CN100412906C

CN100412906C - Method for correcting digital tongue picture colour cast

Info

Publication number: CN100412906C
Application number: CNB2006101138705A
Authority: CN
Inventors: 白净; 张永红; 吴佳; 崔珊珊; 孙晓静
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2006-10-20
Filing date: 2006-10-20
Publication date: 2008-08-20
Anticipated expiration: 2026-10-20
Also published as: CN1945627A

Abstract

This invention relates to a digital correction method of tongue color deflection, the feature is as the following steps: making white equation emendation on the image of Macbeth color adjusting calibration card from flash lamp and switching to the tonality, saturation, lightness space to get the corresponding adjustment parameters by cooperation with the standard value of color card. Then shoot the tongue picture by digital camera, and make white equation emendation switching to the tonality, saturation, lightness space, which is then adjusted by using the above parameter and returned into the red, green and blue spaces. Finally, use gamma value to adjust the pictures of red, green and blue spaces. The gamma value of curve emendation is combined by the least double multiplication with the light value of Macbeth card standard and the average value of picture light.

Description

Digital tongue picture color cast correction method

Technical Field

The present invention belongs to the field of digital correction method of tongue picture color cast.

Background

From the middle and late 80 s of the 20 th century, the combination of traditional Chinese medicine workers and digital image workers began to attempt to objectively study tongue diagnosis based on computer technology. Research has focused primarily on calibration of tongue photograph color, storage and output, and processing of tongue photographs with modern image processing analysis techniques. Among them, the color of the tongue is the most important information in tongue diagnosis. Although the tongue image shot in the forced flash mode of the digital camera can be shielded from natural light interference by the strong light of the flash lamp, the whole color of the tongue image is deviated when the strong light of the flash lamp irradiates the surface of the tongue, and the whole tongue image is biased to be red-yellow warm tone, so that the deviation needs to be corrected by a certain method.

The method utilizes a Macbeth24 color standard colorimetric card commonly used for correcting the color of a digital photo, the colorimetric card is processed by a special process, the surface of the colorimetric card has a reflection spectrum similar to the skin, and a series of calibration parameters are designed by comparing the color of the Macbeth24 colorimetric card in the photo with the color of a real colorimetric card for calibrating a shot tongue picture.

Disclosure of Invention

The invention aims to correct color cast caused by a strong flash lamp when a tongue picture is shot, and provides a set of general computer method for correcting tongue picture colors.

The invention is characterized in that:

the method comprises the following steps that (1) under the condition of forced flashing, N Macbeth color cards are shot by a digital camera, wherein N is 10-15 Macbeth color cards, and photos are input into a computer;

step (2), the computer corrects the white balance of the N Macbeth color chart photos:

if: the N photo Macbeth color chip photo is denoted as f₁(x，y)，f₂(x，y)…f_N(x, y) each image in the photograph is represented as: f. of_i(x，y)＝[R_i(x，y)，G_i(x，y)，B_i(x，y)]，i＝1…NR_i(x，y)，G_i(x，y)，B_i(x, y) represents the values of red, green, and blue at the sampling point (x, y), respectively; the corresponding picture of each photo after white balance correction is expressed as:

f_i′＝[R_i′(x，y)，G_i′(x，y)，B_i′(x，y)]，

wherein,

wherein, Max (R)_i(x，y))，Max(G_i(x, y)) and Max (B)_i(x, y)) are respectively shown in the image f_i(x, y), a maximum value of a red channel, a green channel, and a blue channel;

step (3) of expressing f in RGB space_i' (x, y) and Macbeth color chip color into HSV space representation, H, S, V denotes hue, saturation and value, respectively:

step (3.1), mixing_i′(x，y)＝[R_i′(x，y)，G_i′(x，y)，B_i′(x，y)]Conversion to HSV space:

