KR100712467B1 - Apparatus for eliminating a color noise from CCD image - Google Patents

Abstract

The present invention is a device for removing color noise of an image including a color data interpolation device and a boundary preserver low pass filter. The color signal separator separates the input composite signal from the R, G, and B signals, and the median filtering unit receives a predetermined number of G signals to output data of the G component of the average value, and the pattern information extracting unit of the G component data of the average value. When the difference is higher than the threshold value, the difference is calculated and judges that there is a boundary component in the input composite signal, and outputs the determination signal. The color signal interpolation unit interpolates and outputs the separated R and B signals based on the determination signal. do. In addition, the area characteristic determination unit receives an input luminance signal and calculates smoothness in a predetermined region, and the adaptive low pass filter adjusts filter coefficients based on the calculated smoothness to low pass filter the luminance signal to preserve boundary components. As a result, color noise generated when interpolating the image signal photographed by the CCD element can be removed, and clear picture quality can be obtained without the problem of blurring the display screen.

Median filter, image data, CCD, RGB, interpolation, color noise, smoothness, sharpness, pixel

Description

Apparatus for eliminating a color noise from CCD image}

1 is a diagram for explaining a process of a general CCD image data interpolation process.

2 is a view for explaining three channels separated from the CFA of RGB.

3 is a block diagram of a color data interpolation apparatus of an apparatus for removing color noise of an image according to an exemplary embodiment of the present invention.

4 is a block diagram of an edge preserving low pass filter of an apparatus for removing color noise of an image according to an exemplary embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

110: color signal separation unit 120: median filter unit

130: pattern information extraction unit 140: color signal interpolation unit

210: region characteristic determination unit 220: adaptive low pass filter

230: Optical low pass filter 240: Multiplexer

The present invention relates to an apparatus for removing color noise of an image, and more particularly, to an apparatus for removing color noise of an image capable of minimizing sharp image quality and color noise in all optical applications using a CCD.

It is the field of image sensor that is most practically applied to the application of the charge-coupled device (CCD) technology, which was invented by Bell et al., Boyle et al. Sensors are used in a wide range of fields such as video cameras, surveillance cameras, medical cameras, industrial instrumentation, facsimile, image scanners and bar code readers.

In addition, the recent application to digital still cameras (Digital Still Camera) has been active, and in response to this CCD image sensor that has optimized the structure for the digital camera has also appeared.

As such, clear image quality and color noise removal are essential for all optical devices using a CCD. Especially, the importance of image quality is emphasized because digital image recording devices such as digital still cameras (DSC) are used to preserve still images.

However, since the RGB image signal captured by the CCD is input in a mosaic form, an interpolation process is performed to completely compensate for the missing color information by inferring the value of neighboring pixels as a basis to obtain an image having final RGB.

Although there are several kinds of color filter arrays (CFAs) of actual CCDs, and the interpolation methods are different from each other, here is an example of performing linear interpolation using CFA of a typical Bayer array. Both pictures are shown.

FIG. 1 is a diagram for explaining a process of a general CCD image data interpolation process, and FIG. 2 is a diagram for explaining three channels in which CFAs of RGB are separated.

Referring to Fig. 1, when the signal from the CCD is read as it is, an image on a mosaic corresponding to the CFA shown in Fig. 1 is obtained. Since it cannot be processed as such, the signal of the CCD filter is separated into three channels of RGB as shown in FIG. Here, the information lacking the blank portion. Looking at the G channel, there are blank pixels in the shape of a checkerboard. To fill this, it is based on the values of known pixels which are in close proximity. For example, various methods may be used, such as copying the left pixel value as it is, averaging two pixels on the left and right, and averaging four pixels on the top, bottom, left, and right sides.

However, when four pixels of up, down, left, and right are used, data for three lines is always processed. Therefore, it is necessary to prepare a corresponding buffer. This may be difficult when the number of pixels is large or when the bit depth is large.

In addition, since interpolation is uniformly performed irrespective of the amount of information included in each channel of RGB, the boundary between adjacent pixels becomes unclear and the contrast of the displayed image is reduced and the image is blurred.

The problem of the conventional interpolation processing method of CCD image data is that first, color noise is generated due to an error in the interpolation process of R and B components with relatively low frequency, and secondly, color noise during such signal processing. The screen is dimmed by low pass filtering, which is done to get rid of.                         

In general, CCDs, which are currently being used, are designed to transmit light projected through a lens to a sensor surface capable of measuring charge by the amount of light through a filter through which R, G, and B colors are transmitted.

