JPS623380A - Smoothing system for picture - Google Patents

Abstract

PURPOSE:To improve the picture smoothing level by using an (nXn) space filter which designates a color having the highest emerging frequency among n<2> pieces of unit areas as a central unit area. CONSTITUTION:An (nXn) space filter (nXn SPF) consists of (nXn) pieces of unit areas D11-Dnn. These areas D11-Dnn consist of a single color display unit of pictures, e.g., a dot of a single color. The area Dcc serves as the central unit area and c=(n+1)/2 is satisfied. The color having the highest emerging frequency among the color information on n<2> pieces of unit areas is designated to the area Dcc. In such a way, a single (nXn) space filter is used to delete the isolated noise and to smooth pictures and graphic forms. Furthermore, this smoothing action is repeated to increase the number of isolated points deleted and to reduce a large isolated point in size.

Description

[Detailed Description of the Invention] [Summary] In an image smoothing method using an n×n spatial filter (n≧3), the color that appears most frequently among each of n2 unit areas is designated as the central unit area. The image is smoothed using an n×n spatial filter. As a result, it is possible to remove isolated noise and smooth image shapes using one n×n spatial filter, and by repeating smoothing, it is possible to increase the number of removed isolated points and reduce the size of large isolated points. This makes it possible to increase the level of smoothing of image shapes.

[Industrial application field]

The present invention relates to an image smoothing method that uses an n×n (n≧3) spatial filter to remove isolated points such as isolated noise and smooth unevenness on an image figure.

In videotex such as the Captain method, the original image input from a camera or scanner is converted into photographic figure information, mosaic figure information, geometric figure information, etc. to create image information for video typhoid fever. There is.

As a result of these graphical transformations, the accuracy of the Videotex image is often reduced compared to the original input image. This is because in photographic figure information, even if a 4x4 dot mini-block has three or more colors, the amount of image information is compressed by converting it into two colors, and in mosaic figure information, the amount of image information is compressed. Even when using a small mosaic, the image information must be created using a limited number of standard mosaics (153 in the case of the captain method) after converting the 8 x 24 dot block into two colors. This is because the amount of information is compressed.

When image information for videotex is created by such graphic conversion processing, the presence of noise in the input original image causes various inconvenient problems.

For example, in the case of photographic and mosaic graphic information, not only is it necessary to perform unnecessary graphic conversion processing for noise, but even after the graphic conversion process, it becomes noise and further deteriorates the image quality. .

In addition, in the case of geometric figure information, the input original image can be accurately reproduced using vectors, but the noise is also accurately reproduced, so it is possible to use a large amount of information to compensate for the noise. reduce image quality,
Moreover, this has the disadvantageous result of lowering the transmission speed.

Noise occurs not only in isolation on the screen, but also in unnecessary protrusions and unevenness around the image. Therefore, by removing these noises by smoothing, it is possible to smooth the image and obtain a high-quality image that is easy to see, and at the same time, it is possible to compress the amount of information.

Therefore, in order to eliminate the aforementioned inconvenient problems caused by noise, improve image quality, and reduce the amount of information required for image transmission, isolated noise is removed.
It is necessary to smooth the image.

Smoothing in the present invention involves both removing these isolated noises and smoothing the unevenness of image figures. Note that smoothing of the image shape includes not only removing unnecessary protrusions and unevenness caused in the image due to noise, but also smoothing unevenness existing in the image itself.

[Conventional technology]

Figure 5 shows spatial filters for removing isolated noise. Figure 5 (A) shows spatial filter A that removes noise consisting of one isolated point, and Figure 5 (B) shows spatial filter A that removes noise consisting of 4 or fewer isolated points. This is a spatial filter B that removes noise.

Spatial filter A is composed of a square area of 3 x 3 dots, and by rewriting the center dot DC to the color of the surrounding 8 dots, it eliminates the noise of an isolated 1 dot (generally 1
(isolated dots).

