CN103279928B - A kind of image enchancing method based on atmospherical scattering model - Google Patents

Abstract

The present invention discloses and a kind of self-adaptation can retain the image enchancing method based on atmospherical scattering model of image dark portion and highlighted part information, comprise step: (1) is based on the histogram image equalization of the dark primary passage principle of atmospherical scattering model: dark primary channel characteristic and the image histogram distribution characteristic control image equilibration scope utilizing atmospherical scattering model, makes the image after equalization be beneficial to and keep dark portion and highlight regions minutia in image; (2) based on the dark primary Postprocessing technique of atmospherical scattering model.

Description

A kind of image enchancing method based on atmospherical scattering model

Technical field

The invention belongs to digital picture and the technical field of digital video image process, it is specifically related to a kind of image enchancing method based on atmospherical scattering model, is mainly used in the video image of safety-security area network shooting machine monitoring.

Background technology

Security protection video supervisory system usually can in the face of various light environment; multiple natural cause causes monitor video image to degenerate; video image contrast gradient and spatial resolution are reduced; graphic information is lost serious; therefore; usually in security protection video supervisory system, it is necessary to the image obtained is carried out image enhancement processing, to obtain more monitoring informations and better vision effect.

Atmospherical scattering model is by McCartney(McCartney) etc. people propose, be the important models in image restoration theory. The physical expressions of this model is: I (x, y)=J (x, y) t (x, y)+A (x, y) [1-t (x, y)]

In formula: J (x, y) is the intensity pattern under desirable meteorological conditions; I (x, y) is actual emanations intensity, is the degeneration image of J (x, y); The transmission plot that t (x.y) is air in environment, represents the transmissivity of light in atmospheric environment; J (x, y) t (x, y) is the yield of radiation of J (x, y) correspondence after air transmission decays; A (x, y) is atmosphere light yield of radiation; A (x, y) [1-t (x, y)] is corresponding atmosphere light scattering strength. As can be seen from atmospherical scattering model, I (x, y) it is actual emanations intensity, i.e. actual observation image, as long as obtaining air transmission plot t (x, and atmosphere light intensity pattern A (x y), y), so that it may with the intensity pattern J (x, y) obtained under desirable meteorological conditions, the i.e. restored image of I (x, y).

On this model basis, He Kaiming propose based on dark primary passage method ingenious solve the recovery of Misty Image, and achieve more effective achievement. The method is that the priori being based upon dark primary passage is assumed on basis, and namely following formula priori is assumed to set up:

$I_{\mathrm{dark}}\left(x\right)=\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(I^c\left(y\right)\right)\right)\→0$

In formula, I_darkX () is image dark primary channel image, I^cY () represents for input picture in c field of definition, �� (x) is for asking for region, and c is for calculating field of definition: r, g, b(image red, green, blue passage). Then atmospheric model can change to:

$\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(\frac{I^c\left(y\right)}{A^c}\right)\right)=t\left(x\right)\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(\frac{J^c\left(y\right)}{A^c}\right)\right)+(1-t\left(x\right))$

In formula, J^c(x, y) is the graphical representation of the intensity pattern under desirable meteorological conditions in c field of definition, and �� (x) is for asking for region, and c is for calculating field of definition: r, g, b(image red, green, blue passage), A^cFor atmospheric scattering intensity.

Then have:

$t\left(x\right)=1-\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(\frac{I^c\left(y\right)}{A^c}\right)\right)$

After obtaining environment Transmission light figure t (x), namely can calculate the intensity pattern J (x, y) under desirable meteorological conditions by atmospherical scattering model, have:

$J\left(x\right)=\frac{I\left(x\right)-A^c}{\max (t\left(x\right),t_0)}+A^c$

To in the experiment of the method, it has been found that when the method uses in monitoring environment, supercompression situation often occur in image dark portion and highlight regions, there is image dark portion and the problem of highlight regions part information loss.

Summary of the invention

The technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, it is provided that a kind of self-adaptation can retain the image enchancing method based on atmospherical scattering model of image dark portion and highlighted part information.

The technical solution of the present invention is: this kind, based on the image enchancing method of atmospherical scattering model, comprises the following steps:

(1) based on the histogram image equalization of dark primary passage principle of atmospherical scattering model: dark primary channel characteristic and the image histogram distribution characteristic control image equilibration scope utilizing atmospherical scattering model, makes the image after equalization be beneficial to and keep dark portion and highlight regions minutia in image;

(2) based on the dark primary Postprocessing technique of atmospherical scattering model.

