JPH0256179A - Picture noise removal processing system - Google Patents

Abstract

PURPOSE:To remove noises generated in a flat part around an edge without blurring the edge by detecting the edge part in a decoding picture element and filtering selectively picture elements around the edge. CONSTITUTION:A filtering picture element deciding circuit 4 decides whether a certain picture element (a) becomes the object of the filtering or not from edge information (e) which is the output of an edge detecting circuit 2 and a Laplacian which is the output of a Laplacian calculating circuit 3. That is to say, the positional relation between the edge part and the picture element (a) is obtained, whether or not the picture element (a) is influenced by the noises generated in the edge part is checked and whether or not the picture element (a) is the picture element positioned around the edge and in the flat part is decided. A filter 5 selectively filters the picture elements around the edge decided by the above deciding circuit 4.

Description

[Detailed Description of the Invention] [Summary] In an image encoding method using orthogonal transform encoding or vector quantization, noise generated near the edge regions of two images has a large amplitude and is difficult to notice for the viewer of the reproduced image. give an unnatural impression. The present invention removes noise generated near such edge regions and obtains high-quality images.

In an image encoding method using orthogonal transform encoding or vector quantization, one image is divided into blocks, and encoding is performed in units of blocks. In this case, distortion caused by encoding will be spread throughout the block. Furthermore, when encoding temporally sudden and sharp edges, large encoding noise is generated. In the case of a person in a flat background, there is a sharp edge near the boundary between the background and the person.

In a block including such a steep edge and a flat area, the noise generated in connection with the edge part also occurs in the flat part around the edge part. The noise that occurs on flat areas is
This was particularly noticeable, giving an unnatural impression to the viewer of the reproduced image.

In the present invention, in order to remove such encoding noise that occurs in flat areas around edges, edge areas are first detected in a decoded image, and then selective filtering is performed on pixels around the edges. . However, if the area around the edge is not a flat area.

Since filtering may actually cause image blurring, such problems are avoided by adaptively selecting pixels to be filtered.

[Industrial application field]

The present invention relates to a highly efficient image noise removal processing method.

[Conventional technology]

BACKGROUND ART Conventionally, by dividing one image into blocks, a method having high coding efficiency such as orthogonal transform coding or vector quantization has been used, and thereby image coding efficiency has been improved.

[Problem that the invention seeks to solve]

However, since one image is encoded by dividing it into blocks, a large amount of encoding noise occurs around the edges of the reproduced image, giving an unnatural impression to the viewer.

Means for Solving Problem c] In the present invention, in order to remove coding noise generated around such edges, first, edge portions in the decoded image are detected, and pixels around the edges are .

Perform selective filtering. However, if the area around the edge is not a flat area, filtering may actually blur the image, so this problem can be avoided by adaptively selecting pixels for filtering. . That is, in order to select pixels to be filtered, edge portions of the input image are detected.

Let e be information representing the edge portion. Next, pixels that are around the edge portion and are in the flat portion are targeted for filtering. For this, the Laplacian 2 of the input image is
Calculate. Whether or not a certain pixel a is to be filtered is determined as follows. First, it is determined whether or not pixel a is located at a position where it is affected by noise generated at the edge.

Using information e, the positional relationship between the edge portion and pixel a is determined. When pixel a is in a position affected by noise generated at the edge, pixels around one pixel a.

(i) Let b be a pixel where the value of Laplacian 2 is less than or equal to a certain threshold Tll and the absolute value of the difference between pixel a and pixel value is less than or equal to a certain threshold Th, and the number of 2 pixels is n.

(+1) and when n is greater than a certain threshold Thn, one pixel a
is determined to be a pixel located around the edge and in a flat area, and filtering is performed using pixel 1 and pixel A.

(Function) By applying the present invention, it is possible to remove noise generated in flat areas around edges without blurring the edges.

〔Example〕

FIG. 1 shows the configuration of an embodiment of the present invention. In the figure, 1 is an input terminal, 2 is an edge detection circuit, 3 is a Laplacian calculation circuit,
4 is a filter for the filtering pixel determination circuit 5, 6 is a switch, and 7 is an output terminal.

An input image is input from input terminal 1. The filtering pixel determination circuit 4 determines whether or not a certain pixel a is to be filtered, based on the edge information e output from the edge detection circuit 2 and the Laplacian output from the Laplacian calculation circuit 3, as follows.

First, in order to determine whether one pixel a is located at a position affected by noise generated at the edge, the positional relationship between the edge and pixel a is determined using information e. When pixel a is located at a position where it is affected by noise generated at the edge.

(1) For pixels around pixel a, the value of Laplacian l is less than or equal to a certain threshold value The, and the absolute value S of the difference between pixel a and the pixel value is a threshold (a pixel whose value is less than or equal to aTh is defined as b, and 2 pixels are Let the number be n.

(it) Then, when n is greater than or equal to a certain threshold Thn, pixel a is determined to be a pixel located around the edge and in a flat area,
Generates a filtering pixel instruction signal.

Further, three pixels a and pixel S are outputted to the filter 5 as filtering input pixels. The filter 5 performs filtering using one pixel a and one pixel S. The switch 6 switches between the output of the filter 5 and the seven images based on the filtering pixel instruction signal number, and outputs the output image from the third output terminal 7.

FIG. 2 shows a flowchart of the main part of the present invention.

In process (2), as described above, a certain pixel a is located at a position where it is affected by the noise generated at the edge portion, but it is nicked. Process (2) checks whether the value of the Laplacian l is less than or equal to the threshold value The and the absolute value S is less than or equal to a certain threshold value Th, as described above. Then, in process (2), as described above, it is determined whether or not pixel a is a pixel located in a flat area around an edge.

Figure 3 shows an image transmission system to which the present invention is applied.
Reference numeral 101 in the figure corresponds to a conventionally known image transmission system, and 102 represents an image noise removal circuit provided in the present invention.

〔Effect of the invention〕

As explained above, according to the present invention, it is possible to remove the noise generated in the flat area around the edge without blurring the edge, and it is possible to obtain a high quality image.

It goes without saying that by using the present invention, it is possible to improve the image quality even in a moving image encoding processing method using an inter-frame differential image encoding processing method.

[Brief explanation of the drawing]

FIG. 1 shows the configuration of an embodiment of the present invention, FIG. 2 is a flowchart of the main part of the present invention, and FIG. 3 shows an image transmission system to which the present invention is applied. 1 is an input terminal, 2 is an edge detection circuit, 3 is a love run calculation circuit, 4 is a filtering pixel determination circuit, 5 is a filter, 6 is a switch, and 7 is an output terminal. Patent applicant Nippon Telegraph and Telephone Corporation

Claims (1)

[Scope of Claims] (1) In an image encoding processing method that divides an image into blocks and encodes each block, an edge portion included in the image is detected, and a flat area near the edge is detected. An image noise removal processing method characterized in that filtering is performed on pixels located in a certain area. (2) The image noise removal processing method according to claim (1), wherein a Laplacian of the image is calculated and filtering is performed using only pixels whose Laplacian is smaller than a certain threshold value. (3) The method is characterized in that the absolute value of the difference between the pixel value of the pixel to be filtered and the pixel values surrounding the pixel used for filtering is calculated, and filtering is performed using only pixels whose value is smaller than a certain threshold value. An image noise removal processing method according to claim (1) or (2). (4) The image noise removal processing method according to claim (2) or (3), wherein filtering is performed only when the number of pixels surrounding the pixel used for filtering is greater than a certain threshold value.