CN100504440C - Tone Adjustment Method Based on Image Overlapping Area in Remote Sensing Image Mosaic - Google Patents

Abstract

This invention relates to a tone adjustment method based on image overlap region for mosaic of remote sensing image, which comprises the following steps: 1. acquiring the pixel brightness values from the basic image and target image; 2. computing the current bright value and the difference distribution compared with the values from step 1, smoothing the histogram in result, and obtaining the opposite diff as difference value on the peak; 3. stating the pixel brightness value less than one, smoothing the result, and obtaining Pos as the opposite brightness value on the peak; 4. regulating the image color by the Pos value. This invention reduces algorithm complexity, and improves practicability.

Description

Hue adjusting method during remote sensing image is inlayed based on image overlap area

Technical field:

The invention belongs to remote sensing image embedding technique field, the hue adjusting method during the remote sensing image that relates to a kind of difference based on adjacent image overlap area brightness value is inlayed.

Background technology:

It is one of important content of remote sensing technology application that remote sensing image is inlayed, and particularly in a big way satellite remote sensing and regional airborne remote sensing, often needs several images to inlay; But, can make atural object of the same name on adjacent image, present different tones owing to obtain the difference of the condition of raw video.If these situations are not taked counter-measure, will make the overall image after inlaying tangible brightness disproportionation occur, the color unbalanced phenomena.This phenomenon has caused spliced image both to be unfavorable for eye-observation, has also influenced the effect to the analysis and the identification of atural object image and specialized information, therefore, in the image mosaic process, needs a kind of technology can adjust the integral color of inlaying image.

Present general hue adjusting method can reduce: histogram method, mean variance method, image information entropy method; The common trait of these methods is that calculated amount is bigger, has often influenced service efficiency, and this shortcoming is especially obvious when several large tracts of land images are carried out batch processing.

Summary of the invention:

The object of the present invention is to provide simple, the hue adjusting method in realizing being easy to inlaying of a kind of algorithm based on the remote sensing image of the difference of adjacent image overlap area brightness value.

Technical scheme provided by the invention is: the hue adjusting method based on image overlap area during remote sensing image is inlayed may further comprise the steps:

One, obtains reference images and handle image overlap brightness value partly from reference images; Treatment step is as follows:

The terrestrial coordinate of I, benchmark image and pending image determines with pending image overlapping areas is arranged in the reference images;

II, take out in the reference images by the brightness value g of the pixel in the resulting zone of step I (x, y);

Two, the difference distribution situation of the brightness value that pending brightness value of image and step 1 obtain on the statistics overlapping region is made smoothing processing to the statistics histogram, obtains the difference diff of histogram peak-peak place correspondence; Treatment step is as follows:

The terrestrial coordinate of I, benchmark image and pending image is determined pending image and reference images overlapping areas;

II, take out in the pending image by the brightness value f of the pixel in step 2, the resulting zone of I (x, y);

III, statistics luminance difference g (x, y)-f (x, histogram distribution situation S[i y)];

IV, level and smooth luminance difference histogram S[i]:

S[min]＝(S[min]+S[min+1])/2

S[n]＝(S[n-1]+S[n]+S[n+1])/3

S[max]＝(S[max]+S[max-1])/2

Wherein, min is the minimum brightness difference, and max is a maximum luminance difference, and n gets the value between min and the max,

Be min＜n＜max;

V, record S[i] luminance difference diffi when maximum;

Three, the brightness value difference equals the distribution situation of the pixel brightness value of diff on the pending image overlap area of statistics, and the statistics histogram is made smoothing processing, obtains the brightness value Pos of histogram peak-peak place correspondence; Treatment step is as follows:

Work as in the overlapping region on I, the pending image of statistics luminance difference g (x, y)-f (x, y)=f during diff (x, brightness histogram distribution situation F[j y)];

II, with the used method of step 2, IV to histogram F[j] carry out smoothing processing, and the level and smooth back of record F[j] brightness value Pos when maximum;

Four, with pending image T (x, y) do adjustment on the integral color:

T (x, y)＜during Pos: T (x, y)=T (x, y) * (Pos+diff)/Pos;

T (x, y)〉during Pos: T (x, y)=[T (x, y)-Pos] * [255-(Pos+diff)]/(255-Pos)+(Pos+diff);

In the formula:

X is the abscissa value of pixel in the image, value is 1,2,3 ..., m; The m number of pixels on the horizontal direction of image for this reason wherein;

Y is the ordinate value of pixel in the image, value is 1,2,3 ..., n; The n number of pixels on the vertical direction of image for this reason wherein;

T (x, y) for this reason in the image position at (x, y) brightness value of the pixel on.

