feature color, use CIEDE2000 aberration algorithm, calculate the difference between each background characteristics color in this feature color and the list of backscatter characteristics of image color, if there is arbitrary difference to be less than color difference threshold 15, the feature color of this gold nano-material is removed from the list of gold nano-material dispersion image feature color; If all differences are all more than or equal to color difference threshold 15, the numerical value of L*, the A* of the feature color of gold nano-material and B* is all rounded to integer.For example, L*, A* and the B* of the feature color of gold nano-material are respectively 31.3,22.5,46.6, be respectively 31,23,47, thereby be convenient to subsequent treatment through rounding up after processing.

Getting rid of in gold nano-material dispersion image feature color after the actual feature color that belongs to background image, for ease of the subimage of testing sample dispersion image being carried out to color judgement fast, need set up gold nano-material feature color look-up table, put into look-up table by the feature color of each gold nano-material and with the color that the aberration of this feature color is less than color difference threshold, its detailed process is:

The table that foundation comprises L*, A* and tri-dimensions of B*, and using this table as gold nano-material feature color look-up table, wherein the span of L* is [0,100], the span of A* is [86,98], the span of B* is [108,94], and the change step of L*, A* and B* is 1; The numerical value of all cells in gold nano-material feature color look-up table is all preset as to 0.

For each the feature color-values in the list of gold nano-material dispersion image feature color, the numerical value of corresponding (L*, A*, B*) cell in gold nano-material feature color look-up table is set to 1.

The span of pressing L* is [35,100], and the span of A* is [86,98], and the span of B* is [1,94] traversal gold nano-material feature color look-up table; For the each (L* in ergodic process, A*, B*) color-values, is used CIEDE2000 aberration algorithm, calculates each (L*, A*, B*) difference between each the gold nano-material feature color in color-values and the list of gold nano-material dispersion image feature color, if there is arbitrary difference to be less than color difference threshold 15, by corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell is set to 1.

Gold nano-material counting in V, testing sample dispersion image

According to step 1) to step 3) method process testing sample dispersion image, obtain meeting the subimage list of gold nano grain appearance and size feature.For each subimage in subimage list, the value of the L* of each non-background pixel, A* and B* is rounded to integer, for example, by (L*, A*, B*) be (31.3,22.5,46.6) round up process after obtain (31,23,47), search the numerical value of corresponding unit lattice in gold nano-material feature color look-up table (31,23,47).If the numerical value of cell is 1, this non-background pixel is gold nano passage pixel.In calculating subimage, gold nano passage pixel accounts for the ratio of non-background pixel, if ratio is greater than 50%, this subimage is gold nano-material subimage.Calculate the quantity of all gold nano-material subimages, this quantity is the gold nano-material quantity in testing sample dispersion image or standard model dispersion image.

The automatic recognition result of all images that same batch of experiment gathered is as table 2.

The gold nano grain image automatic identification result table of table 2 amination surface of glass slide

Sample number into spectrum	Artificial counting result	Auto-counting result	Auto-counting discrimination
				1	176	150	85％
2	173	149	86％
				3	168	148	88％
4	184	158	86％
				Amount to	701	605	85％

As can be seen from Table 2, the present invention can measure the unmodified gold nano grain quantity of amination adsorption, and can reach close discrimination compared with artificial counting method.

Embodiment 2: the imaging time of backscatter image, gold nano-material feature dispersion image and testing sample dispersion image is 620ms, physical size corresponding to each pixel in image is 0.15 μ m × 0.15 μ m, and the size of single image is 4140 × 3096 pixels.Adopt method of counting of the present invention, measure the quantity of the gold nano grain of the surface of glass slide connection that in DNA aptamers-fibrin ferment sandwich method detection system, DNA aptamers is modified, it specifically comprises the following steps:

1) 5 μ L deionized waters are dropped to microscope slide surface, cover the cover glass of finishing thrombin aptamer, be placed under dark field microscope and observe, collect backscatter image.