The image equilibration method equalization scope of the present invention is controlled by original image dark primary channel characteristic, original image property of the histogram, image dark portion and high highlights area information is made to be adjusted to more reasonable region by image equilibration process, self-adaptation can retain image dark portion and highlighted part information, be suitable for using in protection and monitor field.

Accompanying drawing explanation

Fig. 1 shows the schema of the image enchancing method based on atmospherical scattering model according to the present invention.

Embodiment

As shown in Figure 1, this kind, based on the image enchancing method of atmospherical scattering model, comprises the following steps:

(1) based on the histogram image equalization of dark primary passage principle of atmospherical scattering model: dark primary channel characteristic and the image histogram distribution characteristic control image equilibration scope utilizing atmospherical scattering model, makes the image after equalization be beneficial to and keep dark portion and highlight regions minutia in image;

(2) based on the dark primary Postprocessing technique of atmospherical scattering model.

The image equilibration method equalization scope of the present invention is controlled by original image dark primary channel characteristic, original image property of the histogram, image dark portion and high highlights area information is made to be adjusted to more reasonable region by image equilibration process, self-adaptation can retain image dark portion and highlighted part information, be suitable for using in protection and monitor field.

Preferably, step (1) comprises step by step following:

(1.1) by the dark primary passage principle of atmospherical scattering model, computed image primary channel, method of calculation are as follows:

$I_{\mathrm{dark}}\left(x\right)=\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(I^c\left(y\right)\right)\right)$

In formula, I_darkX () is image dark primary channel image, I^cY () is the graphical representation of original image in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage;

By region detection dark primary channel image I_darkX (), demarcates min (I_dark(x)) region, demarcate max (I_dark(x)) region;

(1.2) brightness of image is carried out statistics with histogram and equilibrium treatment

Luminance component image Y is carried out statistics with histogram, and in statistical graph picture, gray scale is the appearance probability of the pixel of i:

$p_x\left(i\right)=\frac{n_i}{n},$ i��0,��,L-1

In formula, L is brightness value all in image, and n is pixel count all in image, n_iFor brightness value is the pixel count of i, p_xI statistic histogram that () is image, normalizes to [0.0,1.0]; Check statistic histogram p_xThe distribution of (i), Computation distribution maximum value y_max, minimum value y_minAnd distribution average y_mean, distribution meansquaredeviation��_y;

C is the accumulative normalization histogram of image, corresponding to p_xI the cumulative probability function of (), is defined as:

$c\left(i\right)=\underset{j=0}{\overset{i}{\mathrm{\Σ}}}{p}_x\left(j\right)$

Design of graphics is as luminance proportion function y_i=T(x_i), with the conversion method of the cumulative probability function of brightness it is:

y_i=T(x_i)=c(i)

Check and demarcated min (I_dark(x)) region, in statistical regions brightness histogram distributionAnd; Check and demarcated max (I_dark(x)) region in zone luminance valueAnd, by following formula, y ' is set_maxAnd y '_min:

$y_{\max }^{\′}=p_1\•y_{\max }+p_2{\•y}_{\max }^{\mathrm{dark}}+p_3\•y_{\max }^{\mathrm{light}}$

$y_{\min }^{\′}=q_1\•y_{\min }+q_2{\•y}_{\min }^{\mathrm{dark}}+q_3\•y_{\min }^{\mathrm{light}}$

Brightness of image equalization methods is:

y��_i=c(i)��(y��_max-y��_min)+y��_min

In formula, y '_maxFor image equilibration high-high brightness, y '_minFor the minimum brightness of image equilibration, y '_iFor output brightness.