When processed image is chromatic image, before above-mentioned steps one, carry out following processing earlier:

A. according to the RGB color space to the conversion method of the HIS double cone model color space, the rgb value of pending each pixel of image is converted to the HSI value;

B. the I component value of the HIS that obtains after the HIS conversion through RGB that proposes each pixel is as brightness value.

When processed image is several images, image and reference images after handling are spliced to form new reference images at every turn, the overlapping region pixel value adopts the pixel value of former benchmark image overlap area.

The present invention can adjust the integral color difference of image.The present invention utilizes the luminance difference statistics with histogram feature of image overlap area to adjust the image integral color, has reduced the complexity of algorithm, and algorithm is simple, realization is easy, has improved practicality.

Description of drawings:

Accompanying drawing is a schematic flow sheet of the present invention.

Embodiment:

Referring to accompanying drawing, the present invention includes following steps:

One, RGB-HIS conversion

According to colorimetry knowledge as can be known, human eye is comparatively responsive to brightness sensation, simultaneously, because the difference on the remote sensing image tone mainly is to cause because of uneven illumination, difference on the form and aspect is also not obvious, all images is handled to the HIS chrominance space by the RGB color space conversion for this reason; Its formula is:

$\mathrm{\θ}=\arccos \left\{\frac{\frac{1}{2}[(R-G)+(R-B)]}{{[{(R-G)}^2+(R-B)(G-B)]}^{1/2}}\right\}$

$S=1-\frac{3}{(R+G+B)}\left[\min (R,G,B)\right]$

$I=\frac{1}{3}(R+B+G)$

In the formula:

R is the red color component value of each pixel of chromatic image;

G is the green component values of each pixel of chromatic image;

B is the blue component value of each pixel of chromatic image.

Two, obtain the image overlap area brightness value

The coordinate information of I, the pending image of taking-up, the taking-up reference images neutralizes, and it has the zone of lap;

II, the brightness value of the pixel in reference images and the pending image overlap area is taken out;

Three, integral color adjustment

I, set up one with the two-dimensional matrix B of the identical size of pending image (x, y), with in the matrix corresponding to the part mark of pending image overlap area;

II, at mark image overlap area in, current brightness value of image be expressed as f (x, y), the brightness value that obtains in the step 2 can be expressed as g (x, y);

Luminance difference g in III, the statistics overlapping region (x, y)-f (x, histogram distribution situation S[i y)], S[i] can obtain by the pixel in the overlapping region is traveled through:

S[i]＝S[i]+1 (i＝g(x，y)-f(x，y))

In the formula: x, y are the ranks number of overlapping area pixel point

IV, level and smooth luminance difference histogram S[i]:

S[min]＝(S[min]+S[min+1])/2

S[n]＝(S[n-1]+S[n]+S[n+1])/3

S[max]＝(S[max]+S[max-1])/2

Wherein, min is the minimum brightness difference, and max is a maximum luminance difference, and n gets the value between min and the max, i.e. min＜n＜max;

V, record S[i] luminance difference diff when maximum;

Work as in the overlapping region on VI, the pending image of statistics luminance difference g (x, y)-f (x, y)=f during diff (x, brightness histogram distribution situation F[j y)];

VII, with used same procedure of IV step to histogram F[j] carry out smoothing processing, and the level and smooth back of record F[j] brightness value Pos when maximum;

VIII, with pending image T (x, y) do adjustment on the integral color:

T (x, y)＜during Pos: T (x, y)=T (x, y) * (Pos+diff)/Pos;

T (x, y)〉during Pos: T (x, y)=[T (x, y)-Pos] * [255-(Pos+diff)]/(255-Pos)+(Pos+diff);

X is the abscissa value of pixel in the image, value is 1,2,3 ..., m; The m number of pixels on the horizontal direction of image for this reason wherein;