2) solution that 50 μ L is contained to fibrin ferment sample drops to the cover glass surface of modifying thrombin aptamer, after reaction 30min, uses deionized water rinsing.The spherical gold nano particle colloidal sols of the finishing thrombin aptamer that is 60nm by 50 μ L particle diameters drops to the cover glass surface of modifying thrombin aptamer, after standing 30min, dries with deionized water rinsing cover glass and by the water of cover glass surface attachment.5 μ L deionized waters are dropped to microscope slide surface, and covered, is placed under dark field microscope and observes, and collects gold nano-material feature dispersion image.

3) measure respectively the fibrin ferment sample solution that 50 μ L contain variable concentrations and drop to the cover glass surface of modifying thrombin aptamer, after reaction 30min, using deionized water rinsing.Measure respectively the spherical gold nano particle colloidal sols of the finishing thrombin aptamer that 50 μ L particle diameters are 60nm and drop to the cover glass surface of modifying thrombin aptamer, after standing 30min, dry with deionized water rinsing cover glass and by the water of cover glass surface attachment.5 μ L deionized waters are dropped to microscope slide surface, and covered, is placed under dark field microscope and observes, and collects the dispersion image of testing sample.

4) to step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains carry out successively graphical analysis, the quantity that obtains gold nano-material in testing sample dispersion image, it specifically comprises:

I, color space conversion

Adopt CIE standard illuminant D65 white point value, respectively by step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains be converted to L*A*B* color space from sRGB color space, correspondence obtains L*A*B* color space image respectively.

II, luminance threshold computing

The L*A*B* color space image that step I is obtained carries out the luminance threshold computing based on high pass filter, brightness in image is changed to less interval and be set to ater, obtain non-background pixel and background pixel, be convenient to successive image analysis, it specifically comprises the following steps:

(I) as shown in Figure 5, it is 3 Gaussian Blur that the luminance channel (as shown in figure (b)) of L*A*B* color space image that step I is obtained (as shown in figure (a)) is carried out radius; For each brightness value of the luminance channel after Fuzzy Processing, deduct this brightness value with 100, obtain the brightness value after negate; After the added luminance of the correspondence position pixel of the L*A*B* color space image that the brightness value after negate and step I are obtained divided by 2, that is ask for brightness value after anti-and the average brightness value of former brightness value, obtain the luminance channel (as shown in figure (c)) after high-pass filtering.

(II) as shown in Figure 6, by (0 of luminance channel after high-pass filtering, 100] brightness value in scope is equally divided into 256 grades, be that luminance difference between every adjacent two-stage is 0.3906, statistics obtains the brightness histogram (as shown in figure (a)) of the luminance channel after high-pass filtering.Adopt 3 slips to average processing to brightness histogram, obtain the brightness histogram (as shown in figure (b)) after level and smooth.In brightness histogram after adopting smoothly, maximum count value 11129 is normalized brightness histogram, obtains the brightness histogram after normalization.Taking brightness value 49.80 maximum in the corresponding brightness value of count value 1 in the brightness histogram after normalization as starting point, taking the luminance difference 0.3906 between adjacent two-stage as step-length, increase successively brightness value, and adopt the rate of change S of line fitting method zequin to the closed interval inside counting value of current brightness value _i(i=2,3,4 ...); From i=3, increase successively the value of i, the rate of change S of count value ₃to S _imean value A _i, current count value rate of change S _irate of change S with previous count value _i-1difference D _iand mean value A _irate of change S with count value _idifference V _i, result of calculation is as shown in table 3; In the process increasing at i, select first difference D _i> 0 and difference V _ithe i value 5 of > 0, using corresponding this i value 5 brightness value 51.37 as luminance threshold.