Preferably, step (2) comprises step by step following:

(2.1) image dark primary Air conduct measurement

Image after image equilibration carries out image dark primary Air conduct measurement, and computed image primary channel method is as follows:

$I_{\mathrm{dark}}\left(x\right)=\underset{c\∈\{r,g,b\}}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(I^c\left(y\right)\right)\right)$

In formula, I_darkX () is image dark primary channel image, I^cY () represents for image equilibration image input picture in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage;

By region detection dark primary channel image I_darkX (), demarcates I_darkThe region of (x)��15 and statistics quantity X_dark; Demarcate max (I_dark(x)) region, statistical regions X_dark;

(2.2) atmosphere light is estimated

Utilize I_darkX () is to atmosphere light A^cEstimate: select in nxn �� (x) region, calculate and meet I_darkAll I of (x) >=15_darkX in (), (r, g, the b) of the original image pixels of the highlighted correspondence of front 0.1% is as atmosphere light A^cEstimation;

(2.3) air transmission plot is calculated

Image transmission graph representation atmosphere light irradiate under, the transmission relation of each part in image, according to following formula computed image transmission plot:

$\tilde{t}\left(x\right)=1-\mathrm{\ω}\underset{c}{\min }\left(\underset{y\∈\mathrm{\Ω}\left(x\right)}{\min }\left(\frac{I^c\left(y\right)}{A^c}\right)\right)$

��=0.95 in formula; I^cY () represents for input picture in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage, A^cFor regional atmospheric light is estimated;

Adopt method pair belowCarry out refinement, obtain t (x);

$(L+\mathrm{\λU})t\left(x\right)=\mathrm{\λ}\tilde{t}\left(x\right)$

L is the stingy figure matrix of Laplce, and U is unit battle array, ��=10^-4;

(2.4) image restoration

By air transmission plot t (x), carry out image restoration by following formula,

$J\left(x\right)=\left\{\begin{array}{cc}\frac{I\left(x\right)-A^c}{\max (t\left(x\right),t_0)}+A^c,& I_{\mathrm{drak}}\left(x\right)\&GreaterEqual;\mathrm{dark}\_\mathrm{th}\\ I\left(x\right),& I_{\mathrm{dark}}\left(x\right)<\mathrm{dark}\_\mathrm{th}\end{array}\right.$

In formula, t₀=0.1, t (x) is air transmission plot, A^cFor regional atmospheric light is estimated, I (x) is input figure

Picture, I_dark(x) for image pixel dark primary value, dark_th be dark primary threshold value.

Hereinafter provide a specific embodiment, comprise the following steps:

One, image dark primary passage pre-detection

By dark primary passage principle in atmospherical scattering model, computed image primary channel, method of calculation are as follows:

$I_{\mathrm{dark}}\left(x\right)=\underset{c\&Element;\{r,g,b\}}{\min }(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{\min }\left(I^c\left(y\right)\right))$

In formula, I_darkX () is image dark primary channel image, I^cY () represents for input picture in c field of definition, �� (x) is for asking for region, and c is for calculating field of definition: r, g, b(image red, green, blue passage).

By region detection dark primary channel image I_darkX (), demarcates min (I_dark(x)) region, demarcate max (I_dark(x)) region.

Two, brightness of image is carried out statistics with histogram and equilibrium treatment

Luminance component image Y is carried out statistics with histogram, and in statistical graph picture, gray scale is the appearance probability of the pixel of i:

$p_x\left(i\right)=\frac{n_i}{n},$ i��0,��,L-1

In formula, L is brightness value all in image, and n is pixel count all in image, n_iFor brightness value is the pixel count of i, p_xI statistic histogram that () is image, normalizes to (0.0,1.0).

Check statistic histogram p_xThe distribution of (i), Computation distribution maximum value y_max, minimum value y_minAnd distribution average y_mean, distribution meansquaredeviation��_y��

C is the accumulative normalization histogram of image, corresponding to p_xI the cumulative probability function of (), is defined as:

$c\left(i\right)=\underset{j=0}{\overset{i}{\mathrm{\&Sigma;}}}{p}_x\left(j\right)$

Design of graphics is as luminance proportion function y_i=T(x_i), with the conversion method of the cumulative probability function of brightness it is:

y_i=T(x_i)=c(i)

Check and demarcated min (I_dark(x)) region, in statistical regions brightness histogram distributionAndCheck and demarcated max (I_dark(x)) region in zone luminance valueAndBy following formula, y ' is set_maxAnd y '_min:

$y_{\max }^{\&prime;}=p_1\&bull;y_{\max }+p_2{\&bull;y}_{\max }^{\mathrm{dark}}+p_3\&bull;y_{\max }^{\mathrm{light}}$

$y_{\min }^{\&prime;}=q_1\&bull;y_{\min }+q_2{\&bull;y}_{\min }^{\mathrm{dark}}+q_3\&bull;y_{\min }^{\mathrm{light}}$

Brightness of image equalization methods is:

y��_i=c(i)��(y��_max-y��_min)+y��_min

Above in formula, y '_maxFor image equilibration high-high brightness, y '_minFor the minimum brightness of image equilibration, y '_iFor output brightness.

