CN104754177A - Chromatic aberration correction and bottom color filtering method of CIS large-breadth scanner - Google Patents

Abstract

The invention discloses a chromatic aberration correction and bottom color filtering method of an CIS (Contact Image Sensor) large-breadth scanner and belongs to the digital image processing field. The CIS large-breadth scanner is limited by fabrication processes and component aging; RGB images acquired by various CIS tubes present inconsistent chromatic aberrations; meanwhile bottom color drawing manuscripts and drawing lines are blended in a common application so that scanned images bring about more distortion after being again printed. In comparison with a traditional method, the chromatic aberration correction and bottom color filtering method is convenient to operate, reduced in software and hardware consumption and good in effect.

Description

A kind of chromatic aberration correction of CIS large format scanner and background color filter method

Technical field

The invention belongs to digital image processing field, be specifically related to a kind of large format scanner image chromatic aberration correction based on contact-type image sensor (Contact Image Sensor, CIS) and image background color filter method.

Background technology

Modern optoelectronic information technical development is rapid, and the physical size of object to be scanned is increasing, and the requirement of people to scanning accuracy is also more and more higher.The scanner of traditional structure cannot meet the demand of the aspect such as large format (A0 and more than), high-resolution (1200dpi) pixel, scanning transmission speed.Therefore, the research and development of high accuracy, wide picture scanner relevant hardware devices and software engineering have realistic price.In CIS large format scanner system, owing to being subject to illumination, humidity, the impact of physical characteristic, there is inevitable color difference (abbreviation aberration) in the image that every root CIS scans.Therefore the chromatic aberration correction for CIS large format scanner just becomes the important technical obtaining high quality graphic.

The chromatic aberration correction of traditional large format scanner mainly adopts and the method such as contiguous statistics theoretical based on complex frequency spectrum chrominance principle, and computational process is complicated, realizes difficulty larger.In order to reduce computation complexity and technology realizes difficulty, The present invention gives and a kind ofly have that computational complexity is little concurrently and the chromatic aberration calibrating method of the feature such as performance is good.Meanwhile, conventional belt background color drawing draft is eliminated background colour and is mostly adopted the method for binaryzation that scan image is converted into black and white image, have lost the details of original image drawing lines.For this situation, in order to keep drawing details, the present invention also creates a kind of background color filtering technique.

Summary of the invention

The present invention comprises chromatic aberration correction and background color filters two parts.

1. chromatic aberration correction

CIS large format scanner is limited to the problem such as manufacture craft and component aging, and the RGB image that each CIS pipe collects presents the inconsistent phenomenon of aberration.In order to solve this technical problem, being the calibration of aberration consistency, the present invention proposes following solution:

Prepare a common white correction paper (all uniform white) and a common black correction paper (all uniform black).Scanning corrects paper, the original image information that memory scanning is arrived in vain.According to image information, calculate the mean value of R, G, B tri-passages pixel value separately in every root CIS respectively as shown in formula (1).

${\stackrel{\‾}{I}}_k=({\stackrel{\‾}{r}}_k,{\stackrel{\‾}{g}}_k,{\stackrel{\‾}{b}}_k),k=\mathrm{1,2}...N---\left(1\right)$

Wherein represent white and correct paper each CIS pipe pixel mean value, k is CIS pipe number, and N is CIS radical.

Scan black correction paper, the original image information that memory scanning is arrived.According to image information, calculate the mean value of R, G, B tri-passages pixel value separately in every root CIS respectively, as shown in formula (2).

${\stackrel{\‾}{I}}_k^{*}=({\stackrel{\‾}{r}}_k^{*},{\stackrel{\‾}{g}}_k^{*},{\stackrel{\‾}{b}}_k^{*}),k=\mathrm{1,2}...N---\left(2\right)$

Wherein represent black correction paper each CIS pipe pixel mean value.

The scanning sequency of black, white correction paper can exchange.For the RGB image of 8 bit depth, the theoretical colour of white is (255,255,255), and the theoretical colour of black is (0,0,0).The mapping of scan values to theoretical value is done respectively, as shown in formula (3) to the pixel value of R, G, B tri-passages of each root CIS.

