JP2001274995A - Noise eliminator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simultaneously carry out the noise elimination and outline emphasize and to optimize the noise elimination processing and outline emphasis processing when both noise elimination and outline emphasis are simultaneously carried out. SOLUTION: An HPF 12 transmits input image data D1 through a high sensitive area, and a memory 13 outputs the HPF output D2 after inverting the positive and negative codes at a comparatively low level and outputs the output D2 in the same code at a comparatively high level. The output D3 of the memory 13 is applied to an adder 11 and added to the data D1 so that the noise elimination and outline emphasis are simultaneously carried out. Meanwhile, the cutoff frequency (fcut) of the HPF 12 is set at a high level when the data D1 has a small size and at a low level when the data D1 has a large size respectively. Furthermore, the input/output gain of the memory 13 is increased and decreased when the size of the data D1 is small and large respectively.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise eliminator for eliminating noise in an image, and more particularly to a noise eliminator suitable for a digital full-color printer.

[0002]

2. Description of the Related Art As a conventional example for removing noise from an image, a method using a recursive filter as disclosed in, for example, Japanese Patent Application Laid-Open No.
A method using a median filter has been proposed as disclosed in Japanese Patent Publication No. 200579. Here, a common problem in removing noise in an image is “how to remove noise (high-frequency small-amplitude signal) while preserving the outline of the image”. Since the “image outline” is a component having a large amplitude change, the noise elimination device includes a noise detection circuit and an LPF.
(A median filter is also a kind of LPF.) As shown in FIG. 5, the highest region of the small amplitude signal in the input image signal is attenuated (particularly when luminance noise is removed). A conventional example of enhancing the outline of an image is disclosed in, for example, JP-A-10-290368.

Incidentally, not only printers but also image processing systems equipped with a noise removal circuit often include an outline emphasis circuit. TV and VTR are representative. The purpose of the contour emphasizing circuit is to improve the sharpness of the image. For this reason, as shown in FIG. 6, the highest region of the input image signal having the larger amplitude signal than the small amplitude signal is emphasized. If is set too large, image noise becomes noticeable. Generally, a Laplacian filter 1 is used as shown in FIG. 7 to enhance the outline of an image. However, since this also amplifies noise, a coring processing unit 2 is used.
The output of the coring processing unit 2 and the input image data are added (shown in FIG. 3) together with (correction restriction of the small amplitude signal). That is, as shown in FIGS. 5 and 6, noise removal and contour emphasis are in front and back.

[0004]

However, noise,
In particular, removal of luminance noise is performed on a visually sensitive region (hereinafter, a high sensitivity region), and enhancement of an outline is performed on a region from the high sensitivity region to the highest region. Is most effective. For this reason, in the case of a printer, as shown in FIG. 8, the noise removing unit 4 is arranged before the pixel density converting unit 5 and the contour emphasizing unit 6 is arranged after the pixel density converting unit 5. Here, in the case of a printer, since the resolution of the input image is generally smaller than the resolution of the output image,
The effect of performing the noise removal processing on the highest area close to the high sensitivity area of the input image is more effective, and the amount of calculation (circuit scale) can be reduced. On the other hand, the contour emphasis processing is prioritized to apply to the highest area of the output image, and is therefore arranged after the pixel density conversion unit 5. Therefore, conventionally, the noise removal unit 4
There is an advantage in that the contour emphasizing unit 6 and the contour emphasizing unit 6 are independent.

However, in recent years, the resolution of an input image of a printer has tended to increase, for example, as in a high-pixel-count CCD of a digital camera, and the resolution of the input image is closer to the resolution of the output image. In addition, if the output resolution of the printer is improved to be far from the high sensitivity area, there is a problem that sufficient effects cannot be obtained by the edge enhancement processing using the Laplacian filter 1. For this reason, there is a problem in that the noise removal unit 4 and the outline emphasizing unit 6 cannot be sufficiently dealt with simply by arranging them before and after the pixel density conversion unit 5 independently.

SUMMARY OF THE INVENTION In view of the above problems, an object of the present invention is to provide a noise eliminator capable of simultaneously processing noise elimination and contour enhancement. Another object of the present invention is to provide a noise elimination device that can optimize the noise elimination processing and the outline enhancement processing when the noise elimination and the outline enhancement are simultaneously performed.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention inverts the sign of a signal in a high sensitivity area at a relatively low level and inputs the same sign at a relatively high level. This is added to the image data. When the size of the input image data is small, the cutoff frequency of the filter is increased. When the size of the input image data is large, the cutoff frequency of the filter is lowered. The output gain is increased, and when the size is large, the input / output gain of the memory is reduced.