However, it is impossible to obtain all of the three primary colors at positions corresponding to one pixel by means of color filters arranged on a plane. Therefore, all color components per pixel are interpolated using the surrounding values through an interpolation process, and color noise is generated by an error generated at this time.

By optically and numerically filtering to remove this noise, blurring of the image occurs. In order to compensate for this, edge enhancement is performed, but since the blurring of the image has already occurred, it is impossible to recover the lost data. In particular, there is a visually disturbing adverse effect such as a ringing effect near the boundary. It is likely to occur.

SUMMARY OF THE INVENTION The present invention has been made in an effort to solve such a conventional problem, and to provide an apparatus for removing color noise of an image capable of minimizing clear image quality and color noise in an optical application device using a solid-state image sensor.

Apparatus for removing color noise of an image according to one feature for achieving the above object,

A color signal separator for separating the input composite signal into R, G, and B signals;                     

A median filtering unit receiving the predetermined number of G signals and outputting data of G components having an average value;

A pattern information extraction unit for calculating a difference between the data of the G component of the average value and determining that a boundary component exists in the input composite signal when the difference is higher than a threshold value, and outputting a determination signal; And

And a color signal interpolation unit configured to output data of the interpolated R and B components by interpolating the separated R and B signals, respectively, based on the determination signal.

According to the apparatus for removing color noise of an image, it is possible to improve the image quality of CCD image data while maintaining the original shape of a conventional signal processing method.

Then, embodiments will be described so that those skilled in the art can easily implement the present invention.

3 is a block diagram of a color data interpolation apparatus of an apparatus for removing color noise of an image according to an exemplary embodiment of the present invention.

Referring to FIG. 3, the color data interpolation apparatus of the apparatus for removing color noise of an image according to an exemplary embodiment of the present invention may include a color signal separation unit 110 and a G signal that separate an input composite signal from R, G, and B signals. The median filter unit 120 that receives a predetermined number and outputs the data of the G component of the average value, calculates the difference between the data of the G component of the average value and determines that the boundary component exists in the input composite signal when the difference is higher than the threshold value. The pattern information extracting unit 130 outputs the determination signal, and the color signal interpolation unit 140 interpolating and outputting the separated R and B signals based on the determination signal.

Hereinafter, the operation of the color data interpolation apparatus of the apparatus for removing color noise of an image according to the present invention including such a structure will be described in detail.

The color signal separation unit 110 receives a signal in which an RGB signal is composited, separates them, outputs the separated G signal to the median filter unit 120, and outputs the separated R and B signals to the color signal complement. Output to the executive 140. In this case, since each of the RGB signals output from the CCD color filter generally has a ratio of 1: 2: 1, the R and B signals that are compositely input based on the G signal having more information than the R or B signals are included. Interpolate

The median filter unit 120 receives a predetermined number of G signals, calculates an average value of the provided G component signals, and outputs the data pixels of the G components. In this case, the median filter generally arranges the input signals in the order of magnitude. When the number of the arranged signals is odd, the median filter outputs one signal having the center size from the signals arranged in the order of magnitude, and the number of the arranged signals When the number is even, the filter outputs the average value of two measured signals located in the middle of the signals arranged in magnitude order.

The pattern information extracting unit 130 calculates a difference between the average values of the data pixels of the G component output from the median filter unit 120 and determines that a boundary component exists when the difference is greater than the threshold, and determines the determination signal as the color signal interpolation unit 140. ) At this time, the extraction of the pattern information is performed by, for example, calculating a difference between neighboring green data in a 2 × 2 area and determining that a boundary component exists when the difference exceeds a predetermined threshold. The boundary component determines four types of horizontal, vertical and two diagonal directions.                     

The color signal interpolator 140 interpolates the R and B components based on the determination signal output from the pattern information extractor 130 and outputs the interpolated R component signal and the B component signal.

According to such a color data interpolation apparatus, the data generated in the conventional interpolation process by interpolating the data of the R component and the data of the B component based on the data of the G component having more information than the data of the R or B component in a predetermined region. Color noise can be reduced.

4 is a block diagram of an edge preserving low pass filter of an apparatus for removing color noise of an image according to an exemplary embodiment of the present invention.

Referring to FIG. 4, the boundary preserved low pass filter of the apparatus for removing color noise of an image according to an exemplary embodiment of the present invention may include a region characteristic determiner 210, an adaptive low pass filter 220, and an optical low pass filter 230. And a multiplexer 240 to calculate the smoothness of the region based on the pattern structure in the predetermined region with respect to the luminance signal Y, determine the filter coefficients, and low-pass the luminance signal based on the determined filter coefficients. Filter through.