Spatial filter B is composed of a 4×4 dot filter, and the four dots Dcl-Dc4 in the center are connected to the surrounding 12 dots.
Dots D,~DI! By rewriting the color to
Isolated noise of 4 dots or less (generally isolated points of 4 dots or less) is removed.

When removing isolated noise, spatial filter A, which removes noise from one isolated point, cannot remove isolated noise consisting of two or more points, so it is necessary to use a spatial filter that corresponds to the size of the isolated point to be removed. , it is necessary to prepare multiple spatial filters.

Furthermore, it is not possible to remove unnecessary protrusions or irregularities formed in an image due to noise or to smooth irregularities in an image using a spatial filter that removes such isolated points. Therefore, conventional methods include a method in which the average level of n×n dots is set as the value of the center dot, and a median filter method in which the center level value when the n×n dots are arranged in level order is set as the value of the center dot. Smoothing was performed using .

[Problem that the invention seeks to solve]

Conventional image smoothing methods require the preparation of multiple spatial filters depending on the size of isolated points to be removed, and when smoothing the image shape, another spatial filter is required. However, there is a problem in that it is difficult to remove isolated points of any size with a single spatial filter and to perform any level of smoothing on the image shape.

The present invention makes it possible to remove isolated points such as isolated noise of any size using a single n×n (n is an odd number of 3 or more) spatial filter, and also to smooth the image shape to any level. The purpose is to provide an image smoothing method.

[Means for solving problems]

The means taken by the present invention to solve the above-mentioned problems in the conventional image smoothing method will be explained with reference to FIG.

FIG. 1 shows the configuration of an n×n spatial filter used in the present invention.

In Fig. 1, H×n SPF is the entire n×n spatial filter, with n×n unit areas (D + I"" D-)
Consists of.

Here, n is an odd number of 3 or more. Unit area (D l
j""D-) is a unit of displaying a single color of an image, and is composed of, for example, one dot of a single color. D Ce is the central unit area, where C=n+1/2.

The n×n5PF of the present invention specifies the color with the highest frequency of appearance among the color information of the n1 unit areas constituting it as the color of the central unit area Dec.

[For production]

The operation of the present invention will be explained with reference to FIGS. 2 and 3, taking the case where n=3 as an example.

FIG. 2 is an explanatory diagram of isolated point removal processing using n×n5pp' according to the present invention, and FIG. 3 is an explanatory diagram of image shape smoothing processing using the same filter.

First, isolated point removal processing such as isolated noise using the n×n5PF of the present invention will be explained with reference to FIG.

Figure 2 (A) shows the n×n5 of the present invention when n=3.
PF (3X3SPF) is shown.

When this 3×3 SPF is used, the isolated point N1 having the size of one unit area e1 shown in FIG. 2(B) can be removed by - times of scanning. Similarly, isolated points N with a size of 64 x 4 unit areas 6 shown in FIG. 2(C) are obtained.
Up to 4 can be removed in - times of scanning. The isolated point N4 is in each unit area e.

Even when ~e4 is arranged in a straight line, it can be removed by - times of scanning. In addition, in the case of the isolated point N4, it is assumed that the surroundings are all the same color.

In the case of a 3x3 isolated point N as shown in Figure 2(D),
If the surroundings are the same color, the first scan will result in el
sea, e, je'l are removed, e2,
e4, es teh, and ea remain.The second scan removes e2, e4, e6, and ell, leaving e5.The third scan removes e, and the isolated point N is completely removed. be done.

In general, when the surroundings are of the same color, by repeating scanning with n×n5PF, it is possible to remove isolated points in an n×n configuration, and it is also possible to reduce the size of isolated points larger than n×n.

