CN105118029A - Medical image enhancement method based on human eye visual characteristic - Google Patents

Abstract

The present invention provides a medical image enhancement method based on a human eye visual characteristic, belonging to the field of digital image processing. The method comprises the steps of (S1) inputting a medical image needed to be enhanced, converting the medical image to an HSV space from an RGB space, and obtaining a brightness component v (x, y), a hue component h (x, y) and a saturation component s (x, y), (S2) defining an adaptive correction function according to a brightness histogram cumulative distribution function, obtaining adjustment parameters k and c, (S3) according to the brightness component v (x, y) and the adjustment parameters k and c, using a modified TAN function to calculate and obtain an adjusted brightness component v' (x, y), (S4) according to the image mean value I<-> after brightness adjustment, judging whether the parameter c needs to be adjusted finely, if so, carrying out fine adjustment to obtain a new parameter c', replacing c with c', then returning to the step S3, and if not, going to step (S5).

Description

A kind of medical image enhancement method based on human-eye visual characteristic

Technical field

The invention belongs to digital image processing field, be specifically related to a kind of medical image enhancement method based on human-eye visual characteristic.

Background technology

Medical image enhancement process needs the mainly brightness regulation of solution and colour detail to strengthen problem, and the image enchancing method of routine exists a lot of weak point, is difficult to the object realizing medical image enhancement.Retinex algorithm is to realize color constancy for original intention, human-eye visual characteristic is utilized to carry out brightness adjustment and colour detail enhancing to image, obtained the color of this pixel by the relative relationship between light and dark calculated between each pixel, there is good high dynamic range compression effect and for medical image, there is good performance equally.Retinex algorithm proposed till now since 1963, domestic and international researcher successively to propose based on Retinex, the random walk Retinex of Variation Model, pyramid iteration Retinex, retina by center, territory/variously to improve one's methods around Retinex and other, to have become in real image reproduction algorithm an important branch.

Although Retinex algorithm is a kind of outstanding algorithm for image enhancement, also there is a lot of defect in it simultaneously, and such as cannot meet the brightness change of different light in the luminance compression of log-domain, adaptive faculty is not high.Multiple dimensioned superposition causes calculation of complex, and speed is slow.The color component of RGB tri-passages is processed respectively, easily causes color quantization noise, produce the phenomenon such as halation and albefaction, on image enhaucament generation impact in various degree.

Summary of the invention

The object of the invention is to solve the difficult problem existed in above-mentioned prior art, a kind of medical image enhancement method based on human-eye visual characteristic is provided, the problem of effective solution image light and shade inequality, and strengthen image local details, the color representation of former figure can be kept simultaneously.

The present invention is achieved by the following technical solutions:

Based on a medical image enhancement method for human-eye visual characteristic, comprising:

S1, input needs the medical image strengthened, and it is transformed into HSV space from rgb space, obtains luminance component v (x, y), chrominance component h (x, y) and saturation degree component s (x, y);

S2, according to image brightness histogram cumulative distribution function definition adaptive correction function, obtains regulating parameter k and c;

S3, according to luminance component v (x, y) and regulating parameter k and c, the luminance component v ' (x, y) after utilizing revised TAN function to calculate adjustment;

S4, according to carrying out the image average after brightness adjustment judge whether to need to finely tune parameter c, if so, then carry out finely tuning obtaining new parameter c ', and replace c with c ', then return S3, if not, then enter S5;

S5, strengthens the luminance component v " (x, y) after obtaining local detail enhancing to local detail;

S6, by v, " chrominance component that (x, y) and S1 obtain is converted into RGB and obtains output image together with saturation degree component.

In described S1, it is transformed into HSV space from rgb space to be achieved in that

If (r, g, b) is the red, green and blue coordinate of a color respectively, they are value real numbers between 0 to 1; If max is the maximum in r, g and b, if min is the reckling in r, g and b, obtain:

$s=\left\{\begin{array}{cc}0,& \begin{array}{cc}if& max=0\end{array}\\ \frac{max-min}{max},& otherwise\end{array}\right.$

V=max, h, s, v represent chrominance component h (x, y), saturation degree component s (x, y) and luminance component v (x, y) respectively.