The selection reckoner of luminance threshold in the operation of table 3 high pass filter

i	L*	S i	A i	D i	V i
						1	49.80	?	?	?	?
2	50.20	-0.00081	?	?	?
						3	50.59	-0.1917	-0.09625	-0.1909	0.09544
4	50.98	-0.3157	-0.1694	-0.1240	0.1463
						5	51.37	-0.2894	-0.1994	0.02635	0.08997
6	51.76	-0.2413	-0.2078	0.04805	0.03354
						7	52.16	-0.1986	-0.2062	0.04274	-0.007671

By each brightness value in the luminance channel after high-pass filtering and luminance threshold comparison, if brightness value is greater than luminance threshold, the pixel in its corresponding L*A*B* color space image is non-background pixel, and all brightness values in non-background pixel are less than to 25 pixel are made as ater; If brightness value is less than or equal to luminance threshold, the pixel in the L*A*B* color space image that brightness value is corresponding is background pixel, and background pixel is set to ater (as shown in Fig. 5 (d)).

III, image are cut apart

Cut apart by image, the original image that comprises multiple continuous shapes can be converted to a series of subimages that only comprise single continuous shape, thereby be convenient to shape recognition.The non-background pixel that adopts four floodings that are communicated with to obtain step II is cut apart, and obtains all four subregion lists that are communicated with; Give up the impact with the noise getting rid of imaging process and may introduce of subimage that all non-background pixel quantity is less than 9 (3 × 3); For each subimage, area and the axial ratio of calculating pixel composition figure, and according to used gold nano-material pattern and size, screening meets the subimage of gold nano-material shape characteristic.

Adopt the axial ratio of the method calculating pixel composition figure of spatial moment, adopt the theoretical span of far-field optics imaging resolution formula reference axis ratio and the theoretical span of gold nano-material imaging area.In the present embodiment, microscopical numerical aperture used is: NA=0.47, the color of image of gold nano-material is green to red, corresponding optical wavelength is 500nm～700nm, the uncertainty of image border is 1 pixel, and image broadening is 1 pixel, the theoretical span of the axial ratio of the image that spherical gold nano grain becomes that particle diameter is 60nm is: [0.51,1], the theoretical span of gold nano-material imaging area is [13,79].

IV, color character extract

Gold nano-material has the scattering properties that concentrates on specific band because of its peculiar surface plasmon resonance effect, thereby is scattering in picture and is shown as feature color in details in a play not acted out on stage, but told through dialogues, and this feature color changes with the pattern of gold nano-material and the variation of particle diameter.

The gold nano-material that the particle diameter that the present embodiment uses is 10nm～1000nm, its feature color comprise as particle diameter be less than the green of the spherical gold nano grain performance of 70nm, as particle diameter in the yellow of the spherical gold nano grain performance of 70nm～100nm and as length-diameter ratio be greater than 2 gold nanorods, the redness that spherical gold nano grain and the gold nanoshell of composite structure etc. of particle diameter more than 100nm shows.And other impurity particle in flaw or detection system on slide has comparatively average scattering properties in all band of visible ray conventionally, thereby be scattering into and in picture, be shown as white or light blue in details in a play not acted out on stage, but told through dialogues.Therefore can be according to gold nano-material and impurity the feature on color, gold nano-material and impurity are distinguished.

For simplifying the color description to subimage, adopt the average color of subimage as the color character of subimage, L*, the A* that the average color of subimage is all non-background pixels and the color mean value of B* color channel.As shown in Figure 4, because the subimage of backscatter image is generally white or light blue, and Luminance Distribution is comparatively concentrated, therefore use average color can cover preferably the most of pixel in subimage.And the subimage of gold nano-material dispersion image can have larger Luminance Distribution, now average color only can cover the partial pixel of low brightness area.The calculating that therefore need to again average color for the highlight area of gold nano-material dispersion image is to embody the color character of highlight area in gold nano-material dispersion image.

For calculating the representative degree of average color for subimage, adopt CIEDE2000 aberration algorithm, difference in calculating subimage between each non-background pixel and average color, if difference is less than color difference threshold 15, thinks that this non-background pixel is consistent with average color.The pixel quantity that calculated difference is less than color difference threshold accounts for the ratio of non-background pixel, obtains the coverage rate of average color.