Hereinafter, the boundary preservation low pass filter operation of the apparatus for removing color noise of an image including such a structure will be described in more detail.

The region characteristic determiner 210 calculates the smoothness of the region with respect to the luminance data included in the predetermined region, and provides the adaptive low pass filter 220 when the calculated smoothness is lower than the predetermined value, and calculates the smoothness. If is greater than a predetermined value, it is provided to the optical low pass filter 230 that is commonly used. The smoothness calculated here is basically determined by the variance (σ) of the brightness A in the area A and the similar values connected to each other. Based on this, is there a boundary component in the area or is it a flat surface? It is determined whether or not it is a realm.

The adaptive low pass filter 220 adjusts the filter coefficients when the smoothness is low by the region characteristic determiner 210, removes the luminance signal input by the adjusted filter coefficients, and provides the multiplexer 240 after the low pass filtering is removed. In addition, when the smoothness is high by the region characteristic determination unit 210, the optical low pass filter 230 removes the luminance signal from the low pass filtering and provides the multiplexer 240 after low pass filtering.

The multiplexer 240 outputs any one of luminance signals output from the adaptive low pass filter 220 or the optical low pass filter 230 in synchronization with the region characteristic determiner 210.

In the above, the output path of the luminance signal inputted according to the smoothness by the region characteristic determining unit 210 is different from the adaptive low pass filter 220 or the optical low pass filter 230, but the optical low pass filter is used. If 230 is removed and the filter coefficients adaptively adjusted according to the smoothness are configured to include the filter coefficients of the optical low pass filter, the number of components may be reduced.

As mentioned above, in general, the conventional low pass filtering applied to digital image data is performed by convolution using a mask having a predetermined size. In the present invention, prior to the multiplication and accumulation operation of the mask, The process of adaptively determining the coefficients is added to determine the filter coefficients based on the pattern structure in a given region.

The interpolation device and boundary preservation low pass filter of the color data according to the present invention can be applied alone or jointly to the signal processing of CCD image data.

As an example, it is applied to the process of separating and interpolating a composite RGB image signal, and it is generally used by interpolating and outputting R and B signals having relatively low information based on G signals having relatively much information. As compared with the interpolation process, generation of color noise due to an error generated during the interpolation process can be reduced.

In order to compensate for the problem that image blur occurs by optically and numerically filtering the luminance signal to remove such color noise, it is impossible to recover an image that has already been blurred, such as edge enhancement. It is also possible to prevent visually annoying adverse effects such as ringing effects near the boundary.

As described above, the present invention can minimize the color noise in the final image data by improving the interpolation performance of the red and blue pixel data having a smaller number of pixel data passing through the color filter of the CCD than green.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be able to variously modify and change the present invention without departing from the spirit and scope of the invention as set forth in the claims below. It will be appreciated.

 As described above, in accordance with the present invention, an apparatus for removing color noise of an image that performs a new color interpolation method based on the geometry of image data, and an optical low range generally used to fundamentally prevent blur of an image is proposed. Instead of employing a pass filter, an adaptive low pass filter having a filter coefficient that varies according to the smoothness of a surface may be employed to smooth the image while preserving boundary components of the image.

Accurate calculation of color signals also helps to minimize color noise and similar arisings that can occur in black and white areas, smoother flat areas, and maximum sharpness allowed by the resolution of the lens around the perimeter. Can be maintained.

Claims (4)

In the device for removing the color noise of the image based on the composite signal and the luminance signal of the image, A color signal separator for separating the input composite signal into R, G, and B signals; A median filtering unit receiving the predetermined number of G signals and outputting G component data having an average value; A pattern information extracting unit which calculates a difference between the sequentially output G component data and checks that a boundary component exists in the composite signal when the difference is higher than a threshold and outputs a determination signal; A color signal interpolation unit outputting data of the interpolated R and B components by interpolating the R and B signals separated from the color signal separation unit based on the determination signal output from the pattern information extraction unit, respectively; An area characteristic determiner configured to receive an input luminance signal and calculate smoothness within a predetermined area; And A boundary preservation low pass filter having an adaptive low pass filter for preserving boundary components of an input luminance signal by low pass filtering the luminance signal according to a filter coefficient adjusted based on the smoothness calculated by the region characteristic determination unit. Apparatus for removing color noise of an image comprising a. delete The method of claim 1, And an optical low pass filter for general low pass filtering the luminance signal when the calculated luminance signal has high smoothness. 4. The apparatus of claim 3, wherein the luminance signal is provided to an adaptive low pass filter whose filter coefficient is adjusted when the calculated luminance signal has low smoothness.

Images