V_i′(x，y)＝Max(R_i′(x，y)，G_i′(x，y)，B_i′(x，y))

said H_i′(x，y)，S_i′(x，y)，V_i' (x, y) respectively indicate the values of three channels of hue, saturation and lightness of the i-th image at the sampling point (x, y) after conversion, and R_i′(x，y)，G_i′(x，y)，B_i' (x, y) respectively shows the values of the red, green and blue channels of the ith image at the sampling point (x, y) after the white balance is adjusted, and the HSV channel of the image after the white balance is corrected is shown as:

f_i′(x，y)＝[H_i′(x，y)，S_i′(x，y)，V_i′(x，y)]

step (3.2), the standard value M of the Macbeth color card is set_jTurning into the space of the HSV (hue, saturation and value),

M_j＝[MR_j，MG_j，MB_j]，j＝1～24

{MH}_{j} = \cos^{- 1} [\frac{({MR}_{j} - {MG}_{j}) + ({MR}_{j} - {MB}_{j})}{2 \sqrt{{({MR}_{j} - {MG}_{j})}^{2} + ({MR}_{j} - {MG}_{j}) ({MR}_{j} - {MB}_{j})}}]

then

{MS}_{j} = \frac{Max ({MR}_{j}, {MG}_{j}, {MB}_{j}) - Min ({MR}_{j}, {MG}_{j}, {MB}_{j})}{Max ({MR}_{j}, {MG}_{j}, {MB}_{j})}

MV_j＝Max(MR_j，MG_j，MB_j)

Wherein, MR_j，MG_j，MB_jRespectively representing the values of the red, green and blue channels of the jth color block of the color card, MH_j，MS_j，MV_jRespectively representing the values of three channels of hue, saturation and lightness of the j color block after conversion;

obtaining: HSV space expression of the standard value of the color card: m_j＝[MH_j，MS_j，MV_j]，j＝1～24；

And (4) respectively solving each adjusting parameter through the values of the standard Macbeth color chart and the color chart photo according to the following steps:

step (4.1), calculating the average value of the H channel of the color tone of the Macbeth color chart photo after white balance adjustment, and comparing the average value with the average color tone of a standard color chart to obtain a color tone adjustment parameter H:

the average hue of the N Macbeth photos is

Hav_iTo adjust the average value of the hue H channel of the i-th image after white balance,

the Macbeth color chart has a standard hue average value of

The hue adjustment parameter is h-Hav-Hst;

and (4.2) calculating the average value of a brightness V channel of the picture of the Macbeth color chart after the white balance is adjusted, and comparing the average value with the average brightness of a standard color chart to obtain a brightness adjusting parameter V:

the lightness average value of N Macbeth photos is

Vpav_iTo adjust the average value of the lightness V channel of the i-th image after white balance,

the standard lightness average of the Macbeth color chart is

The brightness adjustment parameter is v ═ Vav-Vst;

and (4.3) calculating the average value of the saturation S channel of the Macbeth color chart photo after white balance adjustment, comparing the average value with the saturation of a color chart standard value, and obtaining a saturation adjustment parameter S through fitting:

in N Macbeth color chart photos, the average value of the saturation of the color block j is

Wherein, fS_ijI is 1 to N, j is 1 to 24,

the standard saturation of each color block in the color card is MS₁，MS₂…MS₂₄，

The saturation adjustment parameter s is:

and (5) calculating a curve correction value gamma:

in N Macbeth color chart photos, the average brightness value of a color block j is

Wherein, fV_ijThe lightness of the color block j in the ith picture, i is 1-N, j is 1-24,

the average brightness of each color block in the color card is Vbav₁，Vbav₂…Vbav₂₄，

The curve correction value gamma is:

<math> <mrow> <mi>gamma</mi> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>*</mo> <mi>ln</mi> <mrow> <mo>(</mo> <msub> <mi>MV</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </mfrac> <mo>;</mo> </mrow> </math>

taking a tongue picture by using a digital camera to obtain an image It (x, y) ([ IR (x, y), IG (x, y), IB (x, y) ], wherein IR (x, y), IG (x, y), and IB (x, y) respectively represent red, green, and blue values at the image sampling point (x, y);

and (7) correcting the white color in the tongue picture It (x, y) to be pure white, and obtaining a corrected tongue picture It' (x, y):

It′(x，y)＝[IR′(x，y)，IG′(x，y)，IB′(x，y)]

where Max (IR (x, y)), Max (IG (x, y)), and Max (IB (x, y)) respectively represent the maximum values of the red, green, and blue channels in the image It (x, y);

step (8), converting the tongue picture represented in RGB space to the expression of hue H, saturation S and lightness V space

IV′(x，y)＝Max(IR′(x，y)，IG′(x，y)，IB′(x，y))；

Step (9), judging whether the tongue picture is over exposed after correcting the white balance, and carrying out brightness balance on the underexposed picture:

step (9.1), performing brightness histogram distribution on the photo:

a histogram histv (t) ═ Num (IV '(x, y) ═ t), t ═ 0 to 255, and indicates the number of points whose value is t in the lightness graph IV' (x, y) represented by the t-th component in the histogram;

step (9.2), solving peak PeakV of lightness graph:

PeakV＝Max(HistV(t))，t＝0～255；

step (9.3), judging whether PeakV is less than threshold V

Wherein, the threshold V is a set value of 100, if PeakV is larger than or equal to the threshold V, the IV' (x, y) is not processed, otherwise, the next step is carried out;

step (9.4), histogram equalization is carried out on the image IV' (x, y) to obtain

<math> <mrow> <msup> <mi>IV</mi> <mo>′</mo> </msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>Σ</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mrow> <mi>IV</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> </mrow> </munderover> <mi>HistV</mi> <mrow> <mo>(</mo> <msup> <mi>IV</mi> <mo>′</mo> </msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math>

k represents a counter of lightness values;

and (10) correcting the image in the HSV space according to the adjusting parameters h, s and v set in the step (4):

IH_d(x，y)＝IH′(x，y)-h

IS_d(x，y)＝IS′(x，y)*s；

IV_d(x，y)＝IV′(x，y)-v

step (11), converting the image in the HSV space into the RGB space:

step (11.1), four temporary variables f, aa, bb, cc are set to aid in transformation:

wherein f is IH_d-floor(IH_d) Function floor (IH)_d) Means taking ratio IH_dSmall maximum integer, so f denotes IH_dThe fractional part of (a) is,

aa＝IV_d*(1-S_d)

bb＝IV_d*(1-(IS_d*f))

cc＝IV_d*(1-(IS_d*(1-f)))

step (11.2), according to IH_dRange determination of (IR)_d，IG_d.IB_dThe value of (c):

if it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mn>0</mn> <mo>~</mo> <mfrac> <mi>π</mi> <mn>6</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝IV_d，IG_d＝cc，IB_d＝aa

If it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>6</mn> </mfrac> <mo>~</mo> <mfrac> <mi>π</mi> <mn>3</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝bb，IG_d＝IV_d，IB_d＝aa

If it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>3</mn> </mfrac> <mo>~</mo> <mfrac> <mi>π</mi> <mn>2</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝aa，IG_d＝IV_d，IB_d＝cc

If it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>2</mn> </mfrac> <mo>~</mo> <mfrac> <mrow> <mn>2</mn> <mi>π</mi> </mrow> <mn>3</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝aa，IG_d＝bb，IB_d＝IV_d

If it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mrow> <mn>2</mn> <mi>π</mi> </mrow> <mn>3</mn> </mfrac> <mo>~</mo> <mfrac> <mrow> <mn>5</mn> <mi>π</mi> </mrow> <mn>6</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝cc，IG_d＝aa，IB_d＝IV_d

If it is

<math> <mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mrow> <mn>5</mn> <mi>π</mi> </mrow> <mn>6</mn> </mfrac> <mo>~</mo> <mn>2</mn> <mi>π</mi> <mo>)</mo> <mo>,</mo> </mrow> </math>

Then IR_d＝IV_d，IG_d＝aa，IB_d＝bb

And (12) adjusting the image by using the curve correction value gamma:

IR_d′(x，y)＝IR_d(x，y)^gamma

IG_d′(x，y)＝IG_d(x，y)^gamma

IB_d′(x，y)＝IB_d(x，y)^gamma

resulting adjusted image It_d(x，y)＝[IR_d′(x，y)，IG_d′(x，y)，IB_d′(x，y)]The adjusted picture of the tongue picture is obtained.