$\left\{\begin{array}{c}{a}_k^R\left[\begin{array}{c}{\stackrel{\‾}{r}}_k^{*}\\ {\stackrel{\‾}{r}}_k\end{array}\right]+b_k^R=\left[\begin{array}{c}0\\ 255\end{array}\right]\\ a_k^G\left[\begin{array}{c}{\stackrel{\‾}{g}}_k^{*}\\ {\stackrel{\‾}{g}}_k\end{array}\right]+b_k^G=\left[\begin{array}{c}0\\ 255\end{array}\right]\\ a_k^B\left[\begin{array}{c}{\stackrel{\‾}{b}}_k^{*}\\ {\stackrel{\‾}{b}}_k\end{array}\right]+b_k^B=\left[\begin{array}{c}0\\ 255\end{array}\right]\end{array}\right.---\left(3\right)$

Wherein for the parameter of correspondence mappings function, its concrete value is calculated by formula (4) and obtains.

$\left\{\begin{array}{c}{a}_k^R=\frac{255}{{\stackrel{\‾}{r}}_k-{\stackrel{\‾}{r}}_k^{*}},b_k^R=\frac{255\·{\stackrel{\‾}{r}}_k^{*}}{{\stackrel{\‾}{r}}_k^{*}-{\stackrel{\‾}{r}}_k}\\ a_k^G=\frac{255}{{\stackrel{\‾}{g}}_k-{\stackrel{\‾}{g}}_k^{*}},b_k^G=\frac{255\·{\stackrel{\‾}{g}}_k^{*}}{{\stackrel{\‾}{g}}_k^{*}-{\stackrel{\‾}{g}}_k}\\ a_k^B=\frac{255}{{\stackrel{\‾}{b}}_k-{\stackrel{\‾}{g}}_k^{*}},b_k^B=\frac{255\·{\stackrel{\‾}{b}}_k^{*}}{{\stackrel{\‾}{b}}_k^{*}-{\stackrel{\‾}{b}}_k}\end{array}\right.---\left(4\right)$

The mapping function of the value of mapping function parameter just can set up each root CIS R, G, B tri-passages pixel value separately after trying to achieve, as shown in formula (5).

$\left\{\begin{array}{c}{r}_k^{*}(i,j)=a_k^R\·r_k(i,j)+b_k^R\\ g_k^{*}(i,j)=a_k^G\·g_k(i,j)+b_k^G\\ b_k^{*}(i,j)=a_k^B\·b_k(i,j)+b_k^B\end{array}\right.---\left(5\right)$

Wherein, r _k(i, j), g _k(i, j), b _k(i, j) is respectively the pixel value of original image pixels point R, G, B tri-passages of corresponding CIS pipe, for the new pixel point value after function process, i, j are pixel position.So far, the calibration of chromatic aberration correction is completed.

When carrying out image aberration Concordance, target image is scanned, each pixel value that R, G, B tri-passages of the image scanned by each CIS pipe are corresponding substitutes into formula (5), obtain the pixel value after chromatic aberration correction, cover original image, form new image, the image namely after chromatic aberration correction.

The equipment of CIS broad width scanning instrument is installed in initialization, only needs the configuration carrying out a chromatic aberration correction parameter.Then regularly upgrade, to adapt to the chromatic aberration that CIS pipe causes due to senile cause.

Compared with traditional colour difference correcting method, the present invention is simple to operate, and software and hardware consumption is few, respond well.

