JP2001238209A - Time space and control method, time spatial filter, and storage medium recording time space band limit program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a time spatial filter that increase an information amount whose band is limited even from a video image including multiple edge pixels at the outside of a target area. SOLUTION: An edge detection section 103 detects an edge of a received image and an edge information storage section 104 stores edge information. An edge density measurement section 105 measures the edge density of an edge pixel and an edge target area discrimination section 105 determines whether or not the edge pixel is in a target area. When the edge pixel is in the target area, a filter control section 107 sets a frequency characteristic of a filter F1 to a characteristic C1. When the edge pixel is in a non-target area, the filter control section 107 sets the frequency characteristic of the filter F1 to a characteristic C2. A filter selection section 108 selects the filter F1 when a pixel receiving filtering is an edge pixel. The filter selection section 108 selects a filter F2 when a pixel receiving filtering is not an edge pixel.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a band limiting method and a filter device for a moving image signal.

[0002]

2. Description of the Related Art In general, since high-frequency components of an image have little effect on vision, even if high-frequency components are deleted, visual deterioration is small. For this reason, in coding moving images, it has been practiced to improve the coding efficiency by applying a filter to the input image in advance before coding to limit the band.

[0003] Filters that limit the band include a spatial filter that uses correlation between pixels in a screen and a time filter that uses correlation between pixels between frames.

One of the spatial filters is a moving average filter. In this filter, as shown in FIG. 9, a square area of N pixels vertically and N pixels horizontally (N taps) is set around the pixel to be obtained, and the image information of the pixels included in the square is expressed by the formula (1). ) Is performed to obtain image information of the pixel. By performing this process on each pixel in the screen, a moving average filter is applied to the entire screen. The frequency characteristic of this filter is characterized by the number of taps N and the weighting coefficient W, and the visual characteristic after band limitation depends on these values. Generally the number of taps
As N is larger, the average of a wider range of pixels is averaged, so the difference between adjacent pixels is reduced, and higher frequency components are deleted.

[0005]

(Equation 1) Further, an edge in which image information between adjacent pixels greatly changes is generally considered to be a region of interest because it gives remarkable visual information. Therefore, there is an edge-preserving filter that limits the band while preserving the edges in order to maintain the visual quality of the image. In addition, a method has been proposed in which filters having different frequency characteristics are applied to edge pixels and other pixels. In such a filter, band limitation is performed to reduce high-frequency components of pixels other than edges without reducing high-frequency components of edge pixels.

[0006] A variety of edge pixel detection methods are known. For example, Tsukahara, Hasegawa, Kitajima, "Removal of Additive White Noise by Adaptive Wiener Filter Considering Image Edge Preservation", IEICE. Theory D-II, vol. J81-DI
I, No. 4, pp. 623-631, 1998. and Yagi et al., "Practical image processing learned in C language", pp. 44-47, Ohmsha, 1993.

[0007]

However, there are cases where edge pixels are not always considered as a region of interest. Edges provide noticeable visual information, but edge pixels in regions that are not necessarily required in the image content may not be considered as regions of interest. In other words, in a method in which all edge pixels are considered as a region of interest and high-frequency components are not reduced,
If there are many edge pixels even in an area that is not the attention area (non-interest area), the amount of information reduced by the band limitation is small, and the effect of the band limitation is reduced. For example, an edge pixel of a stadium spectator seat in a video relay of a game such as tennis or soccer may not be the attention area. However, if the band is limited so as not to reduce the high-frequency components of the edge pixels even in the non-interest area, the number of edge pixels is large at the audience seat, and the high-frequency components are not reduced. Therefore, in an image in which the audience seats occupy a large area on the screen, the amount of band-limited information is small, and the effect of the band limitation is reduced.

An object of the present invention is to provide a large amount of band-limited information even for an image having many edge pixels outside the region of interest.
Therefore, a space-time band control method with a high band-limiting effect,
It is an object of the present invention to provide a spatio-temporal filter device and a recording medium recording a spatio-temporal band limiting program.