The use result proves that the adjusted image reflects the real color of the tongue more exactly than the unadjusted image through visual inspection of a doctor, and the adjusted image is accepted by the doctor.

Drawings

FIG. 1: an algorithm flow chart;

FIG. 2: a Macbeth color chart real object graph;

FIG. 3: a Macbeth color chart diagram;

FIG. 4: a Macbeth color chart truth table;

FIG. 5: histogram of the photograph: (a) exposing a lightness histogram of the photograph which is not enough, (b) making a lightness histogram after histogram equalization;

FIG. 6: taking a picture of the tongue;

FIG. 7: adjusting the tongue picture after white balance and brightness homogenization;

FIG. 8: and adjusting the tongue picture result.

Detailed Description

The hardware of the method consists of a digital camera and a computer. The software for adjusting the algorithm is implemented by Matlab. The input tongue picture is shot by doctor in Chinese medicine with a digital camera, and the patient stretches out the tongue when shooting, so that doctors can conveniently take the picture of the whole tongue.

The Macbeth color card consists of 24 color blocks, is distributed into four rows and six columns, and is arranged from left to right and from top to bottom, as shown in FIG. 2. Sequentially represented as M_j＝[MR_j，MG_j，MB_j]And j is 1-24, and the sequence is schematically shown in FIG. 3. Wherein, MR_j，MG_j，MB_jThe values of the red, green and blue color channels in the jth patch are shown, respectively, and the standard values of the 24 patches of the card are listed in the accompanying description, as shown in FIG. 4.

A Sony (Sony) Cybershot DSC-W1 digital camera is adopted to take pictures indoors, strong light interference is avoided during shooting, and a forced flash lamp mode is used at the same time, wherein the distance is about 30 centimeters. 10 Macbeth color cards were taken and the pictures were entered into the computer.

The white balance of ten images is adjusted, the brightest color in the images is adjusted to (255, 255, 255), and the other colors are linearly adjusted correspondingly. And then transformed into HSV space.

And comparing the image converted into the HSV space with the color of the standard color card, and solving parameters. In our experiments, four parameters were obtained, h-0.02, s-5, v-6 and gamma-1.1, respectively.

A photograph of the tongue is taken and stored in the computer. As shown in fig. 6. After adjusting the white balance of the picture, the picture is converted into HSV space. The histogram of the V channel is obtained, and it is found that the image belongs to the underexposure type, and the lightness of the image is equalized, and the obtained picture is as shown in fig. 7.

The image is adjusted accordingly according to the parameters h 0.02, s-5, v-6, and gamma 1.1 obtained by the system, and the final adjustment result is shown in fig. 8.

The method comprises the following specific implementation steps:

step (1) shooting N Macbeth color cards

The shooting conditions are as follows: the indoor photographing method has the advantages that strong light interference is avoided during photographing, and meanwhile, a forced flash lamp mode is used, and the distance is about 30 centimeters. And shooting N Macbeth color cards, wherein N is 10-15. The photograph is entered into a computer.

Step (2) correcting the white balance of the N Macbeth color chart photos

Representing N photographs stored in a computer as f₁(x，y)，f₂(x，y)…f_N(x, y), wherein, for each image, f₁(x，y)＝[R_i(x，y)，G_i(x，y)，B_i(x，y)]And i is 1 … N. (x, y) represents sample points of an image, R_i(x，y)，G_i(x，y)，B_i(x, y) represent the values of red, green, and blue at the sample point, respectively.