(2) background color filters

The fusion of background color drawing draft and drawing lines is had to bring more distortion by when causing the image after scanning again to print in common application.In order to solve this technical problem, the present invention also creates a kind of background color filtering technique, for there being background color drawing draft to carry out background color filtration.Thought be by by uniform background color projection to white space (255,255,255), black project to black space (0,0,0) nearly, to complete the object that background color filters.Key step is as follows:

The background color drawing draft that has filtered needing background color scans, and obtains the raw information of its scan image.Pixel value corresponding for its R, G, B tri-passages is designated as r respectively _k(i, j), g _k(i, j), b _k(i, j), then low-pass filtering treatment is carried out to the copy of original image.Specifically, first to original image f (x, y) carry out Fourier transform and change F (i into, j), object is selection filter transform function H (i, j), by decay F (i, j) radio-frequency component produces the Fourier transform G (i, j) of output image.As shown in formula (6).

G(i,j)＝H(i,j)·F(i,j) (6)

Adopt Butterworth filter device, transfer function is as shown in formula (7).

$H(i,j)=\frac{1}{1+{\left(\frac{D(i,j)}{D_0}\right)}^{2n}}---\left(7\right)$

$D(i,j)=\sqrt{{(i-\frac{\mathrm{row}}{2})}^2+{(j-\frac{\mathrm{col}}{2})}^2}$

Wherein, D ₀for cut-off frequency is apart from the distance of initial point, n is progression, and D (i, j) is for point (i, j) is to the distance of Fourier transform center (initial point), and row, col are respectively picturedeep and columns.Finally by Fourier inversion, G (i, j) is converted to output image g (i, j), i.e. background image.

Calculate the average of the pixel value obtaining background image R, G, B tri-passages corresponding to every root CIS pipe, be designated as

In order to drawing draft and drawing lines are mapped to black space, prepare the black correction paper of a silent frame size, scanning obtains its original image information.The average calculating image R, G, B tri-passage pixel values corresponding to every root CIS pipe is designated as if this parameter pre-configured, also can skip this step in system.

Will substitute into formula (4), calculate corresponding mapping function parameter, the background color utilizing formula (5) to set up R, G, B tri-passages filters mapping function.Respectively by each pixel value r of R, G, B tri-passages of original image _k(i, j), g _k(i, j), b _k(i, j) substitutes into formula (5) and calculates acquisition form new image.Above-mentioned namely completing is filtered the background color of original image.

The background color filtering technique of the invention effectively can keep the details of original image drawing lines, reduces the distortion because duplicate printing brings, and eliminates background well simultaneously, respond well.

Accompanying drawing explanation

Fig. 1 is CIS large format scanner scanning schematic diagram;

Fig. 2 is embodiment correction parameter calculation flow chart;

Fig. 3 is embodiment image aberration correcting process figure;

Fig. 4 is chromatic aberration correction control flow chart;

Fig. 5 is embodiment background color filtering process figure;

Fig. 6 is background color filtering frame composition;

Fig. 7 is that background color filters exemplary plot;

Embodiment

Below in conjunction with drawings and Examples, the invention will be further described.

The present invention is based on the hardware configuration basis of existing contact-type image sensor (CIS) scanner, propose a kind of pattern colour difference correcting method and background color filtering technique.Background color filtering technique can be applied separately, also can utilize chromatic aberration correction initialization result.Specific embodiment is as described below:

Embodiment

Fig. 1 is CIS large format scanner scanning schematic diagram.

101: scanning area.Be responsible for scanning drawing, wherein comprise 5 CIS pipe, the scanning brought in order to the gap between eliminating because of CIS pipe is imperfect, cross arrangement between each CIS pipe, and has intersection.Data intersection processes by other image mosaic and integration technology.

102: roller bearing.Be in CIS vertical lower, be responsible for transmission drawing.

103: by the target drawing scanned, scanning direction as shown in the figure.

Fig. 2 is correction parameter calculation flow chart, specifically comprises:

Step 201: allocating default parameter.The white mean value correcting paper and black correction paper acquiescence pixel value of input.

Step 202: select to correct paper.Confirm that carrying out white correction paper parameter upgrades or the renewal of black correction paper parameter.

Step 203: scanning corrects paper.Put into correction paper to scan, the original image information that memory scanning is arrived.

Step 204: the average calculating each passage pixel value of every root CIS pipe R, G, B respectively, white is black is cover default parameters.