If the value of n in n×n5PF is increased beyond 3, large isolated points can be removed or reduced, but
On the other hand, the amount of hardware increases, the processing becomes more complex, and the processing speed decreases. Therefore, considering the size of the isolated point to be removed or reduced, the amount of hardware constituting the nxn5PF, ease of processing, processing speed, etc., the nxn5PF is
Next, the image shape smoothing process of the present invention is performed as follows:
□゛“Explain with reference to Figure 3.

m3゜3゜, 321. ,．． From 9e, from 7□86oj
, , , , or a convex portion Np formed in the shape of the image IM itself is shown. e and ~e are respective unit regions forming the convex portion Np.

When scanning once using the same 3X3 SPF shown in Figure 2 (A), as shown in Figure 3 (B),
1°e is removed, leaving 82 to e8. 2nd time
By 2' scanning, as shown in the same figure (C), e2
and e3.

is removed, leaving e4 to e[l. By the third scan, e is removed, and e4 and e, ~e8 are
.

remain. By the fourth scan, as shown in the same figure (E),
e4 is removed and only e6 to e8 remain.

By increasing the number of times of scanning with 3×3 SPF in this manner, it is possible to increase the level of smoothing of the image shape.

For 3x3SPF, e in Figure 3(E).

~ell cannot be removed. To remove this, 7X7 SPF is required. Generally, (2m+1)×(2m+1) SPF is required to remove m convex portions existing in a row on a long horizontal or vertical straight line image. Therefore, the size of n×n5PF is selected in consideration of the level of smoothing.

The above also applies to the case of smoothing a concave portion in an image figure.

As described above, it is possible to remove or reduce isolated points and smooth the image shape using one n×n5PF, and by repeatedly scanning the screen with this n×n5PF, large isolated points can be removed. It is possible to remove or reduce the image shape and increase the level of smoothing of the image shape.

Conventional isolated point removal spatial filters cannot achieve this kind of effect even if smoothing processing is repeated.

〔Example〕

Embodiments of the present invention will be described with reference to FIGS. 2 to 4.

FIG. 4 shows an example of a specific device for implementing the present invention.

(Description of Configuration) In FIG. 4, 110 is an input unit which A/D converts an original image input from a scanner 111 or a camera 112.

120 is a memory unit, which includes a first image memory (IMSI) 1;
21, 2nd image memory (IMSz) 122, 3rd image memory (1MS3'') 123, 4th image memory (
1MS4) 124 and other memories necessary for image processing. 1M3.121 stores the original image information input from the human resources section 110. IMSt 12
2 stores information about a gradation-converted image obtained by gradation-converting the original image. The 1MS 3123 stores image data that has been subjected to smoothing processing according to the present invention. 1MS4
124 stores images that have been subjected to photographic processing, mosaic processing, or geometric processing.

Reference numeral 130 denotes a gradation conversion unit that performs gradation conversion on the input original image. A photographic conversion unit 150 performs photographic processing on the smoothed image. 16G is a mosaic conversion unit that performs mosaic processing on the smoothed image. 17
0 is a geometry conversion unit that performs geometry conversion processing on the smoothed image. The configuration and operation of each of these converters are well known.

180 is an output unit that transmits the final output image to a terminal or subscriber (not shown). 181 is a floppy disk.

190 is a display unit that displays an input original image, a gradation-converted image, an image subjected to figure conversion processing such as photography processing,
The smoothed image and the like are displayed as appropriate. 191
is a display.

140 is a smoothing processing means that performs the smoothing processing of the present invention. Reference numeral 14-1 denotes an n.times.n cutting section, which cuts out each unit region of the gradation-converted image into a square region (see FIG. 1) composed of n.times.n pieces, using it as a central region Dec. Reference numeral 142 denotes a color information detection unit that detects color information for each of n×n unit areas in a square area, and stores and stores the number of each color in memory. 143 is an area color specification part;
The color with the maximum number of colors, that is, the color with the highest frequency of appearance, is detected from among the pieces of color information detected by the color information detection unit 142, and is designated as the color of the central area Dec.

(Description of Operation) The operation shown in FIG. 4 will be explained with reference to FIGS. 2 and 3.

In FIG. 4, an original image input from a scanner 111, a camera 112, etc. is A/D converted in the input section, and then the IM
Stored in SI 121.

The gradation conversion unit 130 reads out the original image information stored in the IMS 121, performs gradation conversion using, for example, 16 colors, and then stores it in the IMSI 122.

n×n cutting section 1 in smoothing processing means 140
41 cuts out a square area composed of n×n unit areas from the tone-converted image information stored in the IMSz 122. The color information detection unit 142 detects n×
Color information for each of the n unit areas is detected, and the number of each color is stored in memory. The area color determining unit 143 detects the color with the highest frequency of appearance among each number of lustrous axes detected by the color information detecting unit 142, and selects it as the central area Dec of the square area (see FIG. 1 or 2). (see figure (A)) and stores it in the unit area of the address of the central area Dec in the IMS 3123. In the above manner, H×n
Filter processing by SPF is performed.

When the above-mentioned filtering process is completed for one unit area,
A new n×n square area is cut out from the IMSz122 centering on the unit area of the next address of the central area D ec171, and the above-described filtering process is performed on the new central area D e c.

When the above-described filter processing is executed for all unit areas in the IMSz 122, the first smoothing process is completed, and the image subjected to the first smoothing process is stored in the IMS 3.

For example, if n=3, isolated point removal processing as explained in FIG. 2 and image figure smoothing processing as explained in FIG. 3 are performed.

By repeating the smoothing process on the image that has been subjected to the smoothing process described above, the number of isolated points removed increases, and the level of smoothing of the image figure can be raised.

When the smoothing process is completed, the smoothed image stored in the 1MS 3123 is subjected to graphic conversion processing in one of the photographic conversion unit 150, mosaic conversion unit 160, and geometric conversion unit 170, and then converted into a 1MS image.
4124.

The image subjected to the figure conversion process is outputted to the output unit 18 as an output image.
0 and transmitted to a terminal device or subscriber (not shown).

The display unit 190 appropriately displays image states during each process such as gradation conversion processing, smoothing processing, figure conversion processing, etc., and observes each processing state. For example, in the case of smoothing processing, the smoothing state is observed, and the number of times of smoothing processing is determined, and in some cases, the smoothing processing is performed on a specific area.

Although an example of the apparatus for implementing the present invention has been described above, the apparatus for implementing the present invention is not limited to this. For example, IMS 121 to IMS 4124 can be shared with each other. Furthermore, the images to be subjected to the smoothing process of the present invention naturally include general images as well as images for videotex.

〔effect〕

As explained above, according to the smoothing process using the n×n5PF of the present invention, the following effects can be obtained.

(b) One spatial filter can remove isolated points and smooth the image shape.

(b) By repeating the smoothing process, large isolated points can be removed or reduced, and the level of smoothing of image figures can be increased.

[Brief explanation of drawings]

Fig. 1: An explanatory diagram of the spatial filter of the present invention, Fig. 2:
. . . An explanatory diagram of an example of isolated point removal processing using the spatial filter of the present invention. FIG. 3 . . . An explanatory diagram of an example of image figure smoothing processing using the spatial filter of the invention. Explanatory diagram of the implementation device, Figure 5...
FIG. 2 is an explanatory diagram of a conventional isolated point removal spatial filter. In Figures 1 to 3, n x n S P F-n x n spatial filter, D
, ~Dlll, DI~Dst6+~e8...Unit area, DeC...Central area, N+, N4, N
q: Isolated point, N9: Convex portion on the image figure.

Claims (1)

[Claims] In an image smoothing method using an n×n spatial filter consisting of a square area made up of n×n (n is an odd number of 3 or more) unit areas, the n^ that makes up the n×n square area is
The color with the highest frequency of appearance in each color information of the two unit areas (D_1_1 to D_n_n) is selected from the central unit area (D_c
n×n spatial filter (n×n
An image smoothing method characterized by performing smoothing using SPF.