CN103747222B - A kind of method that color artifact of image suppresses and device - Google Patents

Abstract

The invention discloses method and the device of the suppression of a kind of color artifact of image, comprise the following steps: the RGB image inputting a rgb color space; RGB image is transformed to a transform domain color space from rgb color space, obtains luminance component and chromatic component; Filtering process is weighted to chromatic component, obtains the colourity after weighted filtering and suppress output valve; Colourity suppressed output valve and luminance component from described transform domain colour space transformation to rgb color space, obtain the RGB image after false color suppression and export.By the RGB image after demosaicing process is transformed into transform domain LC1C2 color space from rgb color space, obtain luminance component and the chromatic component of RGB image, filtering process is weighted to chromatic component, obtain filtered chromatic component, then filtered chromatic component and former luminance component are transformed into rgb color space again from transform domain LC1C2 color space, obtain the RGB image after false color suppression, reach the effect suppressing color artifact of image simply and effectively.

Description

A kind of method that color artifact of image suppresses and device

Technical field

The present invention relates to field of video image processing, particularly relate to method and the device of the suppression of a kind of simple color artifact of image efficiently.

Background technology

In camera system, except digital camera that minority is special can adopt full color filtered sensor, most of CCD or cmos image sensor, in order to Simplified flowsheet, reduce costs, all adopt color filter array (ColorFilterArray, be called for short CFA), according to pixels R, G, B three primary colors are alternately recorded in locus, namely only can record the luminous intensity of a kind of color in R, G, B three primary colors at each location of pixels.The image of such record is that a kind of mosaic image is called Bayer picture format, as shown in Figure 1, has four kinds of conventional Bayer forms.Obtain full-color image, just must recover the component color of disappearance from Bayer format-pattern.This technology from Bayer form recovery full-color image is called demosaicing (Demosaicing) or color interpolation (ColorInterpolation) technology.But in image detail and edge, demosaicing technology easily causes the appearance of false color coloured silk (colorartifact) usually, affects picture quality.Therefore the false colour caused after being necessary to suppress demosaicing, improves picture quality.General false color suppresses usually to need rgb color space to be transformed into the enterprising row relax in HSV or YUV territory, but process is more consuming time like this, also needs larger resource consumption simultaneously, needs larger color space conversion to calculate.Therefore, a kind of simple and effective false color is provided to suppress extremely to be necessary.

Summary of the invention

Technical problem to be solved by this invention be to provide a kind of suppress demosaicing technology to cause simply and effectively image detail and the method that suppresses of the color artifact of image of false colour phenomenon of edge and device.

For realizing above-mentioned technique effect, the invention discloses a kind of method that color artifact of image suppresses, comprising the following steps:

Input the RGB image of a rgb color space;

Described RGB image is transformed to a transform domain color space from described rgb color space, obtains luminance component and the chromatic component of described RGB image;

Filtering process is weighted to described chromatic component, obtains the colourity after weighted filtering and suppress output valve;

Suppress output valve and described luminance component from described transform domain colour space transformation to rgb color space described colourity, obtain the RGB image after false color suppression;

Export the RGB image after the suppression of described false color.

Described method is further improved and is, is weighted filtering process comprises the following steps described chromatic component:

Select the neighborhood window centered by current point;

Filtering process is carried out to the chromatic component in described neighborhood window, obtains the iotaltered chroma component of current point;

According to the weighted filter coefficients required for the scene complexity determination subsequent weight filtering of described neighborhood window;

The colourity utilizing the iotaltered chroma component of described current point and described weighted filter coefficients to calculate current point suppresses output valve.

Described method is further improved and is, described weighted filter coefficients value is less than or equal to 1.

Described method is further improved and is, the mode of described filtering process comprises mean filter and medium filtering.

The invention also discloses the device that a kind of color artifact of image suppresses, comprising:

One RGB image input units, the RGB image of described input is the RGB image of a rgb color space;

One first colour space transformation unit, connects described RGB image input units, completes the conversion of described rgb color space to transform domain color space;

One weighted filtering unit, connects described first colour space transformation unit, is weighted filtering process to the RGB image of described transform domain color space, obtain the RGB image after weighted filtering;

One second colour space transformation unit, connects described weighted filtering unit, completes the conversion of described transform domain color space to rgb color space;

One RGB image output unit, connects described second colour space transformation unit, and the RGB image of described output is the RGB image after false color suppresses.

Described device further improves and is, is connected chrominance component filter unit further between described weighted filtering unit with described first colour space transformation unit.

Described device further improves and is, is also connected a scene complexity computing unit further between described weighted filtering unit with described first colour space transformation unit.