Since the values of three pure white colors of red, green, and blue in the digital photograph are 255, 255, and 255, respectively, the color is the brightest color in the image and is the maximum value of each channel. In order to correct the brightness color cast of the shot image, before correcting the color of the image, the white color in the picture needs to be corrected to be pure white, and the method comprises the following steps:

wherein, Max (R)_i)，Max(G_i) And Max (B)_i) Is shown in the image f_i(x, y) the maximum of the red, green and blue channels. The corresponding picture after adjustment is denoted as f_i′＝[R_i′(x，y)，G_i′(x，y)，B_i′(x，y)]. By adjusting the white balance, we obtain a series of adjusted images f₁′(x，y)，f₂′(x，y)…f_N′(x，y)。

Step (3) f to be expressed in RGB (red, green, blue) space_i' (x, y), i-1 … N and color card color standard value M_j＝[MR_j，MG_j，MB_j]J is 1 to 24 and is converted into HSV (hue, saturation, brightness) space representation

For an image f whose white balance has been corrected_i′(x，y)＝[R_i′(x，y)，G_i′(x，y)，B_i′(x，y)]It is transformed into the HSV space by the following formula.

V_i′(x，y)＝Max(R_i′(x，y)，G_i′(x，y)，B_i′(x，y))

Wherein R is_i′(x，y)，G_i′(x，y)，B_i' (x, y) denotes the values of the three red, green and blue channels of the ith image at the sampling point (x, y) after the white balance adjustment, respectively, H_i′(x，y)，S_i′(x，y)，V_i' (x, y) respectively indicate the values of three channels of hue, saturation and lightness of the i-th image at the sampling point (x, y) after conversion. The HSV channel from which the image after white balance correction is derived is denoted

f_i′(x，y)＝[H_i′(x，y)，S′(x，y)，V_i′(x，y)]

In addition, the standard value M of the color chart is to be matched_j＝[MR_j，MG_j，MB_j]And j is also transferred to the HSV space from 1 to 24.

{MH}_{j} = \cos^{- 1} [\frac{({MR}_{j} - {MG}_{j}) + ({MR}_{j} - {MB}_{j})}{2 \sqrt{{({MR}_{j} - {MG}_{j})}^{2} + ({MR}_{j} - {MG}_{j}) ({MR}_{j} - {MB}_{j})}}]

{MS}_{j} = \frac{Max ({MR}_{j}, {MG}_{j}, {MB}_{j}) - Min ({MR}_{j}, {MG}_{j}, {MR}_{j})}{Max ({MR}_{j}, {MG}_{j}, {MB}_{j})}

MV_j＝Max(MR_j，MG_j，MB_j)

Wherein, MR_j，MG_j，MB_jRespectively representing the red, green and blue of the jth color block of the color cardValues of three channels, MH_j，MS_j，MV_jAnd respectively representing the values of three channels of hue, saturation and brightness of the j th color block after conversion. Thereby obtaining the HSV space expression mode M of the standard value of the color card_j＝[MH_j，MS_j，MV_j]，j＝1～24。

And (4) solving each adjusting parameter through the values of the standard Macbeth color chart and the color chart photo:

step (4.1) of obtaining the average value of the H (tone) channel of the Macbeth color chart photo after white balance adjustment, and comparing the average value with the average tone of the color chart photo to obtain a tone adjustment parameter H

By Hav₁，Hav₂，…Hav_NThe average value of H channels of each adjusted image i is shown, and the average value of the tone of N images is

In addition, the average value of the standard hue value of the color chart is obtained

The difference h between the two is Hav-Hst which is a parameter for adjusting the chromaticity.

Step (4.2) calculating the average value of the V (brightness) channel of the Macbeth color chart photo after white balance adjustment, and comparing the average value with the average brightness of the color chart photo to obtain a brightness adjustment parameter V

With Vpav₁，Vpav₂，…Vpav_NThe average value of the V channels of the adjusted images is shown, and the brightness average value of the N images is

In addition, the average value of the standard values of the color chart is obtained

The difference v-Vst between the two is a parameter for adjusting the brightness.

Step (4.3) calculating the average value of the channels of the Macbeth color chart photo S (saturation) after white balance adjustment, comparing the average value with the saturation of the color chart photo S (saturation), and obtaining a saturation adjustment parameter S through fitting

Photograph f after adjustment of white balance_i' (x, y), 24 rectangular color blocks are taken out according to positions, and as shown in FIG. 3, the saturation in each color block is determined to be fS_i1，fS_i2.…fS_i24. After the saturation of 24 color blocks is respectively obtained for all the collected photos, the average value of each color block for the N pictures is obtained.

For the calculated saturation mean value Sav₁，Sav₂…Sav₂₄And color card standard saturation MS₁，MS₂…MS₂₄Fitting by using a least square method standard, wherein the concrete method comprises the following steps: selecting parameter s, MS_j＝s*Sav_jSo that

The value of (c) is minimal.

The calculation method of s is as follows:

step (5) calculating the average value of the V (lightness) channel of the color chart photo after the white balance is adjusted, comparing the average value with the lightness of the color chart photo, and obtaining a curve correction value gamma through fitting

gamma is derived from the response curve of a CRT (display/television), i.e. its brightness is non-linear with respect to the input voltage;

gamma correction corresponds to processing an image; since the change in gamma causes a change in lightness, the value of gamma is obtained by the lightness V channel:

photograph f after adjustment of white balance_i' (x, y) taking out 24 rectangular color blocks according to positions, and obtaining the value of fV in each color block_i1，fV_i2.…fV_i24. After the brightness of 24 color blocks is respectively obtained for all the collected pictures, the average value of each color block for the N pictures is obtained.

For the obtained brightness average value Vbav₁，Vbav₂…Vbav₂₄And color chart standard lightness MV₁，MV₂…MV₂₄Fitting by using a least square method standard, wherein the concrete method comprises the following steps: selecting parameters gamma, MV_j＝Vbav_j ^gammaSo that

<math> <mrow> <mi>σ</mi> <mo>=</mo> <msqrt> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <msup> <mrow> <mo>(</mo> <msup> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mi>gamma</mi> </msup> <mo>-</mo> <msub> <mi>MV</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </math>

The value of (c) is minimal.

The gamma is calculated by

<math> <mrow> <mi>gamma</mi> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>*</mo> <mi>ln</mi> <mrow> <mo>(</mo> <msub> <mi>MV</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </mfrac> <mo>.</mo> </mrow> </math>

And (6) taking a tongue picture by using a digital camera, and storing the tongue picture in a computer.

The expression method is that It (x, y) ([ IR (x, y), IG (x, y), IB (x, y) ], (x, y) represents a sampling point of an image, and IR (x, y), IG (x, y), and IB (x, y) represent values of red, green, and blue at the sampling point, respectively.

Step (7) correcting the white balance of the tongue picture

The white color in the picture of the tongue picture is corrected to be pure white by the following method:

It′(x，y)＝[IR′(x，y)，IG′(x，y)，IB′(x，y)]

where max (ir), max (ig), and max (ib) represent the maximum values of the red, green, and blue channels in the image It (x, y). By adjusting the white balance, we obtain the adjusted tongue picture as It' (x, y).

Step (8) will represent It in RGB space_i' (x, y) is transformed into HSV (hue, saturation, brightness) space representation

IV′(x，y)＝Max(IR′(x，y)，IG′(x，y)，IB′(x，y))

Step (9) carrying out lightness balance on the underexposed photos

And (3) after the white balance is corrected, analyzing distribution of a brightness histogram, marking the tongue picture if the position where the peak value appears is smaller than a predetermined threshold value, carrying out brightness balance, and then carrying out next adjustment.

The histogram of the lightness channel IV' (x, y) of the image is counted. The value of the lightness channel is 0-255. And (5) counting a histogram HistV (t) of the lightness channel, wherein t is 0-255.

The histogram is defined as histv (t) ═ Num (IV '(x, Y) ═ t), that is, the t-th component of the histogram represents the number of points having a value of t in the lightness map IV' (x, Y). Fig. 5(a) shows a histogram of one image, where Max (peak v (t)) and t are 0 to 255 in the lightness graph. The abscissa is a gray value, the value is from 0 to 255, and the ordinate is the number of pixel points in the image, the gray value of which is equal to the gray of the abscissa.

