CN103106669B - Chinese medicine tongue picture is as environmental suitability color reproduction method - Google Patents

Abstract

Chinese medicine tongue picture belongs to field of medical image processing as environmental suitability color reproduction method.First this method takes special colour table image under same image acquisition condition, and measures the standard colors angle value of color lump in colour table with colorimeter; Use colour table view data to set up gatherer process colorimetric characterization models in a computer, realize the conversion between device color spaces and standard colorimetric space.Then according to measuring the image capture environment and display environment parameter that obtain, in conjunction with the background adaptation parameter that tongue image structure priori is determined, visual adaptation model is utilized to calculate display image chroma tristimulus values.Similar with gatherer process characterization, by showing one group of color lump over the display, and measuring the standard colors angle value of display color lump, setting up procedure for displaying colorimetric characterization models, calculate the display driving value that colourity tristimulus values is corresponding.This method finally realizes the adaptability true colors reproduction of tongue image.<!--1-->

Description

Chinese medicine tongue picture is as environmental suitability color reproduction method

Technical field

The invention belongs to field of medical image processing.Be by a kind ofly utilizing computer technology, digital image processing techniques realize the method that image true colors reproduces and are applied to Tongue field.During eye-observation color, there is adaptivity, can according to different comprising illumination, environmental factor is adjusted to the most real visually-perceptible automatically.But in the research of traditional Chinese medical science objectivity, because the impact comprising device attribute and environment of observation factor causes tongue image color and the inconsistent problem of Human Perception.The method by carrying out characterization for image input-output device, and uses the method for colour vision sensor model to comprising illumination, background, environment of observation factor take in, and reappears the true colors of tongue image.

The novel tongue picture research method of one grown up the eighties in last century with Digital Image Processing and analytical technology for Main Means carries out the research of Evolution of Tongue Inspection of TCM objectivity.Using the tongue image after color reproduction to carry out treatment and analysis can than more comprehensively studying tongue picture feature.

Background technology

Traditional Evolution of Tongue Inspection of TCM is by the knowledge and experience eye observation of doctor according to oneself, and judges, and its diagnostic result is by the restriction of doctor's know-how, the mode of thinking and diagnostic skill.Therefore, to the objectifying of Evolution of Tongue Inspection of TCM, digitizing, namely objective tongue picture analysis method is studied, to Chinese medical discrimination standardization and Evolution of Tongue Inspection of TCM is clinical, teaching, scientific research method modernization there is important theory value and practical significance.Various vision facilities collection and display tongue image is often adopted in Externalization of Application of Tongue Inspection of TCM process.Due to vision facilities self attributes and environment of observation different, cause tongue image color that display shows and the tongue color that human eye is seen is inconsistent, have impact on the judgement of doctor to the state of an illness.

Externalization of Application of Tongue Inspection of TCM research adopts maximum input-output device to be digital camera and liquid crystal display.Vision facilities color space has device dependency, is the main cause of cross-color.First be device characterization---device color spaces and standard colorimetric space are set up mapping relations, re-uses visual adaptation conversion and revise the impact comprising illumination, contextual factor, finally realize true colors reproduction.Device characterization is the key of color reproduction effect.First the nonlinear transformation relation of apparatus for establishing color space RGB and each passage luminous energy Y, then realizes according to digital camera characterization model, colored quantum noise, liquid crystal display characterization model algorithm and optimizes mapping, finally obtains reappearing image.This mapping process considers device attribute, environmental factor etc., and make tongue image color closer to true colors, observer can judge the state of an illness accurately according to tongue image.

Summary of the invention

The present invention is in Tongue field, proposes for tongue image cross-color.Its feature is: on the colorimetric characterization analysis foundation of image acquisition and procedure for displaying, in conjunction with human eye vision adaptive model and Chinese medicine tongue picture structure priori, soil boy structure adaptability Chinese medicine tongue picture is as color reproduction system.The main flow of this method as shown in Figure 1.

First this method takes special colour table image under same image acquisition condition, and measures the standard colors angle value of color lump in colour table with colorimeter; Use colour table view data to set up gatherer process colorimetric characterization models in a computer, realize the conversion between device color spaces and standard colorimetric space.Then according to measuring the image capture environment and display environment parameter that obtain, in conjunction with the background adaptation parameter that tongue image structure priori is determined, visual adaptation model is utilized to calculate display image chroma tristimulus values.Similar with gatherer process characterization, by showing one group of color lump over the display, and measuring the standard colors angle value of display color lump, setting up procedure for displaying colorimetric characterization models, calculate the display driving value that colourity tristimulus values is corresponding.Finally realize the adaptability true colors reproduction of tongue image.This algorithm comprises the steps:

Chinese medicine tongue picture, as environmental suitability color reproduction method, is characterized in that, comprises the steps:

A. the foundation of image acquisition characterization model

Standard colour board is placed in image capture environment, shooting colour table image, and obtains the RGB mean value of each color lump; Use colorimeter measurement obtain each color lump tristimulus values XYZ and tristimulus values normalization is made spectral reflectivity be 1 white blocks Y value be 100;

Unexplained reference in following steps, all tristimulus values be spectral reflectivity be 1 white blocks Y value be the normalized value of 100;

According to the Y in the RGB mean value of gray scale color lump each in colour table and tristimulus values, set up the one-dimensional map look-up table of R, G, B triple channel and Y respectively by cubic spline interpolation algorithm; Use this one-dimensional map look-up table that the original RGB mean value of above-mentioned all color lumps is converted into the linear color space value with brightness, be designated as rgb;

And adopting polynomial regression model to set up the mapping relations of rgb and XYZ, the several pre-measuring colour difference according to test sample book of Polynomial Terms is chosen; Like this, the three-channel one-dimensional map look-up table of R, G, B and polynomial regression model are combined into image acquisition characterization model; The each pixel RGB values of original tongue image, after image acquisition characterization model calculates, is namely converted to each pixel tristimulus values XYZ;

B. the visual adaptation based on tongue image priori converts

1) determination of image capture environment adaptation parameter

(1) determine to adapt to field brightness L _a: refer to the absolute brightness of observing background, using blee estimated brightness as adaptation field brightness; First the absolute brightness L of photometer measurement reference white color lump in image capture environment is used _w, calculate white absolute brightness L _wwith the Y recorded with colorimeter in white tristimulus values _wratio cc; Former tongue image selects a colour of skin point, according to the Y value in this tristimulus that gatherer process characterization model calculates, calculates the product of Y and α, give L _a;

(2) background relative brightness Y is determined _b: represent the brightness that background relative reference is white, using the Y in blee tristimulus values as Y _b;

(3) parameter factors of environmental correclation is determined: refer to the parameter that the outer lighting environment of the entirety of the object of observation is relevant, comprise environmental impact parameter c, color induction factor Nc, fitness factor F, three parameters are set to 0.69,1.0,1.0 respectively;

2) forward adapts to conversion

Forward adapts to conversion and refers to according to collection environmental adaptation parameter, colour stimulus value is converted to the visual experience value after environmental adaptation, i.e. the process of look looks attribute; According to the above-mentioned collection environmental adaptation parameter determined, each pixel tristimulus values XYZ that A step gatherer process characterization model is calculated and the tristimulus values X of reference white _wy _wz _w, by colored quantum noise positive-going transition in colorimetry, calculate the look looks attribute that these colour stimulus are corresponding, comprise lightness J, chroma C and hue angle h;

3) determination of display environment adaptation parameter

(1) determine to adapt to field brightness : display white block first over the display, by its absolute brightness of photometer measurement , and record the Y in white blocks tristimulus values with colorimeter, both calculating ratio ; Former tongue image selects a colour of skin point, according to the Y value in this tristimulus that gatherer process characterization model calculates, calculates the product of Y and β, give ;

(2) background relative brightness is determined : using the Y in blee tristimulus values as ;

(3) determine the parameter factors of environmental correclation: environmental selection environmental impact parameter c, color induction factor Nc, fitness factor F setting parameter residing for display, dark room conditions above-mentioned parameter is set to 0.525,0.8,0.8 respectively; Duskiness environment is set to 0.59,0.95,0.9; General environment is set to 0.69,1.0,1.0;

(4) reverse adaptation conversion

Reverse adaptation conversion refers to according to display environment adaptation parameter, visual experience value is converted to the process of display environment colour stimulus value; According to display environment adaptation parameter, forward being adapted to convert the look looks property value obtained, calculating each pixel tristimulus values for showing by colored quantum noise reciprocal transformation;

Thus, each pixel tristimulus values XYZ that gatherer process characterization model exports adapts to conversion through forward and is converted to visual experience look looks property value, then is converted to the tristimulus values of display needs display through reverse adaptation conversion ;

C. the foundation of display characteristic model

By RGB color space from 0 to 255 equal interval samplings, entreat region to show each RGB color successively in the display, use colorimeter to measure each color tristimulus values XYZ;

Namely only have this channel value non-vanishing each for RGB triple channel pure color, tristimulus values substitutes into the XYZ of following formula, by the linear rgb value that matrix computations pure color is corresponding, and X in formula _{i, max}, Y _{i, max}, Z _{i, max}, wherein i be R G B be single channel maximum export time XYZ, X _{k, max}, Y _{k, max}, Z _{k, max}for stain, each channels drive value is the XYZ value of 0; Each pure color and RGB tri-kinds of pure colors, i.e. three kinds of passages, but each passage can value various, as R passage 50,0,0; 100,0,0 etc.;