[0009]

In order to achieve the above object, according to the present invention, it is determined whether or not an edge pixel is a pixel in a region of interest, and the frequency characteristic of a filter applied to the pixel is changed. That is, when reducing the high-frequency component of the pixel of the non-interest area, it is determined whether or not the edge pixel is the area of interest, and the high-frequency component is not reduced at the edge pixel of the area of interest. Reduce high frequency components. For example, a spectator seat in a live broadcast of a soccer match is a non-attention area, so that high-frequency components are reduced for edge pixels of the spectator seat, and high-frequency components are not reduced because edge pixels indicating a player or a ball are attention areas. .

As a method of determining whether or not an edge pixel in a screen is a region of interest, there are a method of manually performing the method and a method of automatically performing the method based on information obtained from an image. As a method of automatically judging from information obtained from images,
There is a method in which edge pixels in a screen are detected in advance, the density of edge pixels around each pixel is measured, and judgment is made based on the magnitude of the density. For example, a spectator seat of a live game such as a soccer game is considered to have a higher density of edge pixels than a player. Therefore, in this method, a pixel having a higher density of surrounding edge pixels is determined as a non-attention area, and a lower density is determined. The pixel is determined to be a region of interest. Specifically, after detecting an edge pixel in the screen, a square of N pixels horizontally by N pixels centered on the pixel is set for each pixel, and the number of edge pixels included in the square is measured. I do. If the number of edge pixels is, for example, half or more of N × N, the density of peripheral edge pixels is high, so that the region is determined to be a non-attention area, and the high-frequency component of the square central edge pixel is reduced. If the number of edge pixels is less than half, for example, the density of peripheral edge pixels is low, so it is determined to be a region of interest, and high-frequency components are not reduced.

It is also preferable to set not only one type of edge pixel density but also a plurality of ranges, and determine from the edge pixel density of each range whether or not the pixel is a non-interest area. This method is effective when only a region where a pixel having a large edge pixel density exists in a wide range is set as a non-target region. In other words, by measuring the density in the narrow range of the edge and the density in the wide range, it can be used to exclude the pixel of the attention area from the pixels having the high edge density.

After determining the region of interest from the measurement of the edge pixel density, the density of the non-target region pixels around the pixels of the non-target region is measured, and the pixel is excluded from the non-target region from the density. A method of determining whether or not to do so is also suitable. This method excludes a pixel erroneously determined as a non-interest area from the measurement of the edge pixel density.

As described above, according to the present invention, by automatically judging a region of interest by measuring the edge pixel density, the edge pixels of the non-region of interest are reduced in high frequency components, and the edge of the region of interest is reduced. The pixel can change the frequency characteristic of the filter so as not to reduce the high frequency component. Therefore, the amount of information band-limited to a video having many edge pixels in the non-interest area increases,
The effect of the band limitation increases.

[0014]

Next, embodiments of the present invention will be described with reference to the drawings. First Embodiment Referring to FIG. 1, a spatio-temporal filter device according to a first embodiment of the present invention includes an image input unit 101 and a filter processing unit 102.
, An edge detection unit 103, an edge information storage unit 104, an edge density measurement unit 105, an edge attention determination unit 106, a filter control unit 107, a filter selection unit 108, and an image output unit 1.
09.

In this embodiment, the edge density around the pixel is set to 1
The measurement is performed for two ranges, and it is determined whether or not the pixel is the attention area based on the magnitude of the edge density, and the frequency characteristic of the filter applied to the edge pixel is changed. The measurement of the edge density is performed on the range of the square. If the ratio of the edge pixels is less than half, the region is determined to be the region of interest.

The image input unit 101 inputs an image. The edge detection unit 103 detects an edge of the input image.
The edge information storage unit 104 stores the detected edge information. The edge density measuring unit 105 sets, for each edge pixel, a square-shaped range of three pixels vertically and horizontally around the pixel, and measures the number of edge pixels included in the range. The edge attention determination unit 106 checks whether the number of edge pixels measured by the edge density measurement unit 105 is half or more (ie, 5 or more) of the pixels in the range. The pixel is determined to be a pixel. If the number of pixels is half or more, the pixel is determined to be a pixel of the non-interest area. The three vertical and horizontal pixels shown here are merely examples, and it is also possible to set a square of any other size. In the filter selection unit 108, a filter F1 for an edge pixel and a filter F2 for a pixel other than an edge are set. Based on information on the presence or absence of an edge from the edge information storage unit 104, one of the two is selected. Select a filter. The filter control unit 107 controls the frequency characteristics of the filter F1 for edge pixels in the filter selection unit 108 according to whether the edge pixel is a region of interest. If the edge pixel is a region of interest, the characteristic C1 is set to F1, and if the edge pixel is a region of non-interest, a characteristic C2 that removes higher frequency components than C1.
To F1.