Described S2 is achieved in that

According to image brightness histogram cumulative distribution function definition adaptive correction function:

Wherein C _arepresent that gray level is the image cumulative distribution of a, C _brepresent that gray level is the image cumulative distribution of b, a, b represent the ratio shared by dark areas and bright area respectively here; C _aand C _ball be less than 1, meet k < 0.5

T ₁, T ₂be respectively bright, dark statistical threshold.

Get a=50, b=200, get T ₁, T ₂be respectively 0.2,0.8.

Revised TAN function in described S3 is as follows:

$v^{\&prime;}(x,y)=\frac{v_{max}}{tan\&lsqb;(2c-1)k\mathrm{\&pi;}\&rsqb;+tan\left(k\mathrm{\&pi;}\right)}\&times;\&lsqb;tan\left(\frac{2\mathrm{\&pi;}kc}{v_{\max }}v\right(x,y)-k\mathrm{\&pi;})+tan\left(k\mathrm{\&pi;}\right)\&rsqb;$

Wherein 0 < k < 0.5,0≤c≤m _k; v _maxbe 255, m _krelevant to k.

Basis in described S4 carries out the image average after brightness adjustment judge whether to need to finely tune parameter c, if so, then carry out finely tuning obtaining new parameter c ' and be achieved in that

Image average be by the brightness of each pixel in image is added up, then obtain divided by the number of pixel;

$c^{\&prime;}=\left\{\begin{array}{cc}c-0.1& \stackrel{\&OverBar;}{I}<T_3\\ c+0.1& \stackrel{\&OverBar;}{I}>T_4\end{array}\right.$

Wherein, T ₃with T ₄be respectively and judge that integral image crosses dark and excessively bright threshold value.

T ₃with T ₄be respectively 100 and 180.

Described S5 is achieved in that

Operator below and v ' (x, y) convolutional calculation are obtained v " (x, y):

$\left[\begin{array}{ccc}0& -1& 0\\ -1& 5& -1\\ 0& -1& 0\end{array}\right].$

Described S6 is achieved in that

If (h, s, v) is the tone of a color, saturation degree and lightness dimension respectively, they are at the real number of value between 0 to 1;

h _i＝hmod60

$f=\frac{h}{60}-h_i$

p＝v×(1-s)

q＝v×(1-f×s)

t＝v×(1-(1-f)×s)

$(r,g,b)=\left\{\begin{array}{cc}(v,t,p),& h_i=0\\ (q,v,p),& h_i=1\\ (p,v,t),& h_i=2\\ (p,q,v),& h_i=3\\ (t,p,v),& h_i=4\\ (v,p,q),& h_i=5\end{array}\right..$

Compared with prior art, the invention has the beneficial effects as follows: the inventive method introduces the TAN function nonlinear adaptive curve of correction and the two antagonism Side-inhibition Model of ON/OFF of improvement, form a kind of enhancing algorithm of applicable medical image, efficiently solve the problem of image light and shade inequality, and enhance image local details, the color representation of former figure can be kept simultaneously, compared with similar algorithm for image enhancement, this method calculating is easy, universality good, visual effect is outstanding, meets the specific demand of medical image enhancement.

Accompanying drawing explanation

The two antagonism Side-inhibition Model of the ON/OFF that Fig. 1 improves

The step block diagram of Fig. 2 the inventive method

Former figure in Fig. 3 embodiment of the present invention

The figure of Fig. 4 after the inventive method process.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail:

The present invention, for the purpose of image enhaucament, first proposes a kind of non-linear global map model, introduces the brightness regulation mechanism that TAN function is also improved imitation pupil, according to the integral brightness level of the statistical property self-adaptative adjustment image of image.Afterwards according to the lateral inhibition compete mechanism of pathways for vision, propose a kind of two antagonism lateral inhibition response models of improvement, the image after brightness adjustment is carried out to the enhancing of regional area details, improve picture contrast.