Described device further improves and is, the RGB image of described input is the RGB image after demosaicing process

The present invention is owing to have employed above technical scheme, making it have following beneficial effect is: by the RGB image after demosaicing process is transformed into transform domain LC1C2 color space from rgb color space, obtain luminance component and the chromatic component of RGB image, filtering process is weighted to chromatic component, obtain filtered chromatic component, then filtered chromatic component and former luminance component are transformed into rgb color space again from transform domain LC1C2 color space, obtain the RGB image after false color suppression, reach the effect suppressing color artifact of image simply and effectively.

Accompanying drawing explanation

Fig. 1 ~ Fig. 4 is four kinds of conventional Bayer forms.

Fig. 5 is the structural representation of the device that a kind of color artifact of image of the present invention suppresses.

Fig. 6 is the transformation matrix figure of the first colour space transformation unit in the present invention.

Fig. 7 is the transformation matrix figure of the second colour space transformation unit in the present invention.

Fig. 8 is the flow chart of a preferred embodiment of the method that a kind of color artifact of image of the present invention suppresses.

Embodiment

Below in conjunction with accompanying drawing and embodiment, the present invention is further detailed explanation.

Consult shown in 5, the invention discloses the device that a kind of color artifact of image suppresses, comprise a RGB image input units, one first colour space transformation unit 101, chrominance component filter unit 102, scene complexity computing unit 103, weighted filtering unit 104,1 second colour space transformation unit 105 and a RGB image output unit.The RGB image inputted in RGB image input units is the rgb color space image after demosaicing process.

First colour space transformation unit 101 realizes the conversion of rgb color space to LC1C2 transform domain color space, and the transformation matrix A of employing as shown in Figure 6.Wherein L represents luminance component, and C1, C2 represent chroma component information, reflection be color difference signal between R, G, B.Concrete transformation calculations is shown in following formula (1).

$\left[\begin{array}{c}L\\ C1\\ C2\end{array}\right]=A\×\left[\begin{array}{c}R\\ G\\ B\end{array}\right]---\left(1\right)$

Chromatic component filter unit 102 couples of chromatic components C1, C2 carry out filtering calculating.First, select the certain neighborhood window centered by current point, such as select the neighborhood window of 5x5; Secondly, carry out filtering process respectively to chromatic component C1, C2 in neighborhood window, concrete filtering mode can adopt the filtering mode of mean filter or medium filtering, can certainly adopt more complicated filtering method.Filtered components C1', the C2' of central point is obtained after chromatic component filter unit 102 filtering process.

Weighted filter coefficients AlphaC1, AlphaC2(AlphaC1, AlphaC2 value required for the filtering of scene complexity determination subsequent weight of the neighborhood window that scene complexity computing unit 103 is selected by calculating chromatic component filter unit 102 is no more than 1).Following formula (2) is shown in the concrete calculating of AlphaC1, AlphaC2.Wherein, StdL, StdC1, StdC2 represent the standard deviation of the neighborhood window data that chromatic component filter unit 102 is selected.

$\begin{array}{c}\mathrm{AlphaC}1=\frac{\mathrm{StdC}1}{\mathrm{StdL}}\\ \mathrm{AlphaC}2=\frac{\mathrm{StdC}2}{\mathrm{StdL}}\\ \mathrm{AlphaC}1=\min \left(\mathrm{AlphaC}\mathrm{1,1}\right)\\ \mathrm{AlphaC}2=\min \left(\mathrm{AlphaC}\mathrm{2,1}\right)\end{array}---\left(2\right)$

Weighted filtering unit 104 calculates colourity and suppresses component C1o, C2o, calculates and sees following formula (3).When scene is complicated (there is details or edge) then AlphaC1, AlphaC2 value can smaller, at this moment tend to selection C1', C2' output; Otherwise tend to select primary colors degree C1, C2 to export.

C1o=AlphaC1*C1+(1-AlphaC1)*C1'(3)

C2o=AlphaC2*C2+(1-AlphaC2)*C2'

Second colour space transformation unit 105 realizes the conversion of LC1C2 transform domain color space to rgb color space, and the transformation matrix C of employing as shown in Figure 7.Concrete transformation calculations is shown in following formula (4).

$\left[\begin{array}{c}R\\ G\\ B\end{array}\right]=C\×\left[\begin{array}{c}L\\ C1o\\ C2o\end{array}\right]---\left(4\right)$

Consulting shown in Fig. 8, is the concrete embodiment flow chart of the method that a kind of color artifact of image of the present invention suppresses.

Step 201, RGB image input units, the RGB image after the process of input demosaicing.

Step 202, rgb color space, to LC1C2 colour space transformation, complete the transformation calculations from rgb color space to LC1C2 color space, obtain luminance component L and chromatic component C1, C2.