A filter F1 for an edge pixel, wherein a characteristic C1 corresponding to a region of interest and a characteristic C1 corresponding to a non-region of interest
Tables 1 and 2 show examples of the coefficient W (i, j) in the equation (1) of the filter having 2 respectively, and Table 3 shows an example of the filter F2 for pixels other than edges. Note that the filters shown here are merely examples, and different filters may be used. For example, if a median filter is used as F1 set to the characteristic C1, the edge can be stored as it is.

[0018]

[Table 1]

[0019]

[Table 2]

[0020]

[Table 3] The procedure for selecting a filter to be applied to each pixel is described below with reference to FIG.
This will be described with reference to the flowchart shown in FIG.

1. The edge detection unit 103 detects an edge of an input image and stores edge information in an edge information storage unit 104 (step 201).

2. In the case of an edge pixel, the edge density is measured by the edge density measurement unit 105, and the edge attention determination unit 106 determines whether or not the edge pixel is a region of interest (steps 202 and 203).

3. When the edge pixel is a region of interest,
The filter control unit 107 sets the frequency characteristic of the filter F1 to C1.
(Step 204).

4. If the edge pixel is a non-interest area, the filter control unit 107 sets the frequency characteristic of the filter F1 to C2 (step 205).

5. If the pixel to which the filter is applied is an edge pixel, the filter selection unit 108 selects the filter F1 (step 206).

6. If the pixel to which the filter is applied is not an edge pixel, the filter selection unit 108
2 is selected (step 207).

The filter processing unit 102 has been described above.
According to the selection of the filter by the filter selection unit 108 and the selection of the frequency characteristic of the filter F1 by the filter control unit 107, the input image is filtered.

By measuring the density of edge pixels in this manner, a pixel having a high edge pixel density, such as a stadium spectator image in a soccer game relay video, can be determined to be a non-attention area. Therefore, it is possible to reduce the high-frequency components of the edge pixels in the non-interest area. Second Embodiment Referring to FIG. 3, a spatiotemporal filter device according to a second embodiment of the present invention includes an image input unit 101 and a filter processing unit 102.
An edge detection unit 103, an edge information storage unit 104, an edge attention determination unit 106 ', a filter control unit 107, a filter selection unit 108, a first edge density measurement unit 110, and a second edge density measurement unit 111.

This embodiment differs from the first embodiment in that
The edge density around the pixel is measured for two ranges, and it is determined from the two types of edge densities whether the pixel is a target area.

The measurement of the edge density is performed for two types of squares having different sizes. If the ratio of the edge pixels is more than half in both ranges, it is determined that the region is the non-interest area. Is determined to be the attention area.

For example, a square range (A) of three vertical pixels and three horizontal pixels and a square range (B) of seven larger vertical pixels and seven horizontal pixels are defined, and the edge included in each range is defined. Measure pixels. As shown in FIG. 6A, when the range (A) has eight edge pixels and more than half, but in the range (B), there are eight edge pixels and less than half, the vicinity of the pixel It is considered that there is no pixel having a large edge pixel density. Therefore, this pixel can be excluded from the non-interest area. Conversely, as shown in FIG. 6B, when the range (A) has three edge pixels and less than half, but the range (B) has 28 edge pixels and is half or more, A pixel having a high edge pixel density exists near the pixel, and the pixel is considered to have a low edge pixel density. Therefore, this pixel can be excluded from the non-interest area.

The first edge density measuring section 110 sets a square-shaped range of three pixels vertically and horizontally around each pixel, and measures the number of edge pixels included in the range. Also, the second edge density measuring unit 111
Is the vertical and horizontal 7 for each edge pixel
A square-shaped range of pixels is set, and the number of edge pixels included in the range is measured. The three pixels and seven pixels in the vertical and horizontal directions shown here are examples, and can be set to other ranges.