The inventive method as shown in Figure 2, comprising:

Because RGB color space has higher color correlativity, therefore first image (namely needing the medical image strengthened) is transformed into the less HSV space of color correlation from rgb space, obtain luminance component v (x, y), chrominance component h (x, and saturation degree component s (x, y), and only process luminance component y).

The method being transformed into the less HSV space of color correlation from rgb space is specific as follows:

If (r, g, b) is the red, green and blue coordinate of a color respectively, they are at the real number of value between 0 to 1.If max is the maximum in r, g and b.If min is the reckling in r, g and b, can obtain:

$s=\left\{\begin{array}{cc}0,& \begin{array}{cc}if& max=0\end{array}\\ \frac{max-min}{max},& otherwise\end{array}\right.$

v＝max。

In order to the integral brightness level of energy self-adaptative adjustment image, introduce the TAN function with flexible mapping ability here, and can export correction by regulating parameter k and c to this function, formula is:

$v^{\&prime;}(x,y)=\frac{v_{max}}{tan\&lsqb;(2c-1)k\mathrm{\&pi;}\&rsqb;+tan\left(k\mathrm{\&pi;}\right)}\&times;\&lsqb;tan\left(\frac{2\mathrm{\&pi;}kc}{v_{\max }}v\right(x,y)-k\mathrm{\&pi;})+tan\left(k\mathrm{\&pi;}\right)\&rsqb;$

When using of the present invention, luminance component v (x, y) is substituted into above formula, and utilize k and c below to obtain v ' (x, y), v ' (x, y) represents the luminance component after calculating adjustment.

Wherein 0 < k < 0.5,0≤c≤m _k.V _maxgenerally be set to 255, m _krelevant to k (in above formula, as long as ensure as v (x, y) value is when (0,1), obtains v ' (x, y) value is also (0,1) between, so need to limit the span of k and c, after the span of the value of in k and c is determined, the span of another value is also determined thereupon), ensure that the output of TAN function can not be overflowed.Revised TAN function is applicable to the medical image of different light and shadow characteristics.

In order to make this function have better universality, then according to image brightness histogram cumulative distribution function definition adaptive correction function:

Wherein C _arepresent that gray level is the image cumulative distribution of a, C _brepresent that gray level is the image cumulative distribution of b, a=50, b=200 (a, b two be worth do not fix, be empirical value, without concrete scope), represent dark areas and the ratio shared by bright area respectively here.T ₁, T ₂be respectively light and shade statistical threshold, be respectively 0.2 and 0.8 (being empirical value, without concrete scope) here.C _aand C _ball be less than 1, (according to above formula, 0.2 and 0.35 is all less than 0.5, works as C to meet k < 0.5 _aand C _bwhen being all less than 1, be less than 0.5).

The image average after brightness adjustment is carried out according to above formula (brightness of each for image pixel added up, then divided by the number of pixel) judge effect, again processes if desired to small parameter perturbations, specific as follows:

$c^{\&prime;}=\left\{\begin{array}{cc}c-0.1& \stackrel{\&OverBar;}{I}<T_3\\ c+0.1& \stackrel{\&OverBar;}{I}>T_4\end{array}\right.$

T ₃with T ₄be respectively and judge that integral image crosses dark and excessively bright threshold value, be respectively 100 and 180 (empirical value, without concrete scopes) here.

After have adjusted integral image luminance level, need to strengthen local detail further, introduce a kind of two antagonism Side-inhibition Model of ON/OFF of improvement here, as shown in Figure 1, can obtain stable state output according to this model is

$r_p=r_{pon}-r_{poff}=(k_{pon}^{+}+k_{poff}^{-})e_p-\underset{j=1,j\&NotEqual;p}{\overset{n}{\&Sigma;}}(k_{pjon}^{-}+k_{pjoff}^{+})(r_j-r_{pj}^0)$

Consider the time resolution characteristics of the two antagonism of ON/OFF, the Transient Equations of above formula is

