CN106361280A - Optical imaging system adopted by bio-optical imaging device of combined iris and cortical tissues - Google Patents

Abstract

The invention discloses an optical imaging system adopted by a bio-optical imaging device of combined iris and cortical tissues. The optical imaging system comprises a cell identification module and an optical imaging device, wherein the cell identification module is used for determining a biological type; the optical imaging device consists of an illuminating light source unit and an imaging unit. The optical imaging system has the beneficial effects that the optical imaging system has ordinary people adaptability, and meets national level large-scale application; most reliable biological tissue activity detection analysis is performed, and the reliability of a biological assay system is ensured; based on an image analysis assay method, the accuracy and reliability of a biological assay system are improved.

Description

The light that the Photobiology imaging device of a kind of combined iris and cortical tissue adopts studies As device

Technical field

The present invention relates to optical imaging field is and in particular to the Photobiology imaging of a kind of combined iris and cortical tissue fills Put the optical imaging device of employing.

Background technology

At present, demand preferably improves the performance of bioassay system, comprising: solve to lead to iris deformation when heterphoria When making iris be non-circular produce mistake problem, the expression way of more preferable characteristic image, further improve feature extraction and The ability of coding information entropy, and improve the accuracy and reliability that template is estimated, the accuracy and reliability of far/frr.

Content of the invention

For solving the above problems, the present invention is intended to provide the Photobiology imaging device of a kind of combined iris and cortical tissue Using optical imaging device.

The purpose of the present invention employs the following technical solutions to realize:

The optical imaging device that the Photobiology imaging device of a kind of combined iris and cortical tissue adopts, knows including cell Other module and optical imaging device, described cell recognition module is used for determining biological species, described optical imaging device is by illuminating Light source cell and image-generating unit are constituted, and light source unit for lighting is configured to multispectral multi-polarization state combination optical with image-generating unit and becomes As system, described multispectral multi-polarization state combination optical imaging system produces at least four combined imagings, comprising: black light With orthogonal polarisation state combined imaging, black light and parallel polarization states combined imaging, near infrared light and orthogonal polarisation state are combined into Picture, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

The invention has the benefit that having the General Population suitability, meet based on national sizable application；The most reliable The tissue activity detection analysis of property are it is ensured that the reliability of of bioassay system itself；Based on graphical analyses assay method Improve the accuracy and reliability of bioassay system performance.

Brief description

Using accompanying drawing, the invention will be further described, but the embodiment in accompanying drawing does not constitute any limit to the present invention System, for those of ordinary skill in the art, on the premise of not paying creative work, can also obtain according to the following drawings Other accompanying drawings.

Fig. 1 is present configuration schematic diagram；

Fig. 2 is the structural representation of cell recognition module.

Reference:

Cell recognition module 1, Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, Classification and Identification unit 13.

Specific embodiment

In conjunction with following application scenarios, the invention will be further described.

Application scenarios 1

Referring to Fig. 1, Fig. 2, a kind of combined iris of an embodiment of this application scene and the Photobiology of cortical tissue The optical imaging device that imaging device adopts, including cell recognition module and optical imaging device, described cell recognition module is used To determine biological species, described optical imaging device is made up of light source unit for lighting and image-generating unit, light source unit for lighting with become As unit is configured to multispectral multi-polarization state combination optical imaging system, described multispectral multi-polarization state combination optical is imaged System produces at least four combined imagings, comprising: black light and orthogonal polarisation state combined imaging, black light and parallel polarization State combined imaging, near infrared light and orthogonal polarisation state combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

Preferably, described light source unit for lighting and image-generating unit are combined and are configured to the wave-length coverage of at least black light For 300-500nm, the wave-length coverage of near infrared light is 700-900nm, and the polarizer and analyzer are combined and are configured at least to have Orthogonal with parallel polarization state.

This preferred embodiment image quality is higher.

Preferably, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.

This preferred embodiment areas imaging is wider.

Preferably, described cell recognition module 1 includes Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, classification knowledge Other unit 13；Described Methods of Segmentation On Cell Images unit 11 is used for distinguishing the back of the body in the cell image being gathered by cell image acquisition module Scape, nucleus and Cytoplasm；Described feature extraction unit 12 is used for the textural characteristics of cell image are extracted；Described classification Recognition unit 13 is used for utilizing grader to realize to cell image Classification and Identification according to textural characteristics.

This preferred embodiment constructs the unit structure of cell recognition module 1.

Preferably, described Methods of Segmentation On Cell Images unit 11 includes image conversion subunit, noise remove subelement, coarse segmentation Subelement, nuclear centers demarcate subelement, Accurate Segmentation subelement, particularly as follows:

(1) image conversion subunit, for being converted into gray level image by the cell image of collection；

(2) noise remove subelement, for carrying out denoising to gray level image, comprising:

For pixel (x, y), choose its 3 × 3 neighborhood s_x,y(2n+1) the neighborhood l of × (2n+1)_x,y, n be more than Integer equal to 2；

Whether it is that boundary point judges first to pixel, given threshold t, t ∈ [13,26], calculate pixel (x, y) With its neighborhood s_x,yIn each pixel gray scale difference value, and be compared with threshold value t, if gray scale difference value is more than the number of threshold value t More than or equal to 6, then pixel (x, y) is boundary point, and otherwise, pixel (x, y) is non-boundary point；

If (x, y) is boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{q(i,j)\&element;\[q(x,y)-1.5\mathrm{\σ},q(x,y)+1.5\mathrm{\σ}\]}q(i,j)}{k}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, and q (x, y) is the ash of noise reduction preceding pixel point (x, y) Angle value, σ is pixel (x, y) neighborhood l_x,yInterior gray value mark is poor, and q (i, j) ∈ [q (x, y) -1.5 σ, q (x, y)+1.5 σ] represents Neighborhood l_x,yInterior gray value falls within the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ], and k represents neighborhood l_x,yInterior gray value falls within The quantity of the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ]；

If (x, y) is non-boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)q(i,j)}{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, the ash at q (i, j) representative image midpoint (i, j) place Angle value, w (i, j) is neighborhood l_x,yThe corresponding Gauss weight of interior point (i, j)；

(3) coarse segmentation subelement, for slightly being drawn to the background in the cell image after denoising, Cytoplasm, nucleus Point, particularly as follows:

Each pixel (x, y) is represented with four dimensional feature vectors:

$\stackrel{\→}{u}(x,y)=\[h(x,y),h_{ave}(x,y),h_{med}(x,y),h_{sta}(x,y)\]$

In formula, h (x, y) represents the gray value of (x, y), h_ave(x, y) represents its neighborhood s_x,yGray average, h_med(x, y) generation Table its neighborhood s_x,yGray scale intermediate value, h_sta(x, y) represents its neighborhood s_x,yGray variance；

Background, Cytoplasm, nucleus three class are divided into using k-means clustering procedure；

(4) nuclear centers demarcate subelement, for demarcating to nuclear centers:

Nucleus approximate region is obtained by coarse segmentation subelement, if nuclear area comprises n point: (x₁,y₁),…,(x_n, y_n), this region is carried out with intensity-weighted demarcation and geometric center is demarcated, take its meansigma methods as nuclear centers (x_z,y_z):

$x_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{x}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{x}_i}{n})$

$y_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{y}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{y}_i}{n})$

(5) Accurate Segmentation subelement, for carrying out Accurate Segmentation to nucleus, Cytoplasm；

Build from nuclear centers (x_z,y_z) arrive nucleus and Cytoplasm boundary point (x_p,y_p) directed line segmentDistanceRepresent and round downwards；

Along line segment, sampling is carried out with unit length and can obtain dis_pIndividual pointIf adopting The coordinate of sampling point is not integer, and its gray value is obtained by surrounding pixel linear interpolation；

Point (x_i,y_i) place along line segment direction gray scale difference:

hd(x_i,y_i)=h (x_i-1,y_i-1)-h(x_i,y_i)

Definition gray scale difference inhibition function:

$y\left(x\right)=\left\{\begin{array}{ccc}x& if& x\≤0\\ 0.5x& if& x>0\end{array}\right.$

Point (x_i,y_i) place along line segment direction gradient gra (x_i,y_i):

$gra(x_i,y_i)=\frac{\left|y\left(hd\right(x_i,y_i)\right)|+|y\left(hd\right(x_{i+1},y_{i+!})\left)\right|}{2}$

Choose the maximum value point of gradient as nucleus and cytoplasmic precise edge.

This preferred embodiment arranges noise remove subelement, and effective integration center pixel is closed on the space of neighborhood territory pixel Property and grey similarity carrying out noise reduction process, flat site in the picture, in neighborhood, grey scale pixel value is more or less the same, and adopts Gaussian filter is weighted to gray value filtering, and is changing violent borderline region, row bound keeps filtering, is conducive to image The holding at edge；Nucleus and Cytoplasm coarse contour are extracted using k mean cluster, can effectively remove the interference of noise；Setting is thin Subelement is demarcated at karyon center, is easy to subsequently nucleus and Cytoplasm profile are accurately positioned；Accurate Segmentation subelement fills Divide and make use of directional information, overcome the interference to edge graph for the inflammatory cell, can accurately extract nucleus and Cytoplasm side Edge.

Preferably, the described textural characteristics to cell image extract, comprising:

(1) the Gray co-occurrence matrix of cell image, described comprehensive ash is asked for based on improved gray level co-occurrence matrixes method Degree co-occurrence matrix embodies cell textural characteristics in different directions:

Be located at 0 °, 45 °, 90 °, the gray level co-occurrence matrixes on 135 ° of four directions be respectively h (x, y, d, 0 °), h (x, y, d, 45 °), h (x, y, d, 90 °), h (x, y, d, 135 °), corresponding matrix element project is x₁、x₂、x₃、x₄, then Gray is common The computing formula of raw matrix is:

H (x, y, d)=w₁h(x,y,d,0°)+w₂h(x,y,d,45°)+w₃h(x,y,d,90°)+w₄h(x,y,d,135°)

Gray co-occurrence matrix element number is:

$x=\underset{i=1}{\overset{4}{\σ}}{w}_i{x}_i$

In formula, d represents distance, the span of d is [2,4], w_iFor weight coefficient, i=1,2,3,4, it is by four sides To in each direction on the corresponding contrast level parameter of gray level co-occurrence matrixes calculate, if the gray level co-occurrence matrixes on four direction Corresponding contrast level parameter is respectively d_i, average isI=1,2,3,4, then weight coefficient w_iComputing formula be:

$w_i=\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}/\underset{i=1}{\overset{4}{\σ}}\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}$

(2) four textural characteristics parameters needed for being obtained using described Gray co-occurrence matrix and matrix element project: Contrast, variance and, energy and average；

(3) described four textural characteristics parameters are normalized, finally obtain normalized texture eigenvalue.

This preferred embodiment is based on improved gray level co-occurrence matrixes method, asks for cytological map by the way of setting weight coefficient The Gray co-occurrence matrix of picture, and then extract textural characteristics on specified four direction for the cell, solve due to outside dry Disturb the textural characteristics ginseng of the cell that (impact that causes as lighting angle when cell image gathers, flowing interference of gas etc.) causes Numerical value has the problem of bigger difference in different directions, improves the precision of cell image texture feature extraction；Selected contrast, Variance and, energy and four textural characteristics of average, eliminate redundancy and the characteristic parameter repeating；To described four textural characteristics ginseng Number is normalized, and the Classification and Identification facilitating follow-up cell image is processed.

In this application scenarios, given threshold t=13, d=2, image denoising effect improves 5% relatively, cell image The extraction accuracy of feature improves 8%.

Application scenarios 2

Referring to Fig. 1, Fig. 2, a kind of combined iris of an embodiment of this application scene and the Photobiology of cortical tissue The optical imaging device that imaging device adopts, including cell recognition module and optical imaging device, described cell recognition module is used To determine biological species, described optical imaging device is made up of light source unit for lighting and image-generating unit, light source unit for lighting with become As unit is configured to multispectral multi-polarization state combination optical imaging system, described multispectral multi-polarization state combination optical is imaged System produces at least four combined imagings, comprising: black light and orthogonal polarisation state combined imaging, black light and parallel polarization State combined imaging, near infrared light and orthogonal polarisation state combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

Preferably, described light source unit for lighting and image-generating unit are combined and are configured to the wave-length coverage of at least black light For 300-500nm, the wave-length coverage of near infrared light is 700-900nm, and the polarizer and analyzer are combined and are configured at least to have Orthogonal with parallel polarization state.

This preferred embodiment image quality is higher.

Preferably, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.

This preferred embodiment areas imaging is wider.

Preferably, described cell recognition module 1 includes Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, classification knowledge Other unit 13；Described Methods of Segmentation On Cell Images unit 11 is used for distinguishing the back of the body in the cell image being gathered by cell image acquisition module Scape, nucleus and Cytoplasm；Described feature extraction unit 12 is used for the textural characteristics of cell image are extracted；Described classification Recognition unit 13 is used for utilizing grader to realize to cell image Classification and Identification according to textural characteristics.

This preferred embodiment constructs the unit structure of cell recognition module 1.

Preferably, described Methods of Segmentation On Cell Images unit 11 includes image conversion subunit, noise remove subelement, coarse segmentation Subelement, nuclear centers demarcate subelement, Accurate Segmentation subelement, particularly as follows:

(1) image conversion subunit, for being converted into gray level image by the cell image of collection；

(2) noise remove subelement, for carrying out denoising to gray level image, comprising:

For pixel (x, y), choose its 3 × 3 neighborhood s_x,y(2n+1) the neighborhood l of × (2n+1)_x,y, n be more than Integer equal to 2；

Whether it is that boundary point judges first to pixel, given threshold t, t ∈ [13,26], calculate pixel (x, y) With its neighborhood s_x,yIn each pixel gray scale difference value, and be compared with threshold value t, if gray scale difference value is more than the number of threshold value t More than or equal to 6, then pixel (x, y) is boundary point, and otherwise, pixel (x, y) is non-boundary point；

If (x, y) is boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{q(i,j)\&element;\[q(x,y)-1.5\mathrm{\σ},q(x,y)+1.5\mathrm{\σ}\]}q(i,j)}{k}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, and q (x, y) is the ash of noise reduction preceding pixel point (x, y) Angle value, σ is pixel (x, y) neighborhood l_x,yInterior gray value mark is poor, and q (i, j) ∈ [q (x, y) -1.5 σ, q (x, y)+1.5 σ] represents Neighborhood l_x,yInterior gray value falls within the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ], and k represents neighborhood l_x,yInterior gray value falls within The quantity of the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ]；

If (x, y) is non-boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)q(i,j)}{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, the ash at q (i, j) representative image midpoint (i, j) place Angle value, w (i, j) is neighborhood l_x,yThe corresponding Gauss weight of interior point (i, j)；

(3) coarse segmentation subelement, for slightly being drawn to the background in the cell image after denoising, Cytoplasm, nucleus Point, particularly as follows:

Each pixel (x, y) is represented with four dimensional feature vectors:

$\stackrel{\→}{u}(x,y)=\[h(x,y),h_{ave}(x,y),h_{med}(x,y),h_{sta}(x,y)\]$

In formula, h (x, y) represents the gray value of (x, y), h_ave(x, y) represents its neighborhood s_x,yGray average, h_med(x, y) generation Table its neighborhood s_x,yGray scale intermediate value, h_sta(x, y) represents its neighborhood s_x,yGray variance；

Background, Cytoplasm, nucleus three class are divided into using k-means clustering procedure；

(4) nuclear centers demarcate subelement, for demarcating to nuclear centers:

Nucleus approximate region is obtained by coarse segmentation subelement, if nuclear area comprises n point: (x₁,y₁),…,(x_n, y_n), this region is carried out with intensity-weighted demarcation and geometric center is demarcated, take its meansigma methods as nuclear centers (x_z,y_z):

$x_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{x}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{x}_i}{n})$

$y_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{y}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{y}_i}{n})$

(5) Accurate Segmentation subelement, for carrying out Accurate Segmentation to nucleus, Cytoplasm；

Build from nuclear centers (x_z,y_z) arrive nucleus and Cytoplasm boundary point (x_p,y_p) directed line segmentDistanceRepresent and round downwards；

Along line segment, sampling is carried out with unit length and can obtain dis_pIndividual pointIf adopting The coordinate of sampling point is not integer, and its gray value is obtained by surrounding pixel linear interpolation；

Point (x_i,y_i) place along line segment direction gray scale difference:

hd(x_i,y_i)=h (x_i-1,y_i-1)-h(x_i,y_i)

Definition gray scale difference inhibition function:

$y\left(x\right)=\left\{\begin{array}{ccc}x& if& x\≤0\\ 0.5x& if& x>0\end{array}\right.$

Point (x_i,y_i) place along line segment direction gradient gra (x_i,y_i):

$gra(x_i,y_i)=\frac{\left|y\left(hd\right(x_i,y_i)\right)|+|y\left(hd\right(x_{i+1},y_{i+!})\left)\right|}{2}$

Choose the maximum value point of gradient as nucleus and cytoplasmic precise edge.

This preferred embodiment arranges noise remove subelement, and effective integration center pixel is closed on the space of neighborhood territory pixel Property and grey similarity carrying out noise reduction process, flat site in the picture, in neighborhood, grey scale pixel value is more or less the same, and adopts Gaussian filter is weighted to gray value filtering, and is changing violent borderline region, row bound keeps filtering, is conducive to image The holding at edge；Nucleus and Cytoplasm coarse contour are extracted using k mean cluster, can effectively remove the interference of noise；Setting is thin Subelement is demarcated at karyon center, is easy to subsequently nucleus and Cytoplasm profile are accurately positioned；Accurate Segmentation subelement fills Divide and make use of directional information, overcome the interference to edge graph for the inflammatory cell, can accurately extract nucleus and Cytoplasm side Edge.

Preferably, the described textural characteristics to cell image extract, comprising:

(1) the Gray co-occurrence matrix of cell image, described comprehensive ash is asked for based on improved gray level co-occurrence matrixes method Degree co-occurrence matrix embodies cell textural characteristics in different directions:

Be located at 0 °, 45 °, 90 °, the gray level co-occurrence matrixes on 135 ° of four directions be respectively h (x, y, d, 0 °), h (x, y, d, 45 °), h (x, y, d, 90 °), h (x, y, d, 135 °), corresponding matrix element project is x₁、x₂、x₃、x₄, then Gray is common The computing formula of raw matrix is:

H (x, y, d)=w₁h(x,y,d,0°)+w₂h(x,y,d,45°)+w₃h(x,y,d,90°)+w₄h(x,y,d,135°)

Gray co-occurrence matrix element number is:

$x=\underset{i=1}{\overset{4}{\σ}}{w}_i{x}_i$

In formula, d represents distance, the span of d is [2,4], w_iFor weight coefficient, i=1,2,3,4, it is by four sides To in each direction on the corresponding contrast level parameter of gray level co-occurrence matrixes calculate, if the gray level co-occurrence matrixes on four direction Corresponding contrast level parameter is respectively d_i, average isI=1,2,3,4, then weight coefficient w_iComputing formula be:

$w_i=\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}/\underset{i=1}{\overset{4}{\σ}}\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}$

(2) four textural characteristics parameters needed for being obtained using described Gray co-occurrence matrix and matrix element project: Contrast, variance and, energy and average；

(3) described four textural characteristics parameters are normalized, finally obtain normalized texture eigenvalue.