The edge attention judging unit 106 'checks whether or not the number of edge pixels measured by the first edge density measuring unit 110 is half or more (ie, 5 or more) of the pixels in the range. Further, the second edge density measuring unit 111
It is checked whether or not the edge pixels measured in (1) are half or more of the pixels in the range (that is, 25 or more). If the number of pixels measured by the first edge density measurement unit 110 is equal to or greater than half of the pixels in the range, and the number of pixels measured by the second edge density measurement unit 111 is equal to or greater than half of the pixels in the range, The pixel is determined to be a pixel of the non-interest area, and in other cases, the pixel is determined to be a pixel of the attention area.

The method of selecting the filter of each pixel and the method of changing the frequency characteristic of the filter of the edge pixel are the same as in the first embodiment. The filter processing unit 102 is a filter selection unit 1
Selection of filter at 08 and filter control unit 107
Filter processing is performed on the input image in accordance with the selection of the frequency characteristic of the filter F1 in.

As in the present embodiment, by measuring the density of edge pixels in two types of ranges, an area where pixels with a high edge pixel density exist widely, such as a stadium spectator image in a soccer game relay video. Is detected, and it can be determined as a non-attention area. Therefore,
High-frequency components of edge pixels in the non-interest area can be reduced. Third Embodiment Referring to FIG. 4, a filter device according to a third embodiment of the present invention includes an image input unit 101, a filter processing unit 102, an edge detection unit 103, an edge information storage unit 104, a filter control unit 107, and a filter. It comprises a selection unit 108, a first edge density measurement unit 110, a second edge density measurement unit 111, an edge attention determination unit 106, an attention pixel measurement unit 112, and an attention pixel correction unit 113.

In the present embodiment, in the second embodiment, for a pixel determined to be a non-attention region from the edge pixel density, the density of a non-attention region pixel around the pixel is erroneously measured. Pixels determined to be regions can be excluded. The measurement of the pixel density of the non-interest area is performed for a square range. When the ratio of pixels in the non-interest area is more than half, it is determined to be the non-interest area, and when less than half, it is determined to be the attention area.

For example, as shown in FIG. 7, if there is a pixel (P) which is erroneously determined to be a non-interest area from the measurement of the edge pixel density, the pixels of the non-interest area around the pixel (P) It is considered that the density is small. Therefore, the density of the pixels in the non-interest area around the pixel (P) is measured, and since the density is low, the pixel (P) can be determined to be the area of interest.

The operations of the first edge density measuring section 110, the second edge density measuring section 111, and the edge attention judging section 106 are the same as those in the second embodiment.

The pixel-of-interest measurement unit 112 sets a square-shaped range of three pixels vertically and horizontally around the pixel determined as each non-interest area, and sets the non-interest area included in the range. Is measured. Attention pixel correction unit 1
13 checks whether the number of pixels of the non-target area measured by the target pixel measuring unit 112 is equal to or greater than half of the pixels in the range (ie, equal to or greater than 5). When the number of pixels is half or more, the pixel is determined to be a pixel of the non-interest area. The three vertical and horizontal pixels shown here are merely examples, and it is also possible to set a square of any other size.

The filter control unit 107 controls the frequency characteristic of the filter F1 for the edge pixel in the filter selection unit 108 according to the judgment by the edge attention judgment unit 106 or the attention pixel correction unit 113 whether the edge pixel is the attention area. I do. If the edge pixel is the region of interest, set F1 to the characteristic C1, and if the edge pixel is the region of non-interest, C1
F1 is set for the characteristic C2 for removing high frequency components from 1.

The procedure for selecting a filter to be applied to each pixel will be described below with reference to the flowchart shown in FIG.

1. The edge detector 103 detects an edge of the input image and stores the edge information in the edge information storage 104 (step 201).

2. In the case of an edge pixel, the first and second
The edge density is measured by the edge density measuring units 110 and 111, and it is judged by the edge attention judging unit 106 'whether or not the edge pixel is the attention area (step 203).