For the subimage of backscatter image, if the coverage rate of average color is greater than 50%, average color is added to the list of backscatter characteristics of image color.For gold nano-material feature dispersion image, if total coverage rate of average color and high light average color is greater than 50%, average color and high light average color are added to the list of gold nano-material dispersion image feature color.Owing to also may there being the background image causing because of other impurity particle in slide flaw or detection system in gold nano-material feature dispersion image, therefore the color list of gold nano-material dispersion image feature and the list of backscatter characteristics of image color need to be compared, to remove the actual feature color that belongs to background image in the list of gold nano-material dispersion image feature color.

For each the feature color in the list of gold nano-material dispersion image feature color, use CIEDE2000 aberration algorithm, calculate the difference between each background characteristics color in this feature color and the list of backscatter characteristics of image color, if there is arbitrary difference to be less than color difference threshold 15, the feature color of this gold nano-material is removed from the list of gold nano-material dispersion image feature color; If all differences are all more than or equal to color difference threshold 15, the numerical value of L*, the A* of the feature color of gold nano-material and B* is all rounded to integer.For example, L*, A* and the B* of the feature color of gold nano-material are respectively 31.3,22.5,46.6, be respectively 31,23,47, thereby be convenient to subsequent treatment through rounding up after processing.

Getting rid of in gold nano-material dispersion image feature color after the actual feature color that belongs to background image, for ease of the subimage of testing sample dispersion image being carried out to color judgement fast, need set up gold nano-material feature color look-up table, put into look-up table by the feature color of each gold nano-material and with the color that the aberration of this feature color is less than color difference threshold, its detailed process is:

The table that foundation comprises L*, A* and tri-dimensions of B*, and using this table as gold nano-material feature color look-up table, wherein the span of L* is [0,100], the span of A* is [86,98], the span of B* is [108,94], and the change step of L*, A* and B* is 1; The numerical value of all cells in gold nano-material feature color look-up table is all preset as to 0.

For each the feature color-values in the list of gold nano-material dispersion image feature color, the numerical value of corresponding (L*, A*, B*) cell in gold nano-material feature color look-up table is set to 1.

The span of pressing L* is [35,100], and the span of A* is [86,98], and the span of B* is [1,94] traversal gold nano-material feature color look-up table; For the each (L* in ergodic process, A*, B*) color-values, is used CIEDE2000 aberration algorithm, calculates each (L*, A*, B*) difference between each the gold nano-material feature color in color-values and the list of gold nano-material dispersion image feature color, if there is arbitrary difference to be less than color difference threshold 15, by corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell is set to 1.

Gold nano-material counting in V, testing sample dispersion image

According to step 1) to step 3) method process testing sample dispersion image, obtain meeting the subimage list of gold nano grain appearance and size feature.For each subimage in subimage list, the value of the L* of each non-background pixel, A* and B* is rounded to integer, for example, by (L*, A*, B*) be (31.3,22.5,46.6) round up process after obtain (31,23,47), search the numerical value of corresponding unit lattice in gold nano-material feature color look-up table (31,23,47).If the numerical value of cell is 1, this non-background pixel is gold nano passage pixel.In calculating subimage, gold nano passage pixel accounts for the ratio of non-background pixel, if ratio is greater than 50%, this subimage is gold nano-material subimage.Calculate the quantity of all gold nano-material subimages, this quantity is the gold nano-material quantity in testing sample dispersion image or standard model dispersion image.

The automatic recognition result of all images that same batch of experiment gathered is as table 4.

The gold nano grain image automatic identification result table of table 4 thrombin aptamer system

Sample number into spectrum	Artificial counting result	Auto-counting result	Auto-counting discrimination
				1	58	49	84％
2	53	44	83％
				3	50	45	90％
4	89	75	84％
				Amount to	250	213	85％

As can be seen from Table 4, the gold nano grain quantity that is modified with blood coagulation enzyme dna aptamers that the surface of glass slide that the present invention can modify DNA aptamers in DNA aptamers-fibrin ferment sandwich method detection system connects is measured, and can reach close discrimination compared with artificial counting method.