$\begin{array}{c}{r}_p\left(t\right)=r_{pon}\left(t\right)-r_{poff}\left(t\right)=(k_{pon}^{+}+k_{poff}^{-})\&times;(\frac{1}{\mathrm{\&tau;}}{\&Integral;}_0^t\exp \&lsqb;\frac{t}{{\mathrm{\&tau;}}_0}\&rsqb;e_p(t\left)dt\right)\\ -\underset{j=1,j\&NotEqual;p}{\overset{n}{\&Sigma;}}(k_{pjon}^{-}+k_{pjoff}^{+})\&times;\{\frac{1}{\mathrm{\&tau;}}{\&Integral;}_0^t\exp \&lsqb;\frac{t-t^{\&prime;}}{{\mathrm{\&tau;}}_s}\&rsqb;r_j(t)dt-r_{pj}^0)\}\end{array}$

Order $k_{pon}^{+}=k_{poff}^{-},k_{pjon}^{-}=k_{pjoff}^{+},r_{pj}^0=0,$ Output is reduced to

$r_p=2k_{pon}^{+}{e}_p-2\underset{j=1,j\&NotEqual;p}{\overset{n}{\&Sigma;}}{k}_{pjon}^{-}{e}_j$

Reasonable setting with value ( get 2.5, get 0.5, empirical value), obtain best enhancing effect, when specifically using,

When get 2.5, when getting 0.5, above formula becomes

$r_p=5e_p-\underset{j=1,j\&NotEqual;p}{\overset{n}{\&Sigma;}}{e}_j$

Visually make operator

0	-1	0
			-1	5	-1
0	-1	0

This operator and v ' (x, y) convolutional calculation are obtained v " (x, y).

Finally also needing v " (x, y) and original chrominance component are converted into RGB and obtain output image together with saturation degree component, specific as follows:

If (h, s, v) is the tone of a color, saturation degree and lightness dimension respectively, they are at the real number of value between 0 to 1.

h _i＝hmod60

$f=\frac{h}{60}-h_i$

p＝v×(1-s)

q＝v×(1-f×s)

t＝v×(1-(1-f)×s)

$(r,g,b)=\left\{\begin{array}{cc}(v,t,p),& h_i=0\\ (q,v,p),& h_i=1\\ (p,v,t),& h_i=2\\ (p,q,v),& h_i=3\\ (t,p,v),& h_i=4\\ (v,p,q),& h_i=5\end{array}\right.$

As shown in Figure 3 and Figure 4, wherein Fig. 3 is former figure to one embodiment of the present of invention, and brightness of image is uneven, and center is to the left a hot spot; Image detail is not good, is embodied in cell and lacks unity and coherence, and edge contour is not outstanding.Fig. 4 is the result figure after using this method, and brightness disproportionation problem is resolved, and cell is well arranged, and edge contour is clear, has good visual effect.

Technique scheme is one embodiment of the present invention, for those skilled in the art, on the basis that the invention discloses application process and principle, be easy to make various types of improvement or distortion, and the method be not limited only to described by the above-mentioned embodiment of the present invention, therefore previously described mode is just preferred, and does not have restrictive meaning.

Claims (9)

1. based on a medical image enhancement method for human-eye visual characteristic, it is characterized in that: described method comprises:

S1, input needs the medical image strengthened, and it is transformed into HSV space from rgb space, obtains luminance component v (x, y), chrominance component h (x, y) and saturation degree component s (x, y);

S2, according to image brightness histogram cumulative distribution function definition adaptive correction function, obtains regulating parameter k and c

S3, according to luminance component v (x, y) and regulating parameter k and c, the luminance component v ' (x, y) after utilizing revised TAN function to calculate adjustment;

S4, according to carrying out the image average after brightness adjustment judge whether to need to finely tune parameter c, if so, then carry out finely tuning obtaining new parameter c ', and replace c with c ', then return S3, if not, then enter S5;

S5, strengthens the luminance component v " (x, y) after obtaining local detail enhancing to local detail;

S6, by v, " chrominance component that (x, y) and S1 obtain is converted into RGB and obtains output image together with saturation degree component.