3. If the edge pixel is determined to be a non-interest area, the pixel density of the non-interest area is measured by the attention pixel measuring unit 112, and whether or not the pixel is really a non-interest area by the attention pixel correction unit 113 to decide. (Step 208).

4. When the edge pixel is the region of interest,
The filter control unit 107 sets the frequency characteristic of the filter F1 to C1.
(Step 204).

5. If the edge pixel is a non-interest area, the filter control unit 107 sets the frequency characteristic of the filter F1 to C2 (step 205).

6. If the pixel to which the filter is applied is an edge pixel, the filter selector 108 selects the filter F1 (step 206).

7. If the pixel to which the filter is applied is not an edge pixel, the filter selection unit 108
2 is selected (step 207).

The filter processing unit 102 is a filter selection unit 1
Selection of filter at 08 and filter control unit 107
Filter processing is performed on the input image in accordance with the selection of the frequency characteristic of the filter F1 in.

As in the present embodiment, for a pixel once erroneously determined as a non-interest area from the measurement of the edge density, the pixel density of the non-interest area is measured to exclude the pixel from the non-interest area. can do. Therefore,
Pixels in the attention area such as the boundary between the stadium spectator image and the player image in the soccer game relay video can be excluded from the non-interest area.

In the present embodiment, the number of the edge density measuring units is two, but may be one as shown in FIG.

FIG. 8 is a block diagram in the case where the above embodiment is implemented by a computer such as a personal computer.

The input device 301 corresponds to the image input unit 101 and is an input device such as a scanner or a camera for inputting an image. The storage device 302 corresponds to the edge information storage unit 104, and stores edge information. The storage device 303 is a hard disk. The output device 304 corresponds to the image output unit 109, and is a display that outputs a filtered image. The storage device 303 is a hard disk.
The recording medium 305 corresponds to the image input unit 101, the edge information storage unit 104, and the image output unit 10 in the embodiment of FIGS.
This is a recording medium such as a floppy disk, CD-ROM, or magneto-optical disk on which a band limiting program composed of processing of each part except for the part 9 is recorded. The data processing device 306 includes a CPU, various interfaces, etc., reads the band limiting program from the recording medium 305, stores the program in the storage device 303, and executes the program.

[0054]

As described above, according to the present invention,
The frequency characteristics of the filter are such that edge pixels are distinguished into pixels in the region of interest and pixels other than the region of interest, and edge pixels that are not the region of interest reduce high-frequency components, and edge pixels in the region of interest do not reduce high-frequency components Is changed, the amount of information to be band-limited is increased even for an image having a large number of edge pixels outside the region of interest, and the effect of the band restriction is enhanced.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a spatiotemporal filter device according to a first embodiment of the present invention.

FIG. 2 is a flowchart of a filter determination process according to the first embodiment.

FIG. 3 is a configuration diagram of a spatiotemporal filter device according to a second embodiment of the present invention.

FIG. 4 is a configuration diagram of a spatiotemporal filter device according to a third embodiment of the present invention.

FIG. 5 is a flowchart of a filter determination process according to a third embodiment.

FIG. 6 is a diagram illustrating an example of edge pixel density measurement.

FIG. 7 is a diagram illustrating an example in which pixels with a high edge density are excluded.

FIG. 8 is a configuration diagram of a spatiotemporal filter device according to a fourth embodiment of the present invention.

FIG. 9 is a diagram illustrating an example of a spatial filter.

[Explanation of symbols]

 Reference Signs List 101 Image input unit 102 Filter processing unit 103 Edge detection unit 104 Edge information storage unit 105 Edge density measurement unit 106, 106 'Edge attention determination unit 107 Filter control unit 108 Filter selection unit 109 Image output unit 110 First edge density measurement unit 111 Second edge density measuring unit 112 Target pixel measuring unit 113 Target pixel correcting unit 201 to 208 Step 301 Input device 302, 303 Storage device 304 Output device 305 Recording medium 306 Data processing device