This preferred embodiment is based on improved gray level co-occurrence matrixes method, asks for cytological map by the way of setting weight coefficient The Gray co-occurrence matrix of picture, and then extract textural characteristics on specified four direction for the cell, solve due to outside dry Disturb the textural characteristics ginseng of the cell that (impact that causes as lighting angle when cell image gathers, flowing interference of gas etc.) causes Numerical value has the problem of bigger difference in different directions, improves the precision of cell image texture feature extraction；Selected contrast, Variance and, energy and four textural characteristics of average, eliminate redundancy and the characteristic parameter repeating；To described four textural characteristics ginseng Number is normalized, and the Classification and Identification facilitating follow-up cell image is processed.

In this application scenarios, given threshold t=15, d=2, image denoising effect improves 6% relatively, cell image The extraction accuracy of feature improves 8%.

Application scenarios 3

Referring to Fig. 1, Fig. 2, a kind of combined iris of an embodiment of this application scene and the Photobiology of cortical tissue The optical imaging device that imaging device adopts, including cell recognition module and optical imaging device, described cell recognition module is used To determine biological species, described optical imaging device is made up of light source unit for lighting and image-generating unit, light source unit for lighting with become As unit is configured to multispectral multi-polarization state combination optical imaging system, described multispectral multi-polarization state combination optical is imaged System produces at least four combined imagings, comprising: black light and orthogonal polarisation state combined imaging, black light and parallel polarization State combined imaging, near infrared light and orthogonal polarisation state combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

Preferably, described light source unit for lighting and image-generating unit are combined and are configured to the wave-length coverage of at least black light For 300-500nm, the wave-length coverage of near infrared light is 700-900nm, and the polarizer and analyzer are combined and are configured at least to have Orthogonal with parallel polarization state.

This preferred embodiment image quality is higher.

Preferably, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.

This preferred embodiment areas imaging is wider.

Preferably, described cell recognition module 1 includes Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, classification knowledge Other unit 13；Described Methods of Segmentation On Cell Images unit 11 is used for distinguishing the back of the body in the cell image being gathered by cell image acquisition module Scape, nucleus and Cytoplasm；Described feature extraction unit 12 is used for the textural characteristics of cell image are extracted；Described classification Recognition unit 13 is used for utilizing grader to realize to cell image Classification and Identification according to textural characteristics.

This preferred embodiment constructs the unit structure of cell recognition module 1.

Preferably, described Methods of Segmentation On Cell Images unit 11 includes image conversion subunit, noise remove subelement, coarse segmentation Subelement, nuclear centers demarcate subelement, Accurate Segmentation subelement, particularly as follows:

(1) image conversion subunit, for being converted into gray level image by the cell image of collection；

(2) noise remove subelement, for carrying out denoising to gray level image, comprising:

For pixel (x, y), choose its 3 × 3 neighborhood s_x,y(2n+1) the neighborhood l of × (2n+1)_x,y, n be more than Integer equal to 2；

Whether it is that boundary point judges first to pixel, given threshold t, t ∈ [13,26], calculate pixel (x, y) With its neighborhood s_x,yIn each pixel gray scale difference value, and be compared with threshold value t, if gray scale difference value is more than the number of threshold value t More than or equal to 6, then pixel (x, y) is boundary point, and otherwise, pixel (x, y) is non-boundary point；

If (x, y) is boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{q(i,j)\&element;\[q(x,y)-1.5\mathrm{\σ},q(x,y)+1.5\mathrm{\σ}\]}q(i,j)}{k}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, and q (x, y) is the ash of noise reduction preceding pixel point (x, y) Angle value, σ is pixel (x, y) neighborhood l_x,yInterior gray value mark is poor, and q (i, j) ∈ [q (x, y) -1.5 σ, q (x, y)+1.5 σ] represents Neighborhood l_x,yInterior gray value falls within the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ], and k represents neighborhood l_x,yInterior gray value falls within The quantity of the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ]；

If (x, y) is non-boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)q(i,j)}{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, the ash at q (i, j) representative image midpoint (i, j) place Angle value, w (i, j) is neighborhood l_x,yThe corresponding Gauss weight of interior point (i, j)；

(3) coarse segmentation subelement, for slightly being drawn to the background in the cell image after denoising, Cytoplasm, nucleus Point, particularly as follows:

Each pixel (x, y) is represented with four dimensional feature vectors:

$\stackrel{\→}{u}(x,y)=\[h(x,y),h_{ave}(x,y),h_{med}(x,y),h_{sta}(x,y)\]$

In formula, h (x, y) represents the gray value of (x, y), h_ave(x, y) represents its neighborhood s_x,yGray average, h_med(x, y) generation Table its neighborhood s_x,yGray scale intermediate value, h_sta(x, y) represents its neighborhood s_x,yGray variance；

Background, Cytoplasm, nucleus three class are divided into using k-means clustering procedure；

(4) nuclear centers demarcate subelement, for demarcating to nuclear centers:

Nucleus approximate region is obtained by coarse segmentation subelement, if nuclear area comprises n point: (x₁,y₁),…,(x_n, y_n), this region is carried out with intensity-weighted demarcation and geometric center is demarcated, take its meansigma methods as nuclear centers (x_z,y_z):

$x_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{x}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{x}_i}{n})$

$y_z=\frac{1}{2}(\frac{{\mathrm{\σ}}_{i=1}^n{y}_{i}h(x_i,y_i)}{{\mathrm{\σ}}_{i=1}^{n}h(x_i,y_i)}+\frac{{\mathrm{\σ}}_{i=1}^n{y}_i}{n})$