Step 205: whether cycle criterion carries out new correction paper parameter upgrades.Complete herein once in vain correct paper and once black correction paper parameter renewal.

Step 206: the pixel value average obtained according to step 204, utilizes formula (4):

$\left\{\begin{array}{c}{a}_k^R=\frac{255}{{\stackrel{\‾}{r}}_k-{\stackrel{\‾}{r}}_k^{*}},b_k^R=\frac{255\·{\stackrel{\‾}{r}}_k^{*}}{{\stackrel{\‾}{r}}_k^{*}-{\stackrel{\‾}{r}}_k}\\ a_k^G=\frac{255}{{\stackrel{\‾}{g}}_k-{\stackrel{\‾}{g}}_k^{*}},b_k^G=\frac{255\·{\stackrel{\‾}{g}}_k^{*}}{{\stackrel{\‾}{g}}_k^{*}-{\stackrel{\‾}{g}}_k}\\ a_k^B=\frac{255}{{\stackrel{\‾}{b}}_k-{\stackrel{\‾}{g}}_k^{*}},b_k^B=\frac{255\·{\stackrel{\‾}{b}}_k^{*}}{{\stackrel{\‾}{b}}_k^{*}-{\stackrel{\‾}{b}}_k}\end{array}\right.---\left(4\right)$

Obtain each passage of every root CIS pipe R, G, B and divide other mapping function parameter set up the mapping function of original value to corrected value.

$\left\{\begin{array}{c}{r}_k^{*}(i,j)=a_k^R\·r_k(i,j)+b_k^R\\ g_k^{*}(i,j)=a_k^G\·g_k(i,j)+b_k^G\\ b_k^{*}(i,j)=a_k^B\·b_k(i,j)+b_k^B\end{array}\right.---\left(5\right)$

The above-mentioned description to Fig. 2 presents the computational process of correction parameter.Fig. 3 gives the flow chart carrying out image chromatic aberration correction.

Step 301: read in mapping function parameter.This parameter is obtained by Fig. 2.

Step 302: scanning target image obtains the original image information of RGB color mode.

Step 303: by each pixel value r of original image _k(i, j), g _k(i, j), b _k(i, j) substitutes into corresponding mapping function, obtains the new pixel value after overcorrect

Step 304: pixel value pixel value new for step 303 gained being covered original image, namely completes the chromatic aberration correction of image.

Above-mentioned Fig. 3 completes the key step of chromatic aberration correction.Fig. 4 gives its control flow chart.

Step 401: judge whether to carry out correction parameter renewal.When equipment is installed initial, can allocating default parameter.

Step 402: correction parameter calculates.Correct paper by scanning, obtain the error of scan values and theoretical value, set up mapping relations, utilize formulae discovery to go out mapping function parameter.The idiographic flow of this step is provided by Fig. 2.

Step 403: scanning target image.Use CIS large format scanner scanning target image, memory image pattern is RGB color mode.

Step 404: judge whether to carry out chromatic aberration correction.Due to the difference of every root CIS pipe physical characteristic, between each CIS pipe scanning gained image, there is aberration.

Step 405: correcting image.The mapping function parameter obtained by Fig. 2, sets up mapping function, and original image pixels value is substituted into correspondence mappings function one by one, and the new image of generation, is the image after chromatic aberration correction.The idiographic flow of this step is provided by Fig. 3.

Step 406: the image after original image or chromatic aberration correction is exported or is used as other operations.

The above-mentioned description to Fig. 4 completes the description of chromatic aberration correction major control stream.Fig. 5 gives the main flow that background color filters.

Step 501: scanning target image.Scanning needs the target image carrying out background color filtration, and memory image pattern is RGB color mode.

Step 502: Fourier transform is carried out to the original image f (x, y) scanned, is converted to F (x, y).Butterworth filter is selected to carry out filtering process,

$H(i,j)=\frac{1}{1+{\left(\frac{D(i,j)}{D_0}\right)}^{2n}}---\left(7\right)$

$D(i,j)=\sqrt{{(i-\frac{\mathrm{row}}{2})}^2+{(j-\frac{\mathrm{col}}{2})}^2}$

In the present embodiment, D ₀=170, n=2.Obtain G (i, j) after treatment, Fourier inversion is carried out to it and obtains background image g (i, j).

Step 503: in units of CIS pipe, to calculate respectively by step 502 obtain the average of each passage pixel value of R, G, B of background image.

Step 504: judge whether the renewal carrying out parameter.If so, then the renewal of black correction paper parameter is carried out.Black correction paper parameter and chromatic aberration correction black correction paper parameter can share herein.

Step 505: by scanning black correction paper, complete the renewal to parameter.Concrete steps are provided by Fig. 2.

Step 506: utilize black correction parameter and step 503 gained background image parameter, sets up background image color to the mapping of white space and the closely black mapping to black space.By calculating the mapping function parameter obtaining R, G, B tri-passages, set up corresponding mapping function.

Step 507: original image is substituted into step 506 gained correspondence mappings function by pixel, namely the image of the new value composition of acquisition has been the image after background color filters.

Fig. 6 presents the integrated stand composition that background color filters.

Fig. 7 is that engineering drawing background color filters schematic diagram.A () is original engineering drawing, after background color filtration treatment, change (c) into, and can find out that details is not almost lost from (b) (d), background color filter effect is remarkable.

Claims (3)

1. a calibrating method for CIS large format scanner, comprising:

Step one: prepare a common white correction paper (all uniform white) and a common black correction paper (all uniform black), scanning corrects paper in vain, the original image information that memory scanning is arrived, according to image information, calculate the mean value of R, G, B tri-passages pixel value separately in every root CIS respectively as shown in formula (1),

${\stackrel{\‾}{I}}_k=({\stackrel{\‾}{r}}_k,{\stackrel{\‾}{g}}_k,{\stackrel{\‾}{b}}_k),k=\mathrm{1,2}...N---\left(1\right)$

Wherein represent white and correct paper each CIS pipe pixel mean value, k is CIS pipe number, and N is CIS radical,

Scan black correction paper, the original image information that memory scanning is arrived, according to image information, calculate the mean value of R, G, B tri-passages pixel value separately in every root CIS respectively, as shown in formula (2),

${\stackrel{\‾}{I}}_k^{*}=({\stackrel{\‾}{r}}_k^{*},{\stackrel{\‾}{g}}_k^{*},{\stackrel{\‾}{b}}_k^{*}),k=\mathrm{1,2}...N---\left(2\right)$

Wherein represent black correction paper each CIS pipe pixel mean value, the scanning sequency of black, white correction paper can exchange;

Step 2: for the RGB image of 8 bit depth, the theoretical colour of white is (255,255,255), the theoretical colour of black is (0,0,0), the mapping of scan values to theoretical value is done respectively, as shown in formula (3) to the pixel value of R, G, B tri-passages of each root CIS

$\left\{\begin{array}{c}{a}_k^R\left[\begin{array}{c}{\stackrel{\‾}{r}}_k^{*}\\ \stackrel{\‾}{r_k}\end{array}\right]+b_k^R=\left[\begin{array}{c}0\\ 255\end{array}\right]\\ a_k^G\left[\begin{array}{c}{\stackrel{\‾}{g}}_k^{*}\\ {\stackrel{\‾}{g}}_k\end{array}\right]+b_k^G=\left[\begin{array}{c}0\\ 255\end{array}\right]\\ a_k^B\left[\begin{array}{c}{\stackrel{\‾}{b}}_k^{*}\\ \stackrel{\‾}{b_k}\end{array}\right]+b_k^B=\left[\begin{array}{c}0\\ 255\end{array}\right]\end{array}\right.---\left(3\right)$

Step 3: the parameter of correspondence mappings function in computing formula (3) its concrete value is calculated by formula (4) and obtains,

