JPH07283971A - Noise reducer - Google Patents

Abstract

PURPOSE:To suppress block distortion at an edge part and to sufficiently eliminate a white noise on the flat part of an image. CONSTITUTION:Image data is DCT-transformed in an orthogonal transformation circuit 1, and a coefficient in accordance with a white noise component is reduced from a DCT coefficient in a noise coring circuit 2. Inverse DCT transformation is applied to the DCT coefficient outputted from the noise coring circuit 2 in an inverse orthogonal transformation circuit 3. An edge detection circuit 11 detects the presence/absence of the edge of the image data, and performs such control so as to reduce the quantity to be reduced of the DCT coefficient in the noise coring circuit 2 when the edge exists.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise reducer for suppressing a noise component of image data.

[0002]

2. Description of the Related Art In order to suppress a noise component included in image data, image data is orthogonally transformed, and a component corresponding to white noise is reduced from an output obtained by the orthogonal transformation.
There is known a process of performing inverse orthogonal transform to transform the original image data. By performing such processing, it is possible to remove white noise included in the image data.

[0003]

However, if the white noise component is removed from the data obtained by the orthogonal transformation, block distortion will occur in the edge portion containing many high frequency components. In order to suppress the occurrence of block distortion, it is necessary to reduce the amount of data to be removed from the orthogonal transform output, but doing so causes the problem that white noise in the flat part of the image cannot be removed cleanly. there were.

The present invention has been made in view of such a situation, and it is possible to suppress the block distortion in the edge portion and to sufficiently remove the white noise in the flat portion.

[0005]

A noise reducer of the present invention extracts a noise component from an orthogonal transform means (for example, the orthogonal transform circuit 1 in FIG. 1) for orthogonal transforming input image data and an output of the orthogonal transform means. Removal means for removing (eg, FIG. 1)
Noise coring circuit 2), an inverse orthogonal transforming means for performing an inverse orthogonal transform on the output of the removing means (for example, the inverse orthogonal transforming circuit 3 in FIG. 1), the edge of the image data is detected, and the detected result is converted into Control means for controlling the removing means (eg, FIG. 1)
Edge detection circuit 11) of FIG.

When the image data has many edges, the control means can reduce the removal rate of the noise component in the removal means as compared with when the image data has few edges.

Further, selection means for selecting one of the image data that has been subjected to the inverse orthogonal transformation by the inverse orthogonal transformation means and the image data that has not undergone the processing of the orthogonal transformation means, the removal means, and the inverse orthogonal transformation means (for example, FIG. 2 selector 22) is further provided, and when the control means has many edges of image data,
It is possible to control the selection means to select the image data that has not been processed by the orthogonal transformation means, the removal means, and the inverse orthogonal transformation means.

[0008]

In the noise reducer having the above structure, the image data is DCT-transformed by the orthogonal transformation circuit 1, and then the white noise component is removed from the DCT coefficient in the noise coring circuit 2. Then, the DCT coefficient output from the noise coring circuit 2 is subjected to inverse DCT conversion in the inverse orthogonal transform circuit 3. The edge detection circuit 11 detects the presence or absence of an edge, and when there is an edge, D
The removal rate of the noise component of the CT coefficient is reduced.

Therefore, it is possible to suppress the block distortion at the edge portion of the image data and sufficiently remove the white noise.

[0010]

1 is a block diagram showing the configuration of an embodiment of a noise reducer according to the present invention. Image data input from a device (not shown) is supplied to the orthogonal transformation circuit 1 and the edge detection circuit 11. The orthogonal transformation circuit 1 orthogonally transforms the input image data. For this orthogonal transformation,
For example, DCT transform (discrete cosine transform) can be used. The output of the orthogonal transformation circuit 1 is supplied to the noise coring circuit 2 where the component corresponding to white noise is removed. The output of the noise coring circuit 2 is supplied to the inverse orthogonal transform circuit 3, subjected to inverse orthogonal transform, and then output as image data.

The edge detection circuit 11 detects the presence or absence of an edge from the input image data, and controls the noise coring circuit 2 according to the detection result.

Next, the operation will be described. The image data is, for example, DCT-transformed in the orthogonal transformation circuit 1, and the DCT coefficient is supplied to the noise coring circuit 2. Since the white noise appears biased to a predetermined frequency component, the noise coring circuit 2 reduces the DCT coefficient corresponding to the white noise. Then, the DCT coefficient from which the component corresponding to white noise has been reduced is input to the inverse orthogonal transform circuit 3, where it is subjected to inverse DCT transform, restored to the original image data, and then output.