$\left[\begin{array}{c}r\\ g\\ b\end{array}\right]={\left[\begin{array}{ccc}{X}_{R,\max }?X_{k,\min }& X_{G,\max }?X_{k,\min }& X_{B,\max }?X_{k,\min }\\ Y_{R,\max }?Y_{k,\min }& Y_{G,\max }?Y_{k,\min }& Y_{B,\max }?Y_{k,\min }\\ Z_{R,\max }?Z_{k,\min }& Z_{G,\max }?Z_{k,\min }& Z_{B,\max }?Z_{k,\min }\end{array}\right]}^{?1}\left[\begin{array}{c}X?X_{k,\min }\\ Y?Y_{k,\min }\\ Z?Z_{k,\min }\end{array}\right]$

Set up the initial one-dimensional map look-up table of three passages by interpolation fitting between each passage pure color RGB motivation value with corresponding linear rgb value;

Utilize initial one-dimensional map look-up table, all color lump RGB motivation values are converted into linear rgb value, the transformation matrix of recycling following formula calculates corresponding prediction tristimulus X ＇ Y ＇ Z ＇;

$\left[\begin{array}{c}{X}^{\&prime;}\\ Y^{\&prime;}\\ Z^{\&prime;}\end{array}\right]=\left[\begin{array}{cccc}{X}_{R,\max }?X_{k,\min }& X_{G,\max }?X_{k,\min }& X_{B,\max }?X_{k,\min }& X_{k,\min }\\ Y_{R,\max }?Y_{k,\min }& Y_{G,\max }?Y_{k,\min }& Y_{B,\max }?Y_{k,\min }& Y_{k,\min }\\ Z_{R,\max }?Z_{k,\min }& Z_{G,\max }?Z_{k,\min }& Z_{B,\max }?Z_{k,\min }& Z_{k,\min }\end{array}\right]\left[\begin{array}{c}r\\ g\\ b\\ 1\end{array}\right]$

Employing standard aberration formulae discovery prediction tristimulus values and the average color difference measured between tristimulus values, carry out iteration optimization as objective function, progressive updating one-dimensional look-up table and transformation matrix are until objective function is minimum; One-dimensional look-up table final like this and transformation matrix just constitute the characterization model of display, achieve the mapping between display driving value and display tristimulus values;

The tristimulus values of display is needed according to the display obtained after the conversion of previous step visual adaptation , by display characteristic model, calculate the corresponding display driving value of each pixel, obtain final reproduction tongue image.

Beneficial effect

This method is intended to realize tongue image energy accurate reproduction tongue body true colors in Externalization of Application of Tongue Inspection of TCM research, namely no matter uses which kind of vision facilities and can reproduce true tongue color accurately under which kind of illumination with environment of observation.By making accurate vision facilities characterization model, contacting of apparatus for establishing color space and standard colorimetric space, and then the cross-color reducing that device dependency causes; Use the factors such as visual adaptation transfer pair lighting source, background, environment of observation to compensate and revise simultaneously, make reproduction color more meet human-eye visual characteristic.The reproduction tongue image relatively exported and true tongue body can find, the image that this method obtains and true tongue color very close, subjective evaluation result is better than colourity recurrence system.

Example effects: by some observers to the visual experience of primitive color with reproduction color similarity, provide corresponding score value, as table 1, finally will quantize mean value as last evaluation result.

Table 1 subjective assessment grade quantizing table

Respectively the original image shown, colourity reproduction image, look looks reproduction image are carried out subjectivity marking and evaluated.The mean subjective evaluation of three width images is respectively: 4.1,2.6,2.1, and original image belongs to " can accept " grade, and colourity reproduction image is between " similar " and " fairly similar ", and look looks reappear image close to " fairly similar ".

Accompanying drawing explanation

Fig. 1 is the main program flow chart of this method;

Fig. 2 is collecting device R passage one-dimensional look-up table curve exemplary plot;

Fig. 3 is the program flow diagram of gatherer process characterization process;

Fig. 4 is the program flow diagram of positive-going transition in visual adaptation model;

Fig. 5 is display one-dimensional look-up table curve exemplary plot;

Fig. 6 is the program flow diagram of display characteristics process.

Embodiment

This implementation method completes in the Tongue collection be made up of standard D65 light source, digital camera, mechanical platform, computer system and display system.Specific implementation process is as follows:

1. the foundation of image acquisition characterization model

1) Standard colour board (as color checker24) is placed in tongue image acquisition system, uses the digital camera shooting colour table image of system self, and extract the RGB mean value of each color lump.Use colorimeter PR650 measurement to obtain the tristimulus values XYZ of each color lump, and tristimulus values normalization is made spectral reflectivity be 1 white blocks Y value be 100.