Three, image dark primary Air conduct measurement

Image after image equilibration carries out image dark primary Air conduct measurement, and computed image primary channel method is as follows:

$I_{\mathrm{dark}}\left(x\right)=\underset{c\&Element;\{r,g,b\}}{\min }\left(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{\min }\left(I^c\left(y\right)\right)\right)$

In formula, I_darkX () is image dark primary channel image, I^cY () represents for image equilibration image input picture in c field of definition, �� (x) is for asking for region, and c is for calculating field of definition: r, g, b Wei image red, green, blue passages.

By region detection dark primary channel image I_darkX (), demarcates I_darkThe region of (x)��15 and statistics quantity X_dark; Demarcate max (I_dark(x)) region, statistical regions X_dark��

Four, atmosphere light is estimated

Utilize I_darkX () is to atmosphere light A^cEstimate;

Method of estimation: select in nxn �� (x) region, calculates and meets I_darkAll I of (x) >=15_darkX in (), (r, g, the b) of the original image pixels of the highlighted correspondence of front 0.1% is as atmosphere light A^cEstimation;

Five, air transmission plot is calculated

Image transmission graph representation image scene atmosphere light irradiate under, the transmission relation of each part in image, according to following formula computed image transmission plot:

$\tilde{t}\left(x\right)=1-\mathrm{\&omega;}\underset{c}{\min }\left(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{\min }\left(\frac{I^c\left(y\right)}{A^c}\right)\right)$

��=0.95 in upper formula; I^cY () represents for input picture in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage, A^cFor regional atmospheric light is estimated.

Adopt method pair belowCarry out refinement, obtain t (x);

$(L+\mathrm{\&lambda;U})t\left(x\right)=\mathrm{\&lambda;}\tilde{t}\left(x\right)$

L is the stingy figure matrix of Laplce, and U is unit battle array, ��=10^-4��

Six, image restoration

By air transmission plot t (x), carry out image restoration by following formula,

$J\left(x\right)=\left\{\begin{array}{cc}\frac{I\left(x\right)-A^c}{\max (t\left(x\right),t_0)}+A^c,& I_{\mathrm{drak}}\left(x\right)\&GreaterEqual;\mathrm{dark}\_\mathrm{th}\\ I\left(x\right),& I_{\mathrm{dark}}\left(x\right)<\mathrm{dark}\_\mathrm{th}\end{array}\right.$

In formula, t₀=0.1, t (x) is air transmission plot, A^cFor regional atmospheric light is estimated, I (x) is input picture, I_dark(x) for image pixel dark primary value, dark_th be dark primary threshold value.

The above; it it is only the better embodiment of the present invention; the present invention not does any restriction in form, and every any simple modification, equivalent variations and modification above embodiment done according to the technical spirit of the present invention, all still belongs to the protection domain of technical solution of the present invention.

Claims (1)

1. the image enchancing method based on atmospherical scattering model, it is characterised in that, comprise the following steps:

(1) based on the histogram image equalization of dark primary passage principle of atmospherical scattering model: dark primary channel characteristic and the image histogram distribution characteristic control image equilibration scope utilizing atmospherical scattering model, makes the image after equalization be beneficial to and keep dark portion and highlight regions minutia in image;

(2) based on the dark primary Postprocessing technique of atmospherical scattering model;

Step (1) comprises step by step following:

(1.1) by the dark primary passage principle of atmospherical scattering model, computed image primary channel, method of calculation are as follows:

$I_{dark}\left(x\right)=\underset{c\&Element;\{r,g,b\}}{min}\left(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{min}\right(I^c\left(y\right)\left)\right)$

In formula, I_darkX () is image dark primary channel image, I^cY () is the graphical representation of original image in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage;

By region detection dark primary channel image I_darkX (), demarcates min (I_dark(x)) region, demarcate max (I_dark(x)) region;

(1.2) brightness of image is carried out statistics with histogram and equilibrium treatment

Luminance component image Y is carried out statistics with histogram, and in statistical graph picture, gray scale is the appearance probability of the pixel of i:

$p_x\left(i\right)=\frac{n_i}{n},i\&Element;0,1,...,L-1$

In formula, L is the brightness value of image, and n is pixel count all in image, n_iFor brightness value is the pixel count of i, p_xI statistic histogram that () is image, normalizes to [0.0,1.0];

Check statistic histogram p_xThe distribution of (i), Computation distribution maximum value y_max, minimum value y_minAnd distribution average y_mean, distribution meansquaredeviation��_y;

C (i) is the accumulative normalization histogram of image, corresponding to p_xI the cumulative probability function of (), is defined as:

$c\left(i\right)=\underset{j=0}{\overset{i}{\&Sigma;}}{p}_x\left(j\right)$

Design of graphics is as luminance proportion function y_i=T (x_i), with the conversion method of the cumulative probability function of brightness it is:

y_i=T (x_i)=c (i)

Check and demarcated min (I_dark(x)) region, in statistical regions brightness histogram distributionAndCheck and demarcated max (I_dark(x)) region in zone luminance valueAndBy following formula, y ' is set_maxAnd y '_min:

$y_{max}^{\&prime;}=p_1\&CenterDot;y_{max}+p_2\&CenterDot;y_{\max }^{dark}+p_3\&CenterDot;y_{max}^{light}$

$y_{\min }^{\&prime;}=q_1\&CenterDot;y_{\min }+q_2\&CenterDot;y_{min}^{dark}+q_3\&CenterDot;y_{\min }^{light}$

p₁��p₂��p₃For y '_maxThe normalization method weights calculated, p₁+p₂+p₃=1, q₁��q₂��q₃For y '_minThe normalization method weights calculated, q₁+q₂+q₃=1

Brightness of image equalization methods is:

y��_i=c (i) �� (y '_max-y��_min)+y��_min

In formula, y '_maxFor image equilibration high-high brightness, y '_minFor the minimum brightness of image equilibration, y '_iFor output brightness;

Step (2) comprises step by step following:

(2.1) image dark primary Air conduct measurement

Image after image equilibration carries out image dark primary Air conduct measurement, and computed image primary channel method is as follows:

$I_{dark}\left(x\right)=\underset{c\&Element;\{r,g,b\}}{min}\left(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{min}\right(I^c\left(y\right)\left)\right)$

In formula, I_darkX () is image dark primary channel image, I^cY () represents for image equilibration image input picture in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage;

By region detection dark primary channel image I_darkX (), demarcates I_darkThe region of (x)��15 and statistics quantity X_dark; Demarcate max (I_dark(x)) region, statistical regions X_dark;

(2.2) atmosphere light is estimated

Utilize I_darkX () is to atmosphere light A^cEstimate: select in nxn �� (x) region, calculate and meet I_darkAll I of (x) >=15_darkX in (), (r, g, the b) of the original image pixels of the highlighted correspondence of front 0.1% is as atmosphere light A^cEstimation;

(2.3) air transmission plot is calculated

Image transmission graph representation atmosphere light irradiate under, the transmission relation of each part in image, according to following formula computed image transmission plot:

$\tilde{t}\left(x\right)=1-\mathrm{\&omega;}\underset{c}{min}\left(\underset{y\&Element;\mathrm{\&Omega;}\left(x\right)}{min}\right(\frac{I^c\left(y\right)}{A^c}\left)\right)$

��=0.95 in formula; I^cY () represents for input picture in c field of definition, �� (x) for asking for region, c be calculate red passage that field of definition: r is image, green passage that g is image, b be blue passage, A^cFor regional atmospheric light is estimated;

Adopt method pair belowCarry out refinement, obtain t (x);

$(L+\mathrm{\&lambda;}U)t\left(x\right)=\mathrm{\&lambda;}\tilde{t}\left(x\right)$

L is the stingy figure matrix of Laplce, and U is unit battle array, ��=10^-4;

(2.4) image restoration

By air transmission plot t (x), carry out image restoration by following formula,

$J\left(x\right)=\left\{\begin{array}{cc}\frac{I\left(x\right)-A^c}{\max \left(t\right(x),t_0)}+A^c,& I_{drak}\left(x\right)\&GreaterEqual;dark\_th\\ I\left(x\right),& I_{dark}\left(x\right)<dark\_th\end{array}\right.$

In formula, t₀=0.1, t (x) is air transmission plot, A^cFor regional atmospheric light is estimated, I (x) is input picture, I_dark(x) for image pixel dark primary value, dark_th be dark primary threshold value.