That is, according to the present invention, a filter that passes a high-sensitivity region of input image data and a filter that outputs the output of the filter with the sign inverted at a relatively low level and the same sign at a relatively high level A noise output device, and an adder for adding the input image data and the output of the memory.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a noise elimination device according to the present invention, FIG. 2 is an explanatory diagram showing the configuration of the high-pass filter of FIG. 1, and FIG. 3 is an explanatory diagram showing input / output characteristics of the memory of FIG. FIGS. 4A and 4B are explanatory diagrams showing characteristics of the noise elimination device of FIG.

The apparatus shown in FIG. 1 generally determines a unit filter block 10 and a size (number of pixels) of "small", "standard" or "large" in order to determine the resolution of an input image. It is configured by an image determining unit 14 that performs the processing. The unit filter block 10 includes an adder 11 and an HPF
(High-pass filter) 12 and a memory 13. The input image data D1 is provided by an adder 11 and an HPF 12
And the HPF 12 passes through the high sensitivity region (> fcut) of the input image data D1.

The output D2 of the HPF 12 is applied as an address of the memory 13, and the memory 13 outputs the HPF output D2 with the sign inverted at a relatively low level and the same sign at a relatively high level. Then, the output D3 of the memory 13 is applied to the adder 11, and is added to the input image data D1 (= D1 + D3). Also, HPF1
2 and the characteristics of the memory 13 according to the resolution of the input image.
That is, switching is performed according to the input image size “small”, “standard”, and “large”. The unit filter block 10 performs one-dimensional processing and performs processing in the sub-main scanning direction after performing processing in the main scanning direction when processing a two-dimensional image.

The HPF 12 has a memory (not shown), and switches the contents of the memory according to the input image size “small”, “standard”, and “large” to reduce the number of pixels to eight pixels in the main scanning direction as shown in FIG. On the other hand, filters having three types of cutoff characteristics can be selected. The first filter multiplies the pixel value of interest, which is the center pixel of the eight pixels, by a coefficient = 2 for the case where the image size is “small”, and the coefficient = 2 is assigned to the values of 1 + 1 = 2 pixels on both sides thereof. Multiply by −1, multiply the other four pixels by a coefficient = 0, add the results of each multiplication, divide this added value by the number of pixels = 8, and output.

The second filter has a standard image size.
For the case of, the pixel value of interest is multiplied by a coefficient = 4, and the values of 2 + 2 = 4 pixels on both sides thereof are both multiplied by a coefficient = −1,
The other two pixels are both multiplied by a coefficient = 0, the results of each multiplication are added, and the added value is divided by the number of pixels = 8 and output. Third
Filter multiplies the pixel value of interest by a coefficient = 6 for the case where the image size is “large”, multiplies the values of 3 + 3 = 6 pixels on both sides thereof by a coefficient = −1, and adds each multiplication result. Then, the sum is divided by the number of pixels = 8 and output. Therefore, the cutoff frequencies fcu of these three filters are
t is highest for the first filter and lowest for the third filter. Here, when the input image data D1 is 8 bits, the output data D2 of the HPF 12 is 9 bits with ± sign.

Similarly to the HPF 12, the memory 13 is configured such that input / output characteristics such as gain can be switched according to the input image size "small", "standard", or "large". As shown in FIG. 3, in the relatively low noise removal area (−D11 <D2 <+ D11), the memory 13 outputs data D3 which is the inverse of the data D2, and outputs a relatively high contour enhancement area (D2).
<−D11, + D11 <D2), data D of the same sign
3 is output. For this reason, when the adder 11 adds the input image data D1 and the memory output D3, the input image data D1 becomes smaller (D1 + D3 <D1) in the noise removal region (-D11 <D2 <+ D11). Becomes smaller, and the noise removal processing is performed. In addition, the outline emphasis area (D2 <−D11, + D
11 <D2), the input image data D1 becomes large (D
1 + D3> D1), the high-frequency component of the image becomes large,
Therefore, the outline emphasis processing is performed.