On the other hand, the edge detection circuit 11 takes in the input image data and detects whether the image data corresponding to the area to be cored in the noise coring circuit 2 includes an edge. Then, for example, when an edge is included, a logic 1 is output to the noise coring circuit 2, and when an edge is not included, a logic 0 is output to the noise coring circuit 2.

The noise coring circuit 2 reduces the amount of white noise component to be reduced from the orthogonal transform coefficient when the logic 1 is input from the edge detection circuit 11, that is, when the presence of an edge is detected. On the other hand, when the edge detection circuit 11 outputs logic 0, that is, when the edge is not detected, the noise coring circuit 2 outputs the DCT.
A component corresponding to white noise is sufficiently reduced from the coefficient.

Therefore, it is possible to suppress the occurrence of block distortion at the edge portion of the image, and it is possible to sufficiently remove the white noise in the flat portion of the image when there is no edge.

FIG. 2 shows another embodiment of the noise reducer of the present invention. In this embodiment, an edge amount calculation circuit 21 is provided instead of the edge detection circuit 11 in the embodiment of FIG. Edge amount calculation circuit 21
Calculates the edge amount of the input image data, and controls the noise coring circuit 2 and the selector 22 in accordance with the calculation result. The selector 22 selects one of the image data output from the inverse orthogonal transform circuit 3 and the image data not processed by the orthogonal transform circuit 1, the noise coring circuit 2, and the inverse orthogonal transform circuit 3, and outputs the selected image data. It is designed to do.

In other words, in this embodiment, the edge amount calculation circuit 21 detects not only the presence or absence of an edge but also the presence of an edge in multiple stages in the presence of the edge. This edge amount can be calculated, for example, from how much the edge is included in the area to which the coring is applied, or how steep each edge is.

Then, the edge amount calculation circuit 21 outputs the detection result to the noise coring circuit 2. The noise coring circuit 2, when the detection result supplied from the edge amount calculation circuit 21 indicates that the edge amount is large,
The reduction rate of the DCT coefficient is reduced as much as possible, and when the edge amount is small, the reduction rate of the DCT coefficient is increased. Therefore, the noise reduction amount can also be changed in an analog manner in response to the analog change in the edge amount.

Further, the edge amount calculation circuit 21 controls the selector 22 so as to select the output of the inverse orthogonal transform circuit 3 when the edge amount is less than or equal to a preset threshold value. If there is a predetermined threshold value or more, the selector 22 is made to select the input image data itself. This makes it possible to more effectively suppress the occurrence of block distortion when the amount of edges is large.

In the above embodiment, the DCT transform or the inverse DCT transform is used in the orthogonal transform circuit 1 and the inverse orthogonal transform circuit 3, but it is also possible to use Hadamard transform or other orthogonal transform. is there.

[0021]

As described above, according to the noise reducer of the present invention, the edge of the image data is detected and the removing means is controlled in accordance with the detection result, so that the block of the edge portion included in the image is detected. It is possible to suppress the distortion and sufficiently suppress the white noise in the flat portion of the image.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of a noise reducer of the present invention.

FIG. 2 is a block diagram showing the configuration of another embodiment of the noise reducer of the present invention.

[Explanation of symbols]

 1 orthogonal transformation circuit 2 noise coring circuit 3 inverse orthogonal transformation circuit 11 edge detection circuit 21 edge amount calculation circuit 22 selector

Claims (3)

[Claims] 1. An orthogonal transforming unit for orthogonally transforming input image data, a removing unit for removing a noise component from an output of the orthogonal transforming unit, and an inverse orthogonal transforming unit for inverse orthogonal transforming an output of the removing unit. And a control unit that detects an edge of the image data and controls the removing unit in accordance with the detection result. 2. The noise according to claim 1, wherein the control unit reduces the removal rate of the noise component in the removal unit when the number of edges of the image data is large and when the edges are small. Reducer. 3. Selection means for selecting one of image data that has been subjected to inverse orthogonal transformation by the inverse orthogonal transformation means and one of the image data that has not been processed by the orthogonal transformation means, the removing means, and the inverse orthogonal transformation means. Further comprising: when the image data has many edges,
Controlling the selection means, the orthogonal transformation means, the removal means,
The noise reducer according to claim 1 or 2, wherein the image data that is not processed by the inverse orthogonal transforming means is selected.