If PeakV < threshold V, histogram equalization is performed on image IV' (x, y). The equation for equalization is expressed as:

meaning that the gray level of a pixel (x, y) before equalization is IV '(x, y), then the brightness value of this point after equalization is the sum of the histogram from gray value 1 to gray value IV' (x, y). In the equation, k represents a brightness value, and fig. 5(b) is a histogram in which the histogram is changed by histogram equalization.

If PeakV ≧ threshold V, no treatment is performed on IV' (x, y).

Correcting the image in the HSV space according to the set parameters h, s and v;

IH_d(x，y)＝IH′(x，y)-h

IS_d(x，y)＝IS′(x，y)*s

IV_d(x，y)＝IV′(x，y)-v

step (11), converting the image in the HSV space into the RGB space and then using gamma coefficient

The procedure for transferring HSV to RGB space is:

four temporary variables f, aa, bb, cc are set to assist in the operation, where

f＝IH_d-floor(IH_d) Function floor (IH)_d) Means taking ratio IH_dSmall maximum integer, so f denotes IH_dFractional part of

aa＝IV_d*(1-IS_d)

bb＝IV_d*(1-(IS_d*f))

cc＝IV_d*(1-(IS_d*(1-f)))

According to IH_dCan determine the IR_d，IG_d，IB_dValue of (A)

If it is

Then IR_d＝IV_d，IG_d＝cc，IB_d＝aa

If it is

Then IR_d＝bb，IG_d＝IV_d，IB_d＝aa

If it is

Then IR_d＝aa，IG_d＝IV_d，IB_d＝cc

If it is

Then IR_d＝aa，IG_d＝bb，IB_d＝IV_d

If it is

Then IR_d＝cc，IG_d＝aa，IB_d＝IV_d

If it is

Then IR_d＝IV_d，IG_d＝aa，IB_d＝bb

After the image is transferred into the RGB space, the image in the RGB space is adjusted by the curve correction value gamma.

IR_d′(x，y)＝IR_d(x，y)^gamma

IG_d′(x，y)＝IG_d(x，y)^gamma

IB_d′(x，y)＝IB_d(x，y)^gamma

Claims

1. The digital tongue picture color cast correction method is characterized by sequentially comprising the following steps of:

if: the N photo Macbeth color chip photo is denoted as f₁(x，y)，f₂(x，y)…f_N(x，y)

Each of the picturesThe image is represented as: f. of_i(x，y)＝[R_i(x，y)，G_i(x，y)，B_i(x，y)]，i ＝1…N

R_i(x，y)，G_i(x，y)，B_i(x, y) represents the values of red, green, and blue at the sampling point (x, y), respectively;

the corresponding picture of each photo after white balance correction is expressed as:

f_i′＝[R_i′(x，y)，G_i′(x，y)，B_i′(x，y)]，

wherein,

V_i′(x，y)＝Max(R_i′(x，y)，G_i′(x，y)，B_i′(x，y))

f_i′(x，y)＝[H_i′(x，y)，S_i′(x，y)，V_i′(x，y)]

M_j＝[MR_j，MG_j，MB_j]，j＝1～24

{MH}_{j} = \cos^{- 1} [\frac{({MR}_{j} - {MG}_{j}) + ({MR}_{j} - {MB}_{j})}{2 \sqrt{{({MR}_{j} - {MG}_{j})}^{2} + ({MR}_{j} - {MG}_{j}) ({MR}_{j} - {MB}_{j})}}]

then

{MS}_{j} = \frac{Max ({MR}_{j}, {MG}_{j}, {MB}_{j}) - Min ({MR}_{j}, {MG}_{j}, {MB}_{j})}{Max ({MR}_{j}, {MG}_{j}, {MB}_{j})}

MV_j＝Max(MR_j，MG_j，MB_j)