2) set up sampled point according to the Y in the RGB mean value of gray scale color lump each in colour table and tristimulus values, re-use cubic spline interpolation algorithm and set up R, G, B triple channel and the complete one-dimensional look-up table curve of Y on 0-255 respectively.

3) the RGB mean value of all color lumps input one-dimensional look-up table is obtained linear camera response, be designated as rgb; Then use polynomial regression algorithm to set up the mapping relations of rgb and XYZ, the several pre-measuring colour difference according to test sample book of Polynomial Terms is chosen, and the present embodiment adopts 11.Gather characterization model to complete.

4) each pixel RGB values of original tongue image is after image acquisition characterization model calculates, and namely can be converted to each pixel tristimulus values XYZ.

2. the visual adaptation based on tongue image priori converts

1) determination of image capture environment adaptation parameter

(1) field brightness L is adapted to _a: with the absolute brightness L of photometer measurement standard white color lump in image capture environment _w, calculate white absolute brightness L _wwith the Y recorded with colorimeter in white tristimulus values _wratio cc; Former tongue image selects a colour of skin point, according to the Y value in this tristimulus that gatherer process characterization model calculates, calculates the product of Y and α, give L _a.

(2) background relative brightness Y _b: using the Y in blee point tristimulus values as Y _b.

(3) parameter factors of environmental correclation: environmental impact parameter c, color induction factor Nc, fitness factor F tri-parameter are set to 0.69,1.0,1.0 respectively.

2) forward adapts to conversion

(1) by the tristimulus values X of each pixel tristimulus values XYZ and reference white _wy _wz _wbe transformed in cone cellular response space by chromatic adaptation transformation matrix and represent, be designated as (R, G, B) and (R respectively _w, G _w, B _w), namely

$\left[\begin{array}{c}R\\ G\\ B\end{array}\right]=M_{\mathrm{CAT}02}\left[\begin{array}{c}X\\ Y\\ Z\end{array}\right]---\left(1\right)$

$\left[\begin{array}{c}{R}_W\\ G_W\\ B_W\end{array}\right]=M_{\mathrm{CAT}02}\left[\begin{array}{c}{X}_W\\ Y_W\\ Z_W\end{array}\right]---\left(2\right)$

$M_{\mathrm{CAT}02}=\left(\begin{array}{ccc}0.7328& 0.4296& -0.1624\\ -0.7036& 1.6975& 0.0061\\ 0.0030& 0.0136& 0.9834\end{array}\right)---\left(3\right)$

(2) fitness D is calculated.D characterizes human eye to the adaptedness of reference white field, and the computing formula of D is:

$D=F[1-(1/3.6)e^{(-L_A-42)/92}]---\left(4\right)$

(3) the cone cellular response (R after adapting to is obtained by weighting chromatic adaptation conversion _c, G _c, B _c):

$\left\{\begin{array}{c}{R}_C=[Y_{W}D/R_W+(1-D)]R\\ G_C=[Y_{W}D/G_W+(1-D)]G\\ B_C=[Y_{W}D/B_W+(1-D)]B\end{array}\right.---\left(5\right)$

(4) luminance level Adaptation factor is calculated.

$F_L=0.2\&times;k^4\left(5L_A\right)+0.1\&times;{(1-k^4)}^2{\left(5L_A\right)}^{1/3}---\left(6\right)$

$k=1/(5L_A+1)---\left(7\right)$

(5) luminance background factor N is calculated _bb, chroma background gactor N _cb, based on the nonlinear factor z of index, computing formula is as follows:

$N_{\mathrm{bb}}=N_{\mathrm{cb}}=0.725{(1/n)}^{0.2}---\left(8\right)$

$z=1.48+\sqrt{n}---\left(9\right)$

$n=Y_b/Y_w---\left(10\right)$

In formula, Y _bfor background luminance, Y _wfor reference white brightness, n is called background inducible factor.

(6) by (R _c, G _c, B _c) respond spatial alternation to HPE space from cones, namely

$\left[\begin{array}{c}{R}^{\&prime;}\\ C^{\&prime;}\\ B^{\&prime;}\end{array}\right]=M_{\mathrm{HPE}}{M^{-1}}_{\mathrm{CAT}02}\left[\begin{array}{c}{R}_C\\ G_C\\ B_C\end{array}\right]---\left(11\right)$

$M_{\mathrm{HPE}}=\left(\begin{array}{ccc}0.38971& 0.68898& -0.07868\\ -0.22981& 1.8340& 0.04641\\ 0.00000& 0.00000& 1.00000\end{array}\right)---\left(12\right)$

${M^{-1}}_{\mathrm{CAT}02}=\left(\begin{array}{ccc}1.0961& -0.2789& 0.1827\\ 0.4544& 0.4735& 0.0721\\ -0.0096& -0.0057& 1.0153\end{array}\right)---\left(13\right)$

(7) in HPE space, non-linear compression process is carried out to (R ＇, G ＇, B ＇), namely

$\left\{\begin{array}{c}{R}_a^{\&prime;}=\frac{400{(F_L{R}^{\&prime;}/100)}^{0.42}}{27.13+{(F_L{R}^{\&prime;}/100)}^{0.42}}+0.1\\ G_a^{\&prime;}=\frac{400{(F_L{G}^{\&prime;}/100)}^{0.42}}{27.13+{(F_L{G}^{\&prime;}/100)}^{0.42}}+0.1\\ B_a^{\&prime;}=\frac{400{(F_L{B}^{\&prime;}/100)}^{0.42}}{27.13+{(F_L{B}^{\&prime;}/100)}^{0.42}}+0.1\end{array}\right.---\left(1\right)$

(8) look looks attribute is calculated

Hue angle h is

$h=\arctan (b/a)---\left(15\right)$

Wherein chromaticity coordinates a, b is

$a=R_a^{\&prime;}-12G_a^{\&prime;}/11+B_a^{\&prime;}/11---\left(16\right)$

$b=(1/9)(R_a^{\&prime;}+G_a^{\&prime;}-2B_a^{\&prime;})---\left(17\right)$

Lightness J is

$J=100{(A/A_W)}^{\mathrm{cz}}---\left(18\right)$

Wherein netrual colour response A is

$A=[2R_a^{\&prime;}+G_a^{\&prime;}+(1/20)B_a^{\&prime;}-0.305]N_{\mathrm{bb}}---\left(19\right)$

A _wit is white corresponding netrual colour response.

Chroma C is

$c=t^{0.9}\sqrt{J/100}{(1.64-{0.29}^n)}^{0.73}---\left(20\right)$

Wherein the computing method of temporary variable t are

$t=\frac{e\sqrt{a^2+b^2}}{R_a^{\&prime;}+G_a^{\&prime;}+(21/20)B_a^{\&prime;}}---\left(21\right)$

$e=(12500N_c{N}_{\mathrm{cb}}/13)[\cos (\mathrm{h\&pi;}/180+2)+3.8]---\left(22\right)$

So far, lightness J, the chroma C of look looks attribute and hue angle h obtain.

3) determination of display environment adaptation parameter

(1) field brightness is adapted to : show the white blocks driven by 255 over the display, by its absolute brightness of photometer measurement, record the Y in white blocks tristimulus values with colorimeter, both calculating ratio ; Former tongue image selects a colour of skin point, according to the Y value in this tristimulus that gatherer process characterization model calculates, calculates the product of Y and β, give .

(2) background relative brightness : using the Y in blee point tristimulus values as .

(3) parameter factors of environmental correclation: environmental selection environmental impact parameter c, color induction factor Nc, fitness factor F setting parameter residing for display.Dark room conditions above-mentioned parameter is set to 0.525,0.8,0.8 respectively; Duskiness environment is set to 0.59,0.95,0.9; General environment is set to 0.69,1.0,1.0.

4) reverse adaptation conversion

(1) adapted to convert J, C, h of obtaining by forward, calculate t, e, p ₁, p ₂, p ₃(non-formula illustrates that parameter calculates according to forward adaption formula)

$t={\left[\frac{C}{\sqrt{J/100}{(1.64-{0.29}^n)}^{0.73}}\right]}^{1/0.9}---\left(22\right)$

$e=(12500N_c{N}_{\mathrm{cb}}/13)[\cos (\mathrm{h\&pi;}/180+2)+3.8]---\left(23\right)$

$A=A_w{(J/100)}^{1/\left(\mathrm{ca}\right)}$

$p_1=e/t,p_2=(A/N_{\mathrm{bb}})+0.305,p_3=21/20$

(2) a and b is calculated

If t=0, then a=b=0, leaps to next step.It should be noted that h should convert radian to when calculating sine and cosine.

If | sin (h) |>=| cos (h) |, then p ₄=p ₁/ sin (h)

$b=\frac{p_2(2+p_3)(460/1403)}{p_4+(2+p_3)(220/1403)-[\cos \left(h\right)/\sin \left(h\right)]-27/1403+p_3(6300/1403)}---\left(25\right)$

$a=b[\cos \left(h\right)/\sin \left(h\right)]---\left(26\right)$

If | cos (h) | < | sin (h) |, then p ₅=p ₁/ cos (h)

$a=\frac{p_2(2+p_3)(460/1403)}{p_5+(2+p_3)(220/1403)-[27/1403-p_3(6300/1403)][\sin \left(h\right)/\cos \left(h\right)]}---\left(27\right)$

$b=a[\sin \left(h\right)/\cos \left(h\right)]---\left(28\right)$

(3) R ＇ is calculated _a, G ＇ _a, B ＇ _a

$\left\{\begin{array}{c}{R}_a^{\&prime;}=\frac{460}{1403}{p}_2+\frac{451}{1403}a+\frac{288}{1403}b\\ G_a^{\&prime;}=\frac{460}{1403}{p}_2-\frac{891}{1403}a-\frac{261}{1403}b\\ B_a^{\&prime;}=\frac{460}{1403}{p}_2+\frac{220}{1403}a-\frac{6300}{1403}b\end{array}\right.---\left(29\right)$

(4) R ＇, G ＇, B ＇ is calculated

$\left\{\begin{array}{c}{R}^{\&prime;}=\mathrm{sign}(R_a^{\&prime;}-0.1)\frac{100}{F_L}{\left[\frac{27.13|R_a^{\&prime;}-0.1|}{400-|R_a^{\&prime;}-0.1|}\right]}^{1/0.42}\\ G^{\&prime;}=\mathrm{sign}(G_a^{\&prime;}-0.1)\frac{100}{F_L}{\left[\frac{27.13|G_a^{\&prime;}-0.1|}{400-|G_a^{\&prime;}-0.1|}\right]}^{1/0.42}\\ B^{\&prime;}=\mathrm{sign}(B_a^{\&prime;}-0.1)\frac{100}{F_L}{\left[\frac{27.13|B_a^{\&prime;}-0.1|}{400-|B_a^{\&prime;}-0.1|}\right]}^{1/0.42}\end{array}\right.---\left(30\right)$

(5) R is calculated _c, G _c, B _c

$\left[\begin{array}{c}{R}_c\\ G_c\\ B_c\end{array}\right]=M_{\mathrm{CAT}02}{M}_{\mathrm{HPE}}^{-1}\left[\begin{array}{c}{R}^{\&prime;}\\ G^{\&prime;}\\ B^{\&prime;}\end{array}\right]---\left(31\right)$

$M_{\mathrm{HPE}}^{-1}=\left(\begin{array}{ccc}1.910197& -1.112124& 0.201908\\ 0.370950& 0.629054& -0.000008\\ 0.00000& 0.00000& 1.000000\end{array}\right)$

(6) cone cellular response R, G, B is transformed into

$\left\{\begin{array}{c}R=R_C/(Y_{W}D{R}_W+1-D)\\ G=G_C/(Y_W{\mathrm{DG}}_W+1-D)\\ B=B_C/(Y_W{\mathrm{DB}}_W+1-D)\end{array}\right.---\left(32\right)$

(7) tristimulus values X, Y, Z is calculated

$\left[\begin{array}{c}X\\ Y\\ Z\end{array}\right]=M_{\mathrm{CAT}02}^{-1}\left[\begin{array}{c}R\\ G\\ B\end{array}\right]---\left(33\right)$

By reciprocal transformation, calculate each pixel tristimulus values for showing.

Thus, each pixel tristimulus values XYZ that gatherer process characterization model exports adapts to conversion through forward and is converted to visual experience look looks property value, then is converted to the tristimulus values of display needs display through reverse adaptation conversion .

3. the foundation of display characteristic model

1) colour temperature of liquid crystal display is set to 6500K, 2 hours preheating time, measures and carries out in dark room conditions.By RGB color space from 0 to 255 equal interval samplings, as by 5 × 5 × 5 deciles, 216 the sampled point rgb values being spaced apart 50 can be obtained; Entreat the region of 10cm × 10cm to show each RGB color successively in the display, use colorimeter to measure each color tristimulus values XYZ.

2) each pure color (only have this channel value non-vanishing) tristimulus values is substituted into the XYZ of following formula, by the linear rgb value that matrix computations pure color is corresponding, X in formula _{i, max}, Y _{i, max}, Z _{i, max}(i be R G B) be single channel maximum export time XYZ, X _{k, max}, Y _{k, max}, Z _{k, max}for the XYZ value of stain (each channels drive value is 0).

$\left[\begin{array}{c}r\\ g\\ b\end{array}\right]={\left[\begin{array}{ccc}{X}_{R,\max }?X_{k,\min }& X_{G,\max }?X_{k,\min }& X_{B,\max }?X_{k,\min }\\ Y_{R,\max }?Y_{k,\min }& Y_{G,\max }?Y_{k,\min }& Y_{B,\max }?Y_{k,\min }\\ Z_{R,\max }?Z_{k,\min }& Z_{G,\max }?Z_{k,\min }& Z_{B,\max }?Z_{k,\min }\end{array}\right]}^{?1}\left[\begin{array}{c}X?X_{k,\min }\\ Y?Y_{k,\min }\\ Z?Z_{k,\min }\end{array}\right]---\left(34\right)$

3) between each passage pure color RGB motivation value with corresponding linear rgb value, the initial one-dimensional map look-up table of three passages is set up by interpolation fitting.

4) utilize initial one-dimensional map look-up table, all color lump RGB motivation values are converted into linear rgb value, the transformation matrix of recycling following formula calculates corresponding prediction tristimulus X ＇ Y ＇ Z ＇.

$\left[\begin{array}{c}{X}^{\&prime;}\\ Y^{\&prime;}\\ Z^{\&prime;}\end{array}\right]=\left[\begin{array}{cccc}{X}_{R,\max }?X_{k,\min }& X_{G,\max }?X_{k,\min }& X_{B,\max }?X_{k,\min }& X_{k,\min }\\ Y_{R,\max }?Y_{k,\min }& Y_{G,\max }?Y_{k,\min }& Y_{B,\max }?Y_{k,\min }& Y_{k,\min }\\ Z_{R,\max }?Z_{k,\min }& Z_{G,\max }?Z_{k,\min }& Z_{B,\max }?Z_{k,\min }& Z_{k,\min }\end{array}\right]\left[\begin{array}{c}r\\ g\\ b\\ 1\end{array}\right]---\left(35\right)$

5) adopt CIE2000 colour difference formula, computational prediction tristimulus values and the average color difference measured between tristimulus values, carry out iteration optimization as objective function, progressive updating one-dimensional look-up table and transformation matrix are until objective function is minimum.One-dimensional look-up table final like this and transformation matrix just constitute the characterization model of display.

6) tristimulus values of display is needed according to the display obtained after the conversion of previous step visual adaptation , by display characteristic model, the corresponding display driving value of each pixel can be calculated, obtain final reproduction tongue image.

Claims (1)

1. Chinese medicine tongue picture is as environmental suitability color reproduction method, it is characterized in that, comprises the steps:

A. the foundation of image acquisition characterization model

Standard colour board is placed in image capture environment, shooting colour table image, and obtains the RGB mean value of each color lump; Use colorimeter measurement obtain each color lump tristimulus values XYZ and tristimulus values normalization is made spectral reflectivity be 1 white blocks Y value be 100;

Unexplained reference in following steps, all tristimulus values be spectral reflectivity be 1 white blocks Y value be the normalized value of 100;

According to the Y in the RGB mean value of gray scale color lump each in colour table and tristimulus values, set up the one-dimensional map look-up table of R, G, B triple channel and Y respectively by cubic spline interpolation algorithm; Use this one-dimensional map look-up table that the original RGB mean value of above-mentioned all color lumps is converted into the linear color space value with brightness, be designated as rgb;

And adopting polynomial regression model to set up the mapping relations of rgb and XYZ, the several pre-measuring colour difference according to test sample book of Polynomial Terms is chosen; Like this, the three-channel one-dimensional map look-up table of R, G, B and polynomial regression model are combined into image acquisition characterization model; The each pixel RGB values of original tongue image, after image acquisition characterization model calculates, is namely converted to each pixel tristimulus values XYZ;

B. the visual adaptation based on tongue image priori converts

1) determination of image capture environment adaptation parameter

(1) determine to adapt to field brightness L _a: refer to the absolute brightness of observing background, using blee estimated brightness as adaptation field brightness; First the absolute brightness L of photometer measurement reference white color lump in image capture environment is used _w, calculate white absolute brightness L _wwith the Y recorded with colorimeter in white tristimulus values _wratio cc; Former tongue image selects a colour of skin point, according to the Y value in this tristimulus that image acquisition characterization model calculates, calculates the product of Y and α, give L _a;

(2) background relative brightness Y is determined _b: represent the brightness that background relative reference is white, using the Y in blee tristimulus values as Y _b;

(3) parameter factors of environmental correclation is determined: refer to the parameter that the outer lighting environment of the entirety of the object of observation is relevant, comprise environmental impact parameter c, color induction factor Nc, fitness factor F, three parameters are set to 0.69,1.0,1.0 respectively;

2) forward adapts to conversion

Forward adapts to conversion and refers to according to collection environmental adaptation parameter, colour stimulus value is converted to the visual experience value after environmental adaptation, i.e. the process of look looks attribute; According to the above-mentioned collection environmental adaptation parameter determined, each pixel tristimulus values XYZ that A step image acquisition characterization model is calculated and the tristimulus values X of reference white _wy _wz _w, by colored quantum noise positive-going transition in colorimetry, calculate the look looks attribute that these colour stimulus are corresponding, comprise lightness J, chroma C and hue angle h;

3) determination of display environment adaptation parameter

(1) determine to adapt to field brightness first display white block over the display, by its absolute brightness of photometer measurement and record the Y in white blocks tristimulus values with colorimeter, both calculating ratio former tongue image selects a colour of skin point, according to the Y value in this tristimulus that image acquisition characterization model calculates, calculates the product of Y and β, give

(2) background relative brightness is determined using the Y in blee tristimulus values as

(3) determine the parameter factors of environmental correclation: environmental selection environmental impact parameter c, color induction factor Nc, fitness factor F setting parameter residing for display, dark room conditions above-mentioned parameter is set to 0.525,0.8,0.8 respectively; Duskiness environment is set to 0.59,0.95,0.9; General environment is set to 0.69,1.0,1.0;

(4) reverse adaptation conversion

Reverse adaptation conversion refers to according to display environment adaptation parameter, visual experience value is converted to the process of display environment colour stimulus value; According to display environment adaptation parameter, forward being adapted to convert the look looks property value obtained, calculating each pixel tristimulus values for showing by colored quantum noise reciprocal transformation;

Thus, each pixel tristimulus values XYZ that image acquisition characterization model exports adapts to conversion through forward and is converted to visual experience look looks property value, then is converted to the tristimulus values of display needs display through reverse adaptation conversion

C. the foundation of display characteristic model

By RGB color space from 0 to 255 equal interval samplings, entreat region to show each RGB color successively in the display, use colorimeter to measure each color tristimulus values XYZ;

Tristimulus values corresponding for each for RGB triple channel pure color is substituted into the XYZ of following formula, the channel value that each pure color of RGB triple channel and RGB triple channel only have pure color corresponding is non-vanishing, by the linear rgb value that matrix computations pure color is corresponding, and X in formula _{i, max}, Y _{i, max}, Z _{i, max}, wherein i be R G B be single channel maximum export time XYZ, X _{k, min}, Y _{k, min}, Z _{k, min}for stain, each channels drive value is the XYZ value of 0; Each pure color and RGB tri-kinds of pure colors, i.e. three kinds of passages;

$\left[\begin{array}{c}r\\ g\\ b\end{array}\right]={\left[\begin{array}{ccc}{X}_{R,\max }-X_{k,\min }& X_{G,\max }-X_{k,\min }& X_{B,\max }-X_{k,\min }\\ Y_{R,\max }-Y_{k,\min }& Y_{G,\max }-Y_{k,\min }& Y_{B,\max }-Y_{k,\min }\\ Z_{R,\max }-Z_{k,\min }& Z_{G,\max }-Z_{k,\min }& Z_{B,\max }-Z_{k,\min }\end{array}\right]}^{-1}\left[\begin{array}{c}X-X_{k,\min }\\ Y-Y_{k,\min }\\ Z-Z_{k,\min }\end{array}\right]$

Set up the initial one-dimensional map look-up table of three passages by interpolation fitting between each passage pure color RGB motivation value with corresponding linear rgb value;

Utilize initial one-dimensional map look-up table, all color lump RGB motivation values are converted into linear rgb value, the transformation matrix of recycling following formula calculates corresponding prediction tristimulus X ' Y ' Z ';

$\left[\begin{array}{c}{X}^{\&prime;}\\ Y^{\&prime;}\\ Z^{\&prime;}\end{array}\right]=\left[\begin{array}{cccc}{X}_{R,\max }-X_{k,\min }& X_{G,\max }-X_{k,\min }& X_{B,\max }-X_{k,\min }& X_{k,\min }\\ Y_{R,\max }-Y_{k,\min }& Y_{G,\max }-Y_{k,\min }& Y_{B,\max }-Y_{k,\min }& Y_{k,\min }\\ Z_{R,\max }-Z_{k,\min }& Z_{G,\max }-Z_{k,\min }& Z_{B,\max }-Z_{k,\min }& Z_{k,\min }\end{array}\right]\left[\begin{array}{c}r\\ g\\ b\\ 1\end{array}\right]$

Employing standard aberration formulae discovery prediction tristimulus values and the average color difference measured between tristimulus values, carry out iteration optimization as objective function, progressive updating one-dimensional look-up table and transformation matrix are until objective function is minimum; One-dimensional look-up table final like this and transformation matrix just constitute the characterization model of display, achieve the mapping between display driving value and display tristimulus values;

The tristimulus values of display is needed according to the display obtained after the conversion of previous step visual adaptation by display characteristic model, calculate the corresponding display driving value of each pixel, obtain final reproduction tongue image.