(5) Accurate Segmentation subelement, for carrying out Accurate Segmentation to nucleus, Cytoplasm；

Build from nuclear centers (x_z,y_z) arrive nucleus and Cytoplasm boundary point (x_p,y_p) directed line segmentDistanceRepresent and round downwards；

Along line segment, sampling is carried out with unit length and can obtain dis_pIndividual pointIf adopting The coordinate of sampling point is not integer, and its gray value is obtained by surrounding pixel linear interpolation；

Point (x_i,y_i) place along line segment direction gray scale difference:

hd(x_i,y_i)=h (x_i-1,y_i-1)-h(x_i,y_i)

Definition gray scale difference inhibition function:

$y\left(x\right)=\left\{\begin{array}{ccc}x& if& x\≤0\\ 0.5x& if& x>0\end{array}\right.$

Point (x_i,y_i) place along line segment direction gradient gra (x_i,y_i):

$gra(x_i,y_i)=\frac{\left|y\left(hd\right(x_i,y_i)\right)|+|y\left(hd\right(x_{i+1},y_{i+!})\left)\right|}{2}$

Choose the maximum value point of gradient as nucleus and cytoplasmic precise edge.

This preferred embodiment arranges noise remove subelement, and effective integration center pixel is closed on the space of neighborhood territory pixel Property and grey similarity carrying out noise reduction process, flat site in the picture, in neighborhood, grey scale pixel value is more or less the same, and adopts Gaussian filter is weighted to gray value filtering, and is changing violent borderline region, row bound keeps filtering, is conducive to image The holding at edge；Nucleus and Cytoplasm coarse contour are extracted using k mean cluster, can effectively remove the interference of noise；Setting is thin Subelement is demarcated at karyon center, is easy to subsequently nucleus and Cytoplasm profile are accurately positioned；Accurate Segmentation subelement fills Divide and make use of directional information, overcome the interference to edge graph for the inflammatory cell, can accurately extract nucleus and Cytoplasm side Edge.

Preferably, the described textural characteristics to cell image extract, comprising:

(1) the Gray co-occurrence matrix of cell image, described comprehensive ash is asked for based on improved gray level co-occurrence matrixes method Degree co-occurrence matrix embodies cell textural characteristics in different directions:

Be located at 0 °, 45 °, 90 °, the gray level co-occurrence matrixes on 135 ° of four directions be respectively h (x, y, d, 0 °), h (x, y, d, 45 °), h (x, y, d, 90 °), h (x, y, d, 135 °), corresponding matrix element project is x₁、x₂、x₃、x₄, then Gray is common The computing formula of raw matrix is:

H (x, y, d)=w₁h(x,y,d,0°)+w₂h(x,y,d,45°)+w₃h(x,y,d,90°)+w₄h(x,y,d,135°)

Gray co-occurrence matrix element number is:

$x=\underset{i=1}{\overset{4}{\σ}}{w}_i{x}_i$

In formula, d represents distance, the span of d is [2,4], w_iFor weight coefficient, i=1,2,3,4, it is by four sides To in each direction on the corresponding contrast level parameter of gray level co-occurrence matrixes calculate, if the gray level co-occurrence matrixes on four direction Corresponding contrast level parameter is respectively d_i, average isI=1,2,3,4, then weight coefficient w_iComputing formula be:

$w_i=\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}/\underset{i=1}{\overset{4}{\σ}}\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}$

(2) four textural characteristics parameters needed for being obtained using described Gray co-occurrence matrix and matrix element project: Contrast, variance and, energy and average；

(3) described four textural characteristics parameters are normalized, finally obtain normalized texture eigenvalue.

This preferred embodiment is based on improved gray level co-occurrence matrixes method, asks for cytological map by the way of setting weight coefficient The Gray co-occurrence matrix of picture, and then extract textural characteristics on specified four direction for the cell, solve due to outside dry Disturb the textural characteristics ginseng of the cell that (impact that causes as lighting angle when cell image gathers, flowing interference of gas etc.) causes Numerical value has the problem of bigger difference in different directions, improves the precision of cell image texture feature extraction；Selected contrast, Variance and, energy and four textural characteristics of average, eliminate redundancy and the characteristic parameter repeating；To described four textural characteristics ginseng Number is normalized, and the Classification and Identification facilitating follow-up cell image is processed.

In this application scenarios, given threshold t=18, d=3, image denoising effect improves 7% relatively, cell image The extraction accuracy of feature improves 7%.

Application scenarios 4

Referring to Fig. 1, Fig. 2, a kind of combined iris of an embodiment of this application scene and the Photobiology of cortical tissue The optical imaging device that imaging device adopts, including cell recognition module and optical imaging device, described cell recognition module is used To determine biological species, described optical imaging device is made up of light source unit for lighting and image-generating unit, light source unit for lighting with become As unit is configured to multispectral multi-polarization state combination optical imaging system, described multispectral multi-polarization state combination optical is imaged System produces at least four combined imagings, comprising: black light and orthogonal polarisation state combined imaging, black light and parallel polarization State combined imaging, near infrared light and orthogonal polarisation state combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

Preferably, described light source unit for lighting and image-generating unit are combined and are configured to the wave-length coverage of at least black light For 300-500nm, the wave-length coverage of near infrared light is 700-900nm, and the polarizer and analyzer are combined and are configured at least to have Orthogonal with parallel polarization state.

This preferred embodiment image quality is higher.

Preferably, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.

This preferred embodiment areas imaging is wider.

Preferably, described cell recognition module 1 includes Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, classification knowledge Other unit 13；Described Methods of Segmentation On Cell Images unit 11 is used for distinguishing the back of the body in the cell image being gathered by cell image acquisition module Scape, nucleus and Cytoplasm；Described feature extraction unit 12 is used for the textural characteristics of cell image are extracted；Described classification Recognition unit 13 is used for utilizing grader to realize to cell image Classification and Identification according to textural characteristics.

This preferred embodiment constructs the unit structure of cell recognition module 1.

Preferably, described Methods of Segmentation On Cell Images unit 11 includes image conversion subunit, noise remove subelement, coarse segmentation Subelement, nuclear centers demarcate subelement, Accurate Segmentation subelement, particularly as follows:

(1) image conversion subunit, for being converted into gray level image by the cell image of collection；

(2) noise remove subelement, for carrying out denoising to gray level image, comprising:

For pixel (x, y), choose its 3 × 3 neighborhood s_x,y(2n+1) the neighborhood l of × (2n+1)_x,y, n be more than Integer equal to 2；

Whether it is that boundary point judges first to pixel, given threshold t, t ∈ [13,26], calculate pixel (x, y) With its neighborhood s_x,yIn each pixel gray scale difference value, and be compared with threshold value t, if gray scale difference value is more than the number of threshold value t More than or equal to 6, then pixel (x, y) is boundary point, and otherwise, pixel (x, y) is non-boundary point；

If (x, y) is boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{q(i,j)\&element;\[q(x,y)-1.5\mathrm{\σ},q(x,y)+1.5\mathrm{\σ}\]}q(i,j)}{k}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, and q (x, y) is the ash of noise reduction preceding pixel point (x, y) Angle value, σ is pixel (x, y) neighborhood l_x,yInterior gray value mark is poor, and q (i, j) ∈ [q (x, y) -1.5 σ, q (x, y)+1.5 σ] represents Neighborhood l_x,yInterior gray value falls within the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ], and k represents neighborhood l_x,yInterior gray value falls within The quantity of the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ]；

If (x, y) is non-boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)q(i,j)}{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, the ash at q (i, j) representative image midpoint (i, j) place Angle value, w (i, j) is neighborhood l_x,yThe corresponding Gauss weight of interior point (i, j)；

(3) coarse segmentation subelement, for slightly being drawn to the background in the cell image after denoising, Cytoplasm, nucleus Point, particularly as follows:

Each pixel (x, y) is represented with four dimensional feature vectors:

$\stackrel{\→}{u}(x,y)=\[h(x,y),h_{ave}(x,y),h_{med}(x,y),h_{sta}(x,y)\]$

In formula, h (x, y) represents the gray value of (x, y), h_ave(x, y) represents its neighborhood s_x,yGray average, h_med(x, y) generation Table its neighborhood s_x,yGray scale intermediate value, h_sta(x, y) represents its neighborhood s_x,yGray variance；

Background, Cytoplasm, nucleus three class are divided into using k-means clustering procedure；

(4) nuclear centers demarcate subelement, for demarcating to nuclear centers:

Nucleus approximate region is obtained by coarse segmentation subelement, if nuclear area comprises n point: (x₁,y₁),…,(x_n, y_n), this region is carried out with intensity-weighted demarcation and geometric center is demarcated, take its meansigma methods as nuclear centers (x_z,y_z):

$x_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{x}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{x}_i}{n})$

$y_z=\frac{1}{2}(\frac{{\mathrm{\σ}}_{i=1}^n{y}_{i}h(x_i,y_i)}{{\mathrm{\σ}}_{i=1}^{n}h(x_i,y_i)}+\frac{{\mathrm{\σ}}_{i=1}^n{y}_i}{n})$

(5) Accurate Segmentation subelement, for carrying out Accurate Segmentation to nucleus, Cytoplasm；

Build from nuclear centers (x_z,y_z) arrive nucleus and Cytoplasm boundary point (x_p,y_p) directed line segmentDistanceRepresent and round downwards；

Along line segment, sampling is carried out with unit length and can obtain dis_pIndividual pointIf adopting The coordinate of sampling point is not integer, and its gray value is obtained by surrounding pixel linear interpolation；

Point (x_i,y_i) place along line segment direction gray scale difference:

hd(x_i,y_i)=h (x_i-1,y_i-1)-h(x_i,y_i)

Definition gray scale difference inhibition function:

$y\left(x\right)=\left\{\begin{array}{ccc}x& if& x\≤0\\ 0.5x& if& x>0\end{array}\right.$

Point (x_i,y_i) place along line segment direction gradient gra (x_i,y_i):

$gra(x_i,y_i)=\frac{\left|y\left(hd\right(x_i,y_i)\right)|+|y\left(hd\right(x_{i+1},y_{i+!})\left)\right|}{2}$

Choose the maximum value point of gradient as nucleus and cytoplasmic precise edge.

This preferred embodiment arranges noise remove subelement, and effective integration center pixel is closed on the space of neighborhood territory pixel Property and grey similarity carrying out noise reduction process, flat site in the picture, in neighborhood, grey scale pixel value is more or less the same, and adopts Gaussian filter is weighted to gray value filtering, and is changing violent borderline region, row bound keeps filtering, is conducive to image The holding at edge；Nucleus and Cytoplasm coarse contour are extracted using k mean cluster, can effectively remove the interference of noise；Setting is thin Subelement is demarcated at karyon center, is easy to subsequently nucleus and Cytoplasm profile are accurately positioned；Accurate Segmentation subelement fills Divide and make use of directional information, overcome the interference to edge graph for the inflammatory cell, can accurately extract nucleus and Cytoplasm side Edge.

Preferably, the described textural characteristics to cell image extract, comprising:

(1) the Gray co-occurrence matrix of cell image, described comprehensive ash is asked for based on improved gray level co-occurrence matrixes method Degree co-occurrence matrix embodies cell textural characteristics in different directions:

Be located at 0 °, 45 °, 90 °, the gray level co-occurrence matrixes on 135 ° of four directions be respectively h (x, y, d, 0 °), h (x, y, d, 45 °), h (x, y, d, 90 °), h (x, y, d, 135 °), corresponding matrix element project is x₁、x₂、x₃、x₄, then Gray is common The computing formula of raw matrix is:

H (x, y, d)=w₁h(x,y,d,0°)+w₂h(x,y,d,45°)+w₃h(x,y,d,90°)+w₄h(x,y,d,135°)

Gray co-occurrence matrix element number is:

$x=\underset{i=1}{\overset{4}{\σ}}{w}_i{x}_i$

In formula, d represents distance, the span of d is [2,4], w_iFor weight coefficient, i=1,2,3,4, it is by four sides To in each direction on the corresponding contrast level parameter of gray level co-occurrence matrixes calculate, if the gray level co-occurrence matrixes on four direction Corresponding contrast level parameter is respectively d_i, average isI=1,2,3,4, then weight coefficient w_iComputing formula be:

$w_i=\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}/\underset{i=1}{\overset{4}{\σ}}\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}$

(2) four textural characteristics parameters needed for being obtained using described Gray co-occurrence matrix and matrix element project: Contrast, variance and, energy and average；

(3) described four textural characteristics parameters are normalized, finally obtain normalized texture eigenvalue.