Step 203, chromatic component filtering, realize the filtering process of C1, C2 chromatic component.In this example, choose the neighborhood window of the 5x5 centered by current point, and adopt the filtering mode of medium filtering, chromatic component C1', C2' after calculation of filtered.

Step 204, scene complexity calculate, scene complexity StdL, StdC1, StdC2(StdL of the neighborhood window of calculation procedure 203 Unit selection are the standard deviations of the L data window of 5x5, StdC1 is the standard deviation of the C1 data window of 5x5, and StdC2 is the standard deviation of the C2 data window of 5x5); Weighted filter coefficients AlphaC1, AlphaC2 again required for calculated for subsequent weighted filtering.

Step 205, weighted filtering, the colourity calculating current point suppresses output valve C1o, C2o.

Step 206, LC1C2 color space, to rgb color space conversion, suppress output valve C1o, C2o according to the colourity that step 205 calculates, and the luminance component L that step 202 calculates, and conversion obtains R, G, B output that false color suppresses.

Step 207, RGB image export, and export the RGB image after false color suppression to subsequent treatment.

Below embodiment is to invention has been detailed description by reference to the accompanying drawings, and those skilled in the art can make many variations example to the present invention according to the above description.Thus, some details in embodiment should not form limitation of the invention, the present invention by the scope that defines using appended claims as protection scope of the present invention.

Claims (4)

1. a method for color artifact of image suppression, is characterized in that said method comprising the steps of:

Input the RGB image of a rgb color space, the described RGB image of input is the RGB image after demosaicing process;

Described RGB image is transformed to a transform domain color space from described rgb color space, obtains luminance component L and chromatic component C1, C2 of described RGB image;

Filtering process is weighted to described chromatic component C1, C2, obtains the colourity after weighted filtering and suppress output valve C1o, C2o, by following steps, filtering process is weighted to described chromatic component C1, C2:

Select the neighborhood window centered by current point;

Filtering process is carried out to chromatic component C1, C2 in described neighborhood window, obtains iotaltered chroma component C1', C2' of current point;

Calculate the scene complexity of described neighborhood window;

According to weighted filter coefficients AlphaC1, AlphaC2 required for the scene complexity determination subsequent weight filtering of the described neighborhood window calculated, wherein, the computing formula of AlphaC1, AlphaC2 is as follows:

AlphaC1＝min(AlphaC1,1)

AlphaC2＝min(AlphaC2,1)

StdL, StdC1, StdC2 represent the standard deviation of described neighborhood window data;

Utilize iotaltered chroma component C1', C2' of described current point and described weighted filter coefficients AlphaC1, AlphaC2 to calculate described colourity and suppress output valve C1o, C2o, wherein, the computing formula of C1o, C2o is as follows:

C1o＝AlphaC1*C1+(1-AlphaC1)*C1'

C2o＝AlphaC2*C2+(1-AlphaC2)*C2'；

Suppress output valve C1o, C2o and described luminance component L from described transform domain colour space transformation to rgb color space described colourity, obtain the RGB image after false color suppression;

Export the RGB image after the suppression of described false color.

2. the method for claim 1, is characterized in that: described weighted filter coefficients value is less than or equal to 1.

3. the method for claim 1, is characterized in that: the mode of described filtering process comprises mean filter and medium filtering.

4. a device for color artifact of image suppression, is characterized in that comprising:

One RGB image input units, for inputting the RGB image of a rgb color space, the described RGB image of input is the RGB image after demosaicing process;

One first colour space transformation unit, connect described RGB image input units, for completing the conversion of described rgb color space to transform domain color space, described RGB image is transformed to transform domain color space from described rgb color space, obtains luminance component L and the chromatic component of described RGB image;

Chrominance component filter unit, connects described first colour space transformation unit, for selecting the neighborhood window centered by current point, carrying out filtering process to the described chromatic component in described neighborhood window, obtaining the iotaltered chroma component of current point;

One scene complexity computing unit, connects described first colour space transformation unit, for calculating the scene complexity of described neighborhood window, and the weighted filter coefficients required for the scene complexity determination subsequent weight filtering of the described neighborhood window calculated;

One weighted filtering unit, connects described chromatic component filter unit and described scene complexity computing unit, and iotaltered chroma component and described weighted filter coefficients for utilizing described current point calculate colourity and suppress output valve; One second colour space transformation unit, connect described weighted filtering unit, for suppressing output valve and described luminance component from described transform domain colour space transformation to rgb color space described colourity, obtain the RGB image after false color suppression, complete the conversion of described transform domain color space to rgb color space;

One RGB image output unit, connects described second colour space transformation unit, and the RGB image of described output is the RGB image after false color suppresses.