Wherein, MR_j，MG_j，MB_jRespectively representing the values of the red, green and blue channels of the jth color block of the color card, MH_j，MS_j，MV_jRespectively representing the values of three channels of hue, saturation and lightness of the j color block after conversion; obtaining: HSV spatial expression for the colour chip standard value: m_j＝[MH_j，MS_j，MV_j]，j＝1～24；

the average hue of the N Macbeth photos is

Hav_iFor adjusting ith after white balance

The average value of the hue H channel of the image,

the Macbeth color chart has a standard hue average value of

The hue adjustment parameter is h-Hav-Hst;

the lightness average value of N Macbeth photos is

the standard lightness average of the Macbeth color chart is

The brightness adjustment parameter is v ═ Vav-Vst;

Wherein, fS_ijIs composed of

In the ith picture, the saturation of the color patch j, i is 1 to N, j is 1 to 24,

The saturation adjustment parameter s is:

and (5) calculating a curve correction value gamma:

average of color blocks in color cardBrightness of Vbav₁，Vbav₂…Vbav₂₄，

The curve correction value gamma is:

<math><mrow> <mi>gamma</mi> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>*</mo> <mi>ln</mi> <mrow> <mo>(</mo> <mi>M</mi> <msub> <mi>V</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>24</mn> </munderover> <mi>ln</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>Vbav</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </mfrac> <mo>;</mo> </mrow></math>

It′(x，y)＝[IR′(x，y)，IG′(x，y)，IB′(x，y)]

step (8) of converting the tongue picture represented in the RGB space to a picture represented in the hue H, saturation S, lightness V space

IV′(x，y)＝Max(IR′(x，y)，IG′(x，y)，IB′(x，y))；

step (9.1), performing brightness histogram distribution on the photo:

step (9.2), solving peak PeakV of lightness graph:

PeakV＝Max(HistV(t))，t＝0～255；

step (9.3), judging whether PeakV is less than threshold V

HistV(IV′(x.y))，

k represents a counter of lightness values;

IH_d(x，y)＝IH′(x，y)-h

IS_d(x，y)＝IS′(x，y)*s；

IV_d(x，y)＝IV′(x，y)-v

step (11), converting the image in the HSV space into the RGB space:

aa = {IV}_{d} * (1 - {IS}_{d})

bb = {IV}_{d} * (1 - ({IS}_{d} * f))

cc = {IV}_{d} * (1 - ({IS}_{d} * (1 - f)))

step (11.2), according to IH_dRange determination of (IR)_d，IG_d，IB_dThe value of (c):

if it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mn>0</mn> <mo>~</mo> <mfrac> <mi>π</mi> <mn>6</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝IV_d，IG_d＝cc，IB_d＝aa

If it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>6</mn> </mfrac> <mo>~</mo> <mfrac> <mi>π</mi> <mn>3</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝bb，IG_d＝IV_d，IB_d＝aa

If it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>3</mn> </mfrac> <mo>~</mo> <mfrac> <mi>π</mi> <mn>2</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝aa，IG_d＝IV_d，IB_d＝cc

If it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mi>π</mi> <mn>2</mn> </mfrac> <mo>~</mo> <mfrac> <mrow> <mn>2</mn> <mi>π</mi> </mrow> <mn>3</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝aa，IG_d＝bb，IB_d＝IV_d

If it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mrow> <mn>2</mn> <mi>π</mi> </mrow> <mn>3</mn> </mfrac> <mo>~</mo> <mfrac> <mrow> <mn>5</mn> <mi>π</mi> </mrow> <mn>6</mn> </mfrac> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝cc，IG_d＝aa，IB_d＝IV_d

If it is

<math><mrow> <msub> <mi>IH</mi> <mi>d</mi> </msub> <mo>&Element;</mo> <mo>[</mo> <mfrac> <mrow> <mn>5</mn> <mi>π</mi> </mrow> <mn>6</mn> </mfrac> <mo>~</mo> <mn>2</mn> <mi>π</mi> <mo>)</mo> <mo>,</mo> </mrow></math>

Then IR_d＝IV_d，IG_d＝aa，IB_d＝bb

And (12) adjusting the image by using the curve correction value gamma:

IR_d′(x，y)＝IR_d(x，y)^gamma

IG_d′(x，y)＝IG_d(x，y)^gamma

IB_d′(x，y)＝IB_d(x，y)^gamma