This preferred embodiment is based on improved gray level co-occurrence matrixes method, asks for cytological map by the way of setting weight coefficient The Gray co-occurrence matrix of picture, and then extract textural characteristics on specified four direction for the cell, solve due to outside dry Disturb the textural characteristics ginseng of the cell that (impact that causes as lighting angle when cell image gathers, flowing interference of gas etc.) causes Numerical value has the problem of bigger difference in different directions, improves the precision of cell image texture feature extraction；Selected contrast, Variance and, energy and four textural characteristics of average, eliminate redundancy and the characteristic parameter repeating；To described four textural characteristics ginseng Number is normalized, and the Classification and Identification facilitating follow-up cell image is processed.

In this application scenarios, given threshold t=20, d=4, image denoising effect improves 8% relatively, cell image The extraction accuracy of feature improves 6%.

Application scenarios 5

Referring to Fig. 1, Fig. 2, a kind of combined iris of an embodiment of this application scene and the Photobiology of cortical tissue The optical imaging device that imaging device adopts, including cell recognition module and optical imaging device, described cell recognition module is used To determine biological species, described optical imaging device is made up of light source unit for lighting and image-generating unit, light source unit for lighting with become As unit is configured to multispectral multi-polarization state combination optical imaging system, described multispectral multi-polarization state combination optical is imaged System produces at least four combined imagings, comprising: black light and orthogonal polarisation state combined imaging, black light and parallel polarization State combined imaging, near infrared light and orthogonal polarisation state combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

Preferably, described light source unit for lighting and image-generating unit are combined and are configured to the wave-length coverage of at least black light For 300-500nm, the wave-length coverage of near infrared light is 700-900nm, and the polarizer and analyzer are combined and are configured at least to have Orthogonal with parallel polarization state.

This preferred embodiment image quality is higher.

Preferably, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.

This preferred embodiment areas imaging is wider.

Preferably, described cell recognition module 1 includes Methods of Segmentation On Cell Images unit 11, feature extraction unit 12, classification knowledge Other unit 13；Described Methods of Segmentation On Cell Images unit 11 is used for distinguishing the back of the body in the cell image being gathered by cell image acquisition module Scape, nucleus and Cytoplasm；Described feature extraction unit 12 is used for the textural characteristics of cell image are extracted；Described classification Recognition unit 13 is used for utilizing grader to realize to cell image Classification and Identification according to textural characteristics.

This preferred embodiment constructs the unit structure of cell recognition module 1.

Preferably, described Methods of Segmentation On Cell Images unit 11 includes image conversion subunit, noise remove subelement, coarse segmentation Subelement, nuclear centers demarcate subelement, Accurate Segmentation subelement, particularly as follows:

(1) image conversion subunit, for being converted into gray level image by the cell image of collection；

(2) noise remove subelement, for carrying out denoising to gray level image, comprising:

For pixel (x, y), choose its 3 × 3 neighborhood s_x,y(2n+1) the neighborhood l of × (2n+1)_x,y, n be more than Integer equal to 2；

Whether it is that boundary point judges first to pixel, given threshold t, t ∈ [13,26], calculate pixel (x, y) With its neighborhood s_x,yIn each pixel gray scale difference value, and be compared with threshold value t, if gray scale difference value is more than the number of threshold value t More than or equal to 6, then pixel (x, y) is boundary point, and otherwise, pixel (x, y) is non-boundary point；

If (x, y) is boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{q(i,j)\&element;\[q(x,y)-1.5\mathrm{\σ},q(x,y)+1.5\mathrm{\σ}\]}q(i,j)}{k}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, and q (x, y) is the ash of noise reduction preceding pixel point (x, y) Angle value, σ is pixel (x, y) neighborhood l_x,yInterior gray value mark is poor, and q (i, j) ∈ [q (x, y) -1.5 σ, q (x, y)+1.5 σ] represents Neighborhood l_x,yInterior gray value falls within the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ], and k represents neighborhood l_x,yInterior gray value falls within The quantity of the point of interval [q (x, y) -1.5 σ, q (x, y)+1.5 σ]；

If (x, y) is non-boundary point, then carry out following noise reduction process:

$h(x,y)=\frac{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)q(i,j)}{{\mathrm{\σ}}_{(i,j)\&element;l_{x,y}}w(i,j)}$

In formula, h (x, y) is the gray value of pixel (x, y) after noise reduction, the ash at q (i, j) representative image midpoint (i, j) place Angle value, w (i, j) is neighborhood l_x,yThe corresponding Gauss weight of interior point (i, j)；

(3) coarse segmentation subelement, for slightly being drawn to the background in the cell image after denoising, Cytoplasm, nucleus Point, particularly as follows:

Each pixel (x, y) is represented with four dimensional feature vectors:

$\stackrel{\→}{u}(x,y)=\[h(x,y),h_{ave}(x,y),h_{med}(x,y),h_{sta}(x,y)\]$

In formula, h (x, y) represents the gray value of (x, y), h_ave(x, y) represents its neighborhood s_x,yGray average, h_med(x, y) generation Table its neighborhood s_x,yGray scale intermediate value, h_sta(x, y) represents its neighborhood s_x,yGray variance；

Background, Cytoplasm, nucleus three class are divided into using k-means clustering procedure；

(4) nuclear centers demarcate subelement, for demarcating to nuclear centers:

Nucleus approximate region is obtained by coarse segmentation subelement, if nuclear area comprises n point: (x₁,y₁),…,(x_n, y_n), this region is carried out with intensity-weighted demarcation and geometric center is demarcated, take its meansigma methods as nuclear centers (x_z,y_z):

$x_z=\frac{1}{2}(\frac{{\σ}_{i=1}^n{x}_{i}h(x_i,y_i)}{{\σ}_{i=1}^{n}h(x_i,y_i)}+\frac{{\σ}_{i=1}^n{x}_i}{n})$

$y_z=\frac{1}{2}(\frac{{\mathrm{\σ}}_{i=1}^n{y}_{i}h(x_i,y_i)}{{\mathrm{\σ}}_{i=1}^{n}h(x_i,y_i)}+\frac{{\mathrm{\σ}}_{i=1}^n{y}_i}{n})$

(5) Accurate Segmentation subelement, for carrying out Accurate Segmentation to nucleus, Cytoplasm；