Y is the ordinate value of pixel in the image, value is 1,2,3 ..., n; The n number of pixels on the vertical direction of image for this reason wherein;

T (x, y) for this reason in the image position at (x, y) brightness value of the pixel on;

Four, the splicing of image after will handling and reference images is as the reference images of handling next time:

I, can set up an external polygon of this two width of cloth image according to the coordinate of image of handling and reference images;

II, the polygon that is obtained by step I can be set up a corresponding blank image of polygonal region size therewith;

III, with in this blank image with the benchmark image overlapping areas in pixel compose pixel value with respective pixel in the benchmark image;

IV, will be in this blank image compose pixel value with respective pixel on the image after handling by the pixel of III step operation assignment;

V, then can obtain a new reference images through above step, then get back to for second step to circulate and all finish processing until all images.

Claims (3)

1, the hue adjusting method based on image overlap area during a kind of remote sensing image is inlayed is characterized in that: may further comprise the steps:

One, obtain the brightness value of reference images and pending image overlap part from reference images:

The terrestrial coordinate of I, benchmark image and pending image determines with pending image overlapping areas is arranged in the reference images;

II, take out in the reference images by the brightness value g of the pixel in the resulting zone of step I (x, y); G (x, y) in: x is the abscissa value of pixel; Y is the ordinate value of pixel;

Two, the difference distribution situation of the brightness value that pending brightness value of image and step 1 obtain on the statistics overlapping region is made smoothing processing to the statistics histogram, obtains the difference diff of histogram peak-peak place correspondence:

The terrestrial coordinate of I, benchmark image and pending image is determined pending image and reference images overlapping areas;

II, take out in the pending image by the brightness value f of the pixel in step 2, the resulting zone of I (x, y); F (x, y) in: x is the abscissa value of pixel; Y is the ordinate value of pixel;

III, statistics luminance difference g (x, y)-f (x, histogram distribution situation S[i y)];

IV, level and smooth luminance difference histogram S[i]:

S[min]＝(S[min]+S[min+1])/2

S[n]＝(S[n-1]+S[n]+S[n+1])/3

S[max]＝(S[max]+S[max-1])/2

Wherein, min is the minimum brightness difference, and max is a maximum luminance difference, and n gets the value between min and the max, i.e. min＜n＜max;

V, record S[i] luminance difference diff when maximum;

Three, the brightness value difference equals the distribution situation of the pixel brightness value of diff on the pending image overlap area of statistics, and the statistics histogram is made smoothing processing, obtains the brightness value Pos of histogram peak-peak place correspondence:

Work as in the overlapping region on I, the pending image of statistics luminance difference g (x, y)-f (x, y)=f during diff (x, brightness histogram distribution situation F[j y)];

II, with the used method of step 2, IV to histogram F[j] carry out smoothing processing, and the level and smooth back of record F[j] brightness value Pos when maximum;

Four, with pending image T (x, y) do adjustment on the integral color:

T (x, y)＜during Pos: T (x, y)=T (x, y) * (Pos+diff)/Pos;

T (x, y)〉during Pos: T (x, y)=[T (x, y)-Pos] * [255-(Pos+diff)]/(255-Pos)+(Pos+diff);

In the formula:

X is the abscissa value of pixel in the image, value is 1,2,3 ..., m; The m number of pixels on the horizontal direction of image for this reason wherein;

Y is the ordinate value of pixel in the image, value is 1,2,3 ..., n; The n number of pixels on the vertical direction of image for this reason wherein;

T (x, y) for this reason in the image position at (x, y) brightness value of the pixel on.

2, method according to claim 1 is characterized in that: when processed image is chromatic image, carried out following processing earlier before step 1:

A. according to the RGB color space to the conversion method of the HIS double cone model color space, the rgb value of pending each pixel of image is converted to the HSI value;

B. the I component value of the HIS that obtains after the HIS conversion through RGB that proposes each pixel is as brightness value.

3, method according to claim 1 and 2, it is characterized in that: when processed image is several images, image and reference images after handling are spliced to form new reference images at every turn, and the overlapping region pixel value adopts the pixel value of former benchmark image overlap area.

Images