2. the medical image enhancement method based on human-eye visual characteristic according to claim 1, is characterized in that: in described S1, it is transformed into HSV space from rgb space and is achieved in that

If (r, g, b) is the red, green and blue coordinate of a color respectively, they are value real numbers between 0 to 1; If max is the maximum in r, g and b, if min is the reckling in r, g and b, obtain:

$s=\left\{\begin{array}{cc}0,& \begin{array}{cc}if& max=0\end{array}\\ \frac{max-min}{max},& otherwise\end{array}\right.$

V=max, h, s, v represent chrominance component h (x, y), saturation degree component s (x, y) and luminance component v (x, y) respectively.

3. the medical image enhancement method based on human-eye visual characteristic according to claim 2, is characterized in that: described S2 is achieved in that

According to image brightness histogram cumulative distribution function definition adaptive correction function:

Wherein C _arepresent that gray level is the image cumulative distribution of a, C _brepresent that gray level is the image cumulative distribution of b, a, b represent the ratio shared by dark areas and bright area respectively here; C _aand C _ball be less than 1, meet k < 0.5T ₁, T ₂be respectively bright, dark statistical threshold.

4. the medical image enhancement method based on human-eye visual characteristic according to claim 3, is characterized in that: get a=50, b=200, gets T ₁, T ₂be respectively 0.2,0.8.

5. the medical image enhancement method based on human-eye visual characteristic according to claim 4, is characterized in that: the revised TAN function in described S3 is as follows:

$v^{\&prime;}(x,y)=\frac{v_{max}}{tan\&lsqb;(2c-1)k\mathrm{\&pi;}\&rsqb;+tan\left(k\mathrm{\&pi;}\right)}\&times;\&lsqb;tan\left(\frac{2\mathrm{\&pi;}kc}{v_{\max }}v\right(x,y)-k\mathrm{\&pi;})+tan\left(k\mathrm{\&pi;}\right)\&rsqb;$

Wherein 0 < k < 0.5,0≤c≤m _k; v _maxbe 255, m _krelevant to k.

6. the medical image enhancement method based on human-eye visual characteristic according to claim 5, is characterized in that: the basis in described S4 carries out the image average after brightness adjustment judge whether to need to finely tune parameter c, if so, then carry out finely tuning obtaining new parameter c ' and be achieved in that

Image average be by the brightness of each pixel in image is added up, then obtain divided by the number of pixel;

$c^{\&prime;}=\left\{\begin{array}{cc}c-0.1& \stackrel{\&OverBar;}{I}<T_3\\ c+0.1& \stackrel{\&OverBar;}{I}>T_4\end{array}\right.$

Wherein, T ₃with T ₄be respectively and judge that integral image crosses dark and excessively bright threshold value.

7. the medical image enhancement method based on human-eye visual characteristic according to claim 6, is characterized in that: T ₃with T ₄be respectively 100 and 180.

8. the medical image enhancement method based on human-eye visual characteristic according to claim 7, is characterized in that: described S5 is achieved in that

Operator below and v ' (x, y) convolutional calculation are obtained v " (x, y):

$\left[\begin{array}{ccc}0& -1& 0\\ -1& 5& -1\\ 0& -1& 0\end{array}\right].$

9. the medical image enhancement method based on human-eye visual characteristic according to claim 8, is characterized in that: described S6 is achieved in that

If (h, s, v) is the tone of a color, saturation degree and lightness dimension respectively, they are at the real number of value between 0 to 1;

h _i＝hmod60

$f=\frac{h}{60}-h_i$

p＝v×(1-s)

q＝v×(1-f×s)

t＝v×(1-(1-f)×s)

$(r,g,b)=\left\{\begin{array}{cc}(v,t,p),& h_i=0\\ (q,v,p),& h_i=1\\ (p,v,t),& h_i=2\\ (p,q,v),& h_i=3\\ (t,p,v),& h_i=4\\ (v,p,q),& h_i=5\end{array}\right..$