Therefore, the unit filter block 10 can simultaneously perform noise removal and contour enhancement. Further, by changing the data D3 of the memory 13 according to the image size, the noise removal area (−D11 <D2 <+
D11) width, gain, and edge enhancement area (D2 <−D1)
1, + D11 <D2) can be freely set. Further, as long as there is no discontinuity in the data D3 of the memory 13, the image does not become unnatural near the boundary D11 between the noise removal area and the contour emphasis area, for example. Further, when the input image data D1 is 8 bits, the HPF 12
Since the output data D2 is 9 bits with ± sign,
When the data D3 = 8 bits, the capacity of the memory 13 may be 512 bytes.

Here, the HPF 12 is an FIR filter and needs to be symmetrical because it handles image data. Further, the cutoff frequency fcut of the HPF 12 can be changed according to the image size. Then, this unit filter block 10 is used as one one-dimensional filter.
The final result is obtained by processing the main scanning direction and the sub-scanning direction.

Next, the image determining unit 14 will be described. The image determination unit 14 sets the cutoff frequency fcut in the HPF 12 according to the size (the number of pixels) so that the filtering process is performed in the high-sensitivity region, and obtains an appropriate gain with the change in the filter characteristics. The data D3 of the memory 13 is set so as to be stored. Specifically, the noise removal area (-D
The width and gain of 11 <D2 <+ D11) and the gain and characteristics of the outline emphasis area (D2 <-D11, + D11 <D2) are changed (or selected). At this time, if the input image size is large, the cutoff frequency fcut is lowered, and if it is small, it is raised. Here, when the processing result is output to the printer, the output resolution is almost constant in most cases. Therefore, in order to detect a high-sensitivity area, the area can be detected from the input image size (the number of pixels).

Note that the determination parameters of the image determination section 14 are not limited to the input image size. For example, if the noise level in the input image is known in advance, the noise removal area (−
D11 <D2 <+ D11) can be changed. The present invention can be realized by hardware, software, or a combination thereof.

[0019]

As described above, according to the present invention, the sign of the signal in the high sensitivity region is inverted at a relatively low level, and the same sign is added to the input image data at a relatively high level. As a result, noise removal and contour enhancement can be simultaneously performed. Further, according to the present invention, when the size of the input image data is small, the cutoff frequency of the filter is increased, when the size is large, the cutoff frequency of the filter is reduced, and when the size of the input image data is small, When the size is large, the memory input / output gain is reduced, so the noise removal processing and contour enhancement processing are optimized when noise removal and contour enhancement are performed simultaneously. Can be

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of a noise elimination device according to the present invention.

FIG. 2 is an explanatory diagram showing a configuration of a high-pass filter of FIG.

FIG. 3 is an explanatory diagram showing input / output characteristics of the memory of FIG. 1;

FIG. 4 is an explanatory diagram showing characteristics of the noise elimination device of FIG. 1;

FIG. 5 is an explanatory diagram showing characteristics of a conventional noise elimination device.

FIG. 6 is an explanatory diagram showing characteristics of a general edge enhancement device.

FIG. 7 is a block diagram showing a conventional edge enhancement device.

FIG. 8 is a block diagram showing a noise removal and contour enhancement circuit of a conventional printer.

[Explanation of symbols]

 Reference Signs List 11 adder 12 HPF (filter) 13 memory 14 image determination unit (image determination means)

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B057 CA12 CA16 CB12 CB16 CE02 CE03 CE06 CH09 DB02 DB06 5C021 PA06 PA18 PA33 PA39 PA53 PA58 PA62 PA66 PA78 RB03 RB04 XB03 YA01 5C077 LL02 MP07 PP03 PP60 PQ08 PQ12 PQ22

Claims (3)

[Claims] 1. A filter for passing a high-sensitivity region of input image data, and a memory for inverting the output of the filter at a relatively low level, inverting the sign, and outputting the same sign at a relatively high level. A noise removing device comprising: an adding unit configured to add the input image data and an output of the memory. 2. The filter according to claim 1, wherein the cutoff frequency of the filter is increased when the size of the input image data is small, and the cutoff frequency of the filter is decreased when the size of the input image data is large. Noise removal device. 3. The input / output gain of the memory is increased when the size of the input image data is small, and the input / output gain of the memory is decreased when the size of the input image data is large. The noise removing device according to any one of the preceding claims.