$\left\{\begin{array}{c}{a}_k^R=\frac{255}{\stackrel{\‾}{r_k}-{\stackrel{\‾}{r}}_k^{*}},b_k^R=\frac{255\·{\stackrel{\‾}{r}}_k^{*}}{{\stackrel{\‾}{r}}_k^{*}-\stackrel{\‾}{r_k}}\\ a_k^G=\frac{255}{{\stackrel{\‾}{g}}_k-{\stackrel{\‾}{g}}_k^{*}},b_k^G=\frac{255\·{\stackrel{\‾}{g}}_k^{*}}{{\stackrel{\‾}{g}}_k^{*}-{\stackrel{\‾}{g}}_k}\\ a_k^B=\frac{255}{{\stackrel{\‾}{b}}_k-{\stackrel{\‾}{b}}_k^{*}},b_k^B=\frac{255\·{\stackrel{\‾}{b}}_k^{*}}{{\stackrel{\‾}{b}}_k^{*}-{\stackrel{\‾}{b}}_k}\end{array}\right.---\left(4\right)$

Step 4: the value of mapping function parameter tries to achieve the mapping function of rear R, G, B tri-passages setting up each root CIS pixel value separately, as shown in formula (5),

$\left\{\begin{array}{c}{r}_k^{*}(i,j)=a_k^R\·r_k(i+j)+b_k^R\\ g_k^{*}(i,j)=a_k^G\·g_k(i,j)+b_k^G\\ b_k^{*}(i,j)=a_k^B\·b_k(i,j)+b_k^B\end{array}\right.---\left(5\right)$

Wherein, r _k(i, j), g _k(i, j), b _k(i, j) is respectively the pixel value of original image pixels point R, G, B tri-passages of corresponding CIS pipe, for the new pixel point value after function process, i, j are pixel position.

2. based on a kind of chromatic aberration calibrating method according to claim 1, it is characterized in that: target image is scanned, each pixel value that R, G, B tri-passages of the image scanned by each CIS pipe are corresponding substitutes into the formula (5) described in claim 1, obtain the pixel value after chromatic aberration correction, cover original image, form new image, the image namely after chromatic aberration correction.

3., based on a kind of background color filter method according to claim 1, it is characterized in that, comprising:

Step one: the background color drawing draft that has filtered needing background color scans, and obtains the raw information of its scan image, and pixel value corresponding for its R, G, B tri-passages is designated as r respectively _k(i, j), g _k(i, j), b _k(i, j),

Step 2: to original image f (x, y) carry out Fourier transform and change F (i into, j), select a filter transform function H (i, j), by decay F (i, j) radio-frequency component produces the Fourier transform G (i of output image, j), as shown in formula (6)

G(i,j)＝H(i,j)·F(i,j) (6)

Step 3: adopt Butterworth filter device, transfer function as shown in formula (7),

$H(i,j)=\frac{1}{1+{\left(\frac{D(i,j)}{D_0}\right)}^{2n}}---\left(7\right)$

$D(i,j)=\sqrt{{(i-\frac{\mathrm{row}}{2})}^2+{(j-\frac{\mathrm{col}}{2})}^2}$

Wherein, D ₀for cut-off frequency is apart from the distance of initial point, n is progression, D (i, j) for point (i, j) is to the distance of Fourier transform center (initial point), row, col are respectively picturedeep and columns, finally by Fourier inversion by G (i, j) output image g (i, j) is converted to, i.e. background image;

Step 4: the average calculating the pixel value obtaining background image R, G, B tri-passages corresponding to every root CIS pipe, is designated as prepare the black correction paper of a silent frame size, scanning obtains its original image information, and the average calculating image R, G, B tri-passage pixel values corresponding to every root CIS pipe is designated as if this parameter pre-configured, then skip this step in system;

Step 5: will substitute into formula according to claim 1 (4), calculate corresponding mapping function parameter, respectively by each pixel value r of R, G, B tri-passages of original image _k(i, j), g _k(i, j), b _k(i, j) substitutes into formula according to claim 1 (5) and calculates acquisition form new image.