 ──────────────────────────────────────────────────続 き Continued on front page F term (reference) 5C021 PA32 PA38 PA58 PA62 PA66 PA67 PA72 RA02 RB08 YA01 5C059 KK00 LA00 SS20 TA69 TB08 TC02 TC10 TC42 TD08 UA02 UA12 UA17 5C082 AA01 AA27 AA37 BA41 CA21 CA10 CA85 DA87 MM

Claims (12)

[Claims] 1. A spatio-temporal band limiting method for limiting a band of an image, comprising: an edge detecting step of detecting an edge of an input image; an edge information storing step of storing edge information of each pixel; An edge density measuring step of measuring the density of the edge pixel in the above; an edge attention determining step of determining whether the edge pixel is a pixel of the attention area from the edge density; and a plurality of filters having different frequency characteristics set in advance. A filter selection step of selecting a filter for an edge pixel for an edge pixel and a filter for a non-edge pixel for a pixel that is not an edge pixel, and a filter selection step in accordance with the determination of whether or not the edge pixel is a region of interest. A filter control step for controlling the frequency characteristics of the set edge pixel filter; Band limiting process and a filtering processing step of performing a filtering process on the image. 2. A spatio-temporal band limiting method for limiting an image band, comprising: an edge detecting step of detecting an edge of an input image; an edge information storing step of storing edge information of each pixel; A plurality of edge density measurement steps for measuring the density of edge pixels in different ranges, and an edge attention determination step for determining whether the edge pixel is a pixel of the attention area based on the edge densities of the plurality of ranges. A filter selecting step of selecting an edge pixel filter for an edge pixel from among a plurality of filters having different frequency characteristics, and selecting a non-edge pixel filter for a pixel that is not an edge pixel; In the filter selection step according to the determination of whether the edge pixel is a region of interest in A band limiting method comprising: a filter control step of controlling a frequency characteristic of an edge pixel filter; and a filter processing step of performing filter processing on an input image. 3. A spatio-temporal band limiting method for limiting an image band, comprising: an edge detecting step of detecting an edge of an input image; an edge information storing step of storing edge information of each pixel; An edge density measuring step of measuring the density of the edge pixels in, an edge attention determining step of determining whether the edge pixel is a pixel of the attention area based on the edge density, and a pixel determined to be not the attention area by the edge attention determination step. On the other hand, a pixel-of-interest measurement step of measuring the density of a pixel determined not to be the region of interest in the vicinity of the pixel, and determining that the pixel determined not to be the region of interest is not the region of interest in the vicinity A pixel-of-interest correction step of judging whether or not to re-classify the pixel into a region of interest based on the density of the determined pixel; A filter selection step of selecting an edge pixel filter for an edge pixel from among a plurality of filters having different frequency characteristics, and selecting a non-edge pixel filter for a pixel that is not an edge pixel. The frequency of the edge pixel filter preset in the filter selection step is determined according to the determination of whether the edge pixel is a region of interest in the step or the determination of whether the pixel is reclassified to the region of interest in the pixel of interest correction step. A band limiting method including a filter control step of controlling characteristics and a filter processing step of performing filter processing on an input image. 4. A spatio-temporal band limiting method for limiting a band of an image, comprising: an edge detecting step of detecting an edge of an input image; an edge information storing step of storing edge information of each pixel; An edge density measuring step of measuring the density of the edge pixels in different ranges, an edge attention determining step of determining whether the edge pixels are pixels of the attention area from the edge densities of the plurality of ranges, A pixel-of-interest correction step of determining whether to re-classify the pixel into the area of interest based on the density of the pixels determined not to be the area of interest in the vicinity of the determined pixel; An edge pixel filter is selected as an edge pixel from among a plurality of filters having different characteristics, and a non-edge pixel is selected for a pixel which is not an edge pixel. A filter selecting step of selecting a di-pixel filter; and determining whether an edge pixel is a target pixel in the edge target determining step, or determining whether to re-classify a pixel into a target area in the target pixel correcting step. Accordingly, a band limiting method including a filter control step of controlling a frequency characteristic of an edge pixel filter preset in a filter selection step, and a filter processing step of performing filter processing on an input image. 