The various embodiments described above are only for illustrating the present invention; wherein the structure of each parts, connected mode and method step etc. all can change to some extent; every equivalents of carrying out on the basis of technical solution of the present invention and improvement, all should not get rid of outside protection scope of the present invention.

Claims (10)

1. the gold nano-material method of counting on transparent smooth sheet form base surface, it comprises the following steps:

1) during by sample determination, solvent for use adds to transparent smooth sheet form base surface, is then placed under dark field microscope and observes, and collects backscatter image;

2) gold nano-material is added to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the feature dispersion image of gold nano-material;

3) testing sample that contains variable concentrations gold nano-material is added to respectively to transparent smooth sheet form base surface, be then placed under dark field microscope and observe, collect the dispersion image of some testing samples;

4) to step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains carry out successively graphical analysis, obtain the quantity of gold nano-material in testing sample dispersion image.

2. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 1 surface, it is characterized in that: described step 1), step 2) and step 3) in, transparent smooth sheet form base adopts the one in transparent inorganic material and transparent polymer material, transparent inorganic material adopts the one in simple glass and quartz glass, and transparent polymer material adopts the one in dimethyl silicone polymer, organic glass and plastics.

3. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 1 surface; it is characterized in that: described step 2) and step 3) in; gold nano-material adopts the aggregation of gold copper-base alloy particle or gold copper-base alloy particle; gold copper-base alloy particle adopts the one in spherical gold nano grain, gold nanorods and gold nanoshell, and the size of gold copper-base alloy particle and gold copper-base alloy particle aggregate is 10nm～1000nm.

4. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 2 surface; it is characterized in that: described step 2) and step 3) in; gold nano-material adopts the aggregation of gold copper-base alloy particle or gold copper-base alloy particle; gold copper-base alloy particle adopts the one in spherical gold nano grain, gold nanorods and gold nanoshell, and the size of gold copper-base alloy particle and gold copper-base alloy particle aggregate is 10nm～1000nm.

5. the gold nano-material method of counting on a kind of transparent smooth sheet form base surface as claimed in claim 1 or 2 or 3 or 4, it is characterized in that: described step 1), step 2) and step 3) in, the time shutter while gathering details in a play not acted out on stage, but told through dialogues dispersion image is 10ms～1000ms; In the details in a play not acted out on stage, but told through dialogues dispersion image collecting, a size corresponding to pixel is less than or equal to 0.5 μ m × 0.5 μ m.

6. the gold nano-material method of counting on a kind of transparent smooth sheet form base surface as claimed in claim 1 or 2 or 3 or 4, is characterized in that: described step 4) in, dispersion image is carried out to graphical analysis, it specifically comprises the following steps:

I, color space conversion;

Respectively by step 1) in the backscatter image, the step 2 that obtain) in feature dispersion image and the step 3 of the gold nano-material that obtains) in the dispersion image of the testing sample that obtains be converted to L*A*B* color space from sRGB color space, correspondingly respectively obtain the image in L*A*B* color space;

II, luminance threshold computing;

In the L*A*B* color space that step I is obtained, all pixels of image are carried out luminance threshold computing, obtain non-background pixel and background pixel;

III, image are cut apart;

The non-background pixel that adopts four floodings that are communicated with to obtain step II is cut apart, and obtains all four subregion lists that are communicated with; Give up and comprise non-background pixel quantity and be less than 9 subregion, calculate area and the axial ratio of non-background pixel composition figure in each subregion, chosen axis compares in theoretical span and the subregion of area in the theoretical span of gold nano-material imaging area, obtains subimage list;

IV, color character extract;

Each width subimage in the subimage list of the gold nano-material feature dispersion image obtaining to each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III and through the processing of step I～step III carries out color character extraction;

Gold nano-material counting in V, testing sample dispersion image;

For each subimage in each subimage or the list of standard model dispersion image subimage in the testing sample dispersion image subimage list obtaining through the processing of step I～step III, L*, the A* of each the non-background pixel color in subimage and the value of B* are rounded to integer, search corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell; If the numerical value of cell is 1, non-background pixel is gold nano passage pixel; In calculating subimage, gold nano passage pixel accounts for the ratio of non-background pixel, if this ratio is greater than 50%, this subimage is gold nano-material subimage;

Calculate the quantity of all gold nano-material subimages, this quantity is the gold nano-material quantity in testing sample dispersion image or standard model dispersion image.

7. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 6 surface, is characterized in that: in described step III, adopt the axial ratio of the method calculating pixel composition figure of spatial moment, its detailed process is:

The center of gravity of (I) calculating pixel institute composition diagram picture is:

$\left\{\begin{array}{c}\stackrel{\_}{x}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\mathrm{jF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\\ \stackrel{\_}{y}=\frac{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\mathrm{kF}(j,k)}{\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}F(j,k)}\end{array}\right.,$

In formula, J and K are respectively width and the height of figure that pixel forms, and F (j, k) is the value that point (j, k) is located.Wherein, for background pixel, the value that point (j, k) is located is 0; For non-background pixel, the value that point (j, k) is located is 1;

(II) calculates major axis eigenvalue λ _mwith minor axis eigenvalue λ _n, it specifically comprises:

According to the center of gravity of pixel institute composition diagram picture , calculate row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks and intersect moment of inertia U (1,1) and be respectively:

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{x})}^{2}F(j,k),$

$U\left(\mathrm{2,0}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(j-\stackrel{\_}{y})}^{2}F(j,k),$

$U\left(\mathrm{1,1}\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}(j-\stackrel{\_}{x})(k-\stackrel{\_}{y})F(j,k)$

According to row moment of inertia U (2,0), row moment of inertia U (0,2) and ranks intersection moment of inertia U (1,1), calculate two eigenvalue λ ₁and λ ₂:

$\left\{\begin{array}{c}{\mathrm{\λ}}_1=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]+\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\\ {\mathrm{\λ}}_2=\frac{1}{2}[U\left(\mathrm{2,0}\right)+U\left(\mathrm{0,2}\right)]-\frac{1}{2}{[U{\left(\mathrm{2,0}\right)}^2+U{\left(\mathrm{0,2}\right)}^2-2U\left(\mathrm{2,0}\right)U\left(\mathrm{0,2}\right)+4U{\left(\mathrm{1,1}\right)}^2]}^{1/2}\end{array}\right.,$

According to eigenvalue λ ₁and λ ₂, obtain major axis eigenvalue λ _mwith minor axis eigenvalue λ _nbe respectively:

$\left\{\begin{array}{c}{\mathrm{\λ}}_M=\max \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\\ {\mathrm{\λ}}_N=\min \{{\mathrm{\λ}}_1,{\mathrm{\λ}}_2\}\end{array}\right.;$

The major axis eigenvalue λ that (III) obtains according to step (II) _mwith minor axis eigenvalue λ _n, the axial ratio R that obtains pixel composition figure is:

$R=\frac{{\mathrm{\λ}}_N}{{\mathrm{\λ}}_M};$

(IV) according to formula (4), the span of the axial ratio R of pixel composition figure be (0,1]; When pixel composition figure is circular, axial ratio R value is 1.

8. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 6 surface, it is characterized in that: in described step III, according to far-field optics imaging resolution formula, the theoretical span of reference axis ratio and the theoretical span of gold nano-material imaging area, its computation process is:

Gold nano-material is considered as ellipse by (I), the major axis that this is oval and minor axis length are respectively a and b, in conjunction with the numerical aperture NA of light wavelength λ and micro objective, calculate the ideally image axial ratio R' of gold nano-material after microscope imaging and be:

$R^{\′}=\frac{2b+r}{2a+r},$

In formula, r is microscopical optical resolution,

Deviation between (II) setting axial ratio of subimage and the axial ratio of ideal image causes the uncertainty of a marginal existence m pixel of subimage, it is the uncertainty that the major axis of subimage or minor axis exist 2m pixel, and subimage entirety produces the broadening of n pixel, and the theoretical span of axial ratio is:

$[\frac{2b+r-(2m-n)l}{2a+r+(2m+n)l},\frac{2b+r+(2m+n)l}{2a+r+(2m+n)l}],$

In formula, l is the foursquare actual length of side corresponding to single pixel;

(III) gold nano-material imaging area A is:

$A=\mathrm{\π}(a+\frac{r}{2})(b+\frac{r}{2}),$

The uncertainty of the marginal existence m pixel based on subimage and subimage entirety produce the broadening of n pixel, and the theoretical span of gold nano-material imaging area is:

In formula, expression rounds up.

9. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 6 surface, is characterized in that: in described step IV, the detailed process that color character extracts is:

I) for each the width subimage in the subimage list of the backscatter image obtaining through the processing of step I～step III, calculate respectively L*, the A* of all non-background pixels and the color mean value of B* color channel, obtain the average color of subimage; Use CIEDE2000 aberration algorithm, the difference in calculating subimage between all non-background pixels and average color; The span of default color difference threshold is 10～30, and the pixel quantity that calculated difference is less than color difference threshold accounts for the ratio of non-background pixel, obtains the coverage rate of average color; If the coverage rate of average color is greater than 50%, the average color of subimage is added in the feature color list of backscatter image;

II) for each the width subimage in the subimage list of the gold nano-material feature dispersion image obtaining through the processing of step I～step III, by step I) in method, calculate the coverage rate of average color and average color;

If the coverage rate of average color is less than 100%, calculate respectively in all non-background pixel of subimage of gold nano-material feature dispersion image, difference is more than or equal to color difference threshold and brightness is greater than L*, the A* of the pixel of average color brightness and the color mean value of B* color channel, obtains the average color of highlight area; Use CIEDE2000 aberration algorithm, the pixel that the arbitrary difference of average color of calculating and average color or highlight area is less than color difference threshold accounts for the ratio of non-background pixel, if this ratio is greater than 50%, the average color of subimage of gold nano-material feature dispersion image and the average color of highlight area are added to the list of gold nano-material dispersion image feature color;

III) for the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color, use CIEDE2000 aberration algorithm, difference in the feature color of calculating gold nano-material and the list of backscatter characteristics of image color between each background characteristics color, if there is arbitrary difference to be less than color difference threshold, the feature color of gold nano-material is removed from the list of gold nano-material dispersion image feature color, if all differences are all more than or equal to color difference threshold, by the L* of the feature color of gold nano-material, the numerical value of A* and B* is all rounded to integer,

IV) set up gold nano-material feature color look-up table, the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color is put into look-up table, the circumstances in which people get things ready for a trip of going forward side by side enlargement of color photos exhibition.

10. the gold nano-material method of counting on a kind of transparent smooth sheet form base as claimed in claim 9 surface, it is characterized in that: described step IV) in, set up gold nano-material feature color look-up table, the feature color of each gold nano-material in the list of gold nano-material dispersion image feature color is put into look-up table, the circumstances in which people get things ready for a trip of going forward side by side enlargement of color photos exhibition, it specifically comprises the following steps:

(I) sets up the table that comprises L*, A* and tri-dimensions of B*, and using this table as gold nano-material feature color look-up table, wherein the span of L* is [0,100], the span of A* is [86,98], the span of B* is [108,94], and the change step of L*, A* and B* is 1; The numerical value of all cells in gold nano-material feature color look-up table is all preset as to 0;

(II), for each color-values in the list of gold nano-material dispersion image feature color, is set to 1 by the numerical value of corresponding (L*, A*, B*) cell in gold nano-material feature color look-up table;

The span that (III) presses L* is [35,100], and the span of A* is [86,98], and the span of B* is [1,94] traversal gold nano-material feature color look-up table; For the each (L* in ergodic process, A*, B*) color-values, is used CIEDE2000 aberration algorithm, calculates each (L*, A*, B*) difference between each the gold nano-material feature color in color-values and the list of gold nano-material dispersion image feature color, if there is arbitrary difference to be less than color difference threshold, by corresponding (L* in gold nano-material feature color look-up table, A*, B*) numerical value of cell is set to 1.