Build from nuclear centers (x_z,y_z) arrive nucleus and Cytoplasm boundary point (x_p,y_p) directed line segmentDistanceRepresent and round downwards；

Along line segment, sampling is carried out with unit length and can obtain dis_pIndividual pointIf adopting The coordinate of sampling point is not integer, and its gray value is obtained by surrounding pixel linear interpolation；

Point (x_i,y_i) place along line segment direction gray scale difference:

hd(x_i,y_i)=h (x_i-1,y_i-1)-h(x_i,y_i)

Definition gray scale difference inhibition function:

$y\left(x\right)=\left\{\begin{array}{ccc}x& if& x\≤0\\ 0.5x& if& x>0\end{array}\right.$

Point (x_i,y_i) place along line segment direction gradient gra (x_i,y_i):

$gra(x_i,y_i)=\frac{\left|y\left(hd\right(x_i,y_i)\right)|+|y\left(hd\right(x_{i+1},y_{i+!})\left)\right|}{2}$

Choose the maximum value point of gradient as nucleus and cytoplasmic precise edge.

This preferred embodiment arranges noise remove subelement, and effective integration center pixel is closed on the space of neighborhood territory pixel Property and grey similarity carrying out noise reduction process, flat site in the picture, in neighborhood, grey scale pixel value is more or less the same, and adopts Gaussian filter is weighted to gray value filtering, and is changing violent borderline region, row bound keeps filtering, is conducive to image The holding at edge；Nucleus and Cytoplasm coarse contour are extracted using k mean cluster, can effectively remove the interference of noise；Setting is thin Subelement is demarcated at karyon center, is easy to subsequently nucleus and Cytoplasm profile are accurately positioned；Accurate Segmentation subelement fills Divide and make use of directional information, overcome the interference to edge graph for the inflammatory cell, can accurately extract nucleus and Cytoplasm side Edge.

Preferably, the described textural characteristics to cell image extract, comprising:

(1) the Gray co-occurrence matrix of cell image, described comprehensive ash is asked for based on improved gray level co-occurrence matrixes method Degree co-occurrence matrix embodies cell textural characteristics in different directions:

Be located at 0 °, 45 °, 90 °, the gray level co-occurrence matrixes on 135 ° of four directions be respectively h (x, y, d, 0 °), h (x, y, d, 45 °), h (x, y, d, 90 °), h (x, y, d, 135 °), corresponding matrix element project is x₁、x₂、x₃、x₄, then Gray is common The computing formula of raw matrix is:

H (x, y, d)=w₁h(x,y,d,0°)+w₂h(x,y,d,45°)+w₃h(x,y,d,90°)+w₄h(x,y,d,135°)

Gray co-occurrence matrix element number is:

$x=\underset{i=1}{\overset{4}{\σ}}{w}_i{x}_i$

In formula, d represents distance, the span of d is [2,4], w_iFor weight coefficient, i=1,2,3,4, it is by four sides To in each direction on the corresponding contrast level parameter of gray level co-occurrence matrixes calculate, if the gray level co-occurrence matrixes on four direction Corresponding contrast level parameter is respectively d_i, average isI=1,2,3,4, then weight coefficient w_iComputing formula be:

$w_i=\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}/\underset{i=1}{\overset{4}{\σ}}\frac{1}{|d_i-\stackrel{\&overbar;}{d}|+1}$

(2) four textural characteristics parameters needed for being obtained using described Gray co-occurrence matrix and matrix element project: Contrast, variance and, energy and average；

(3) described four textural characteristics parameters are normalized, finally obtain normalized texture eigenvalue.

This preferred embodiment is based on improved gray level co-occurrence matrixes method, asks for cytological map by the way of setting weight coefficient The Gray co-occurrence matrix of picture, and then extract textural characteristics on specified four direction for the cell, solve due to outside dry Disturb the textural characteristics ginseng of the cell that (impact that causes as lighting angle when cell image gathers, flowing interference of gas etc.) causes Numerical value has the problem of bigger difference in different directions, improves the precision of cell image texture feature extraction；Selected contrast, Variance and, energy and four textural characteristics of average, eliminate redundancy and the characteristic parameter repeating；To described four textural characteristics ginseng Number is normalized, and the Classification and Identification facilitating follow-up cell image is processed.

In this application scenarios, given threshold t=26, d=2, image denoising effect improves 7.5% relatively, cytological map Extraction accuracy as feature improves 8%.

Finally it should be noted that above example is only in order to illustrating technical scheme, rather than the present invention is protected The restriction of shield scope, although having made to explain to the present invention with reference to preferred embodiment, those of ordinary skill in the art should Work as understanding, technical scheme can be modified or equivalent, without deviating from the reality of technical solution of the present invention Matter and scope.

Claims (3)

1. the optical imaging device that the Photobiology imaging device of a kind of combined iris and cortical tissue adopts, is characterized in that, bag Include cell recognition module and optical imaging device, described cell recognition module is used for determining biological species, described optical imagery dress Put and be made up of light source unit for lighting and image-generating unit, light source unit for lighting and image-generating unit are configured to multispectral multi-polarization state group Close optical imaging system, described multispectral multi-polarization state combination optical imaging system produces at least four combined imagings, comprising: Black light and orthogonal polarisation state combined imaging, black light and parallel polarization states combined imaging, near infrared light and cross-polarization State combined imaging, and near infrared light and parallel polarization states combined imaging；

Described multispectral multi-polarization state combination optical imaging system, comprising:

Light source unit for lighting is made up of at least black light, near infrared light light source, and the polarizer；

Image-generating unit is made up of at least black light, near infrared light imaging light path, and analyzer.

2. the light that the Photobiology imaging device of a kind of combined iris according to claim 1 and cortical tissue adopts studies As device, it is characterized in that, described light source unit for lighting and image-generating unit are combined and are configured to the wavelength model of at least black light Enclose for 300-500nm, the wave-length coverage of near infrared light is 700-900nm, the polarizer and analyzer are combined and are configured at least to have There is orthogonal with parallel polarization state.

3. the light that the Photobiology imaging device of a kind of combined iris according to claim 2 and cortical tissue adopts studies As device, it is characterized in that, described multispectral multi-polarization state combination optical imaging system, light source unit for lighting and image-generating unit quilt Increase combination configuration further to include: visible wavelength range is 500-700nm, the polarizer and analyzer are combined and are configured to 45 The polarization state of degree.