5. A spatio-temporal filter device for limiting a band of an image, comprising: an edge detection unit for detecting an edge of an input image; an edge information storage unit for storing edge information of each detected pixel; An edge density measurement unit that measures the density of edge pixels around the edge of the image, an edge attention determination unit that determines whether the edge pixel is a pixel of the attention area based on the edge density, and a plurality of filters having different preset frequency characteristics. From among them, a filter selection unit that selects an edge pixel filter for an edge pixel, and a non-edge pixel filter for a pixel that is not an edge pixel, and the edge attention determination unit determines whether the edge pixel is a region of interest. According to the determination, a filter control unit that controls a frequency characteristic of the edge pixel filter preset by the filter selection unit; Spatial filter apparatus when having a filter processing unit that performs filter processing on the image. 6. A spatio-temporal filter device for limiting a band of an image, comprising: an edge detection unit for detecting an edge of an input image; an edge information storage unit for storing edge information of each detected pixel; A plurality of edge density measurement units that measure the density of edge pixels in the vicinity of a different range, and an edge attention determination unit that determines whether an edge pixel is a pixel of a region of interest based on the edge densities of the plurality of ranges. A filter selection unit that selects an edge pixel filter for an edge pixel from among a plurality of filters having different set frequency characteristics, and selects a non-edge pixel filter for a pixel that is not an edge pixel; The frequency of the edge pixel filter preset by the filter selection unit is determined according to the determination unit determining whether the edge pixel is the attention area. A spatiotemporal filter device having a filter control unit for controlling numerical characteristics and a filter processing unit for performing a filter process on an input image. 7. A spatio-temporal filter device for limiting a band of an image, comprising: an edge detection unit for detecting an edge of an input image; an edge information storage unit for storing edge information of each detected pixel; An edge density measuring unit that measures the density of edge pixels around the edge of the image, an edge attention determining unit that determines whether the edge pixel is a pixel of the attention area based on the edge density, and the edge attention determination unit determines that the pixel is not the attention area. A pixel of interest measurement unit that measures the density of a pixel that is determined not to be a region of interest around the pixel, and a pixel of interest that is determined to be not a region of interest. A pixel-of-interest correction unit that determines whether or not to re-classify the pixel into a region of interest based on the density of the pixel determined to be non-existent; A filter selection unit that selects an edge pixel filter as an edge pixel from among a plurality of filters, and selects a non-edge pixel filter as a non-edge pixel; Judgment whether or not the region, or in the target pixel correction unit,
A filter control unit for controlling a frequency characteristic of an edge pixel filter preset in the filter selection step in accordance with a determination as to whether or not pixels are to be re-classified into a region of interest; a filter processing unit for performing a filter process on an input image And a spatio-temporal filter device comprising: 8. A spatio-temporal filter device for limiting a band of an image, comprising: an edge detection unit for detecting an edge of an input image; an edge information storage unit for storing edge information of each detected pixel; An edge density measuring unit that measures the density of edge pixels around different areas in different ranges; an edge attention determining unit that determines whether an edge pixel is a pixel in a region of interest based on a plurality of ranges of edge densities; For a pixel that is determined not to be a region of interest by the determination unit, a pixel of interest measurement unit that measures the density of pixels that are determined to be not a region of interest around the pixel, and a pixel that is determined to be not a region of interest A pixel-of-interest correction unit that determines whether or not to re-classify the pixel into a region of interest based on the density of the pixels that are determined not to be the region of interest in the vicinity thereof; A filter selection unit that selects an edge pixel filter for an edge pixel from among a plurality of filters having different set frequency characteristics, and selects a non-edge pixel filter for a pixel that is not an edge pixel; In the determination unit, whether the edge pixel is a region of interest, or in the pixel of interest correction unit,
A filter control unit for controlling a frequency characteristic of an edge pixel filter preset in the filter selection step in accordance with a determination as to whether or not pixels are to be re-classified into a region of interest; a filter processing unit for performing a filter process on an input image And a spatio-temporal filter device comprising: 9. A band limiting program for limiting a band of an image, comprising: an edge detecting procedure for detecting an edge of an input image; an edge information storing procedure for storing edge information of each pixel; An edge density measurement procedure for measuring pixel density; an edge attention determination procedure for determining whether an edge pixel is a pixel of an attention area based on the edge density; and an edge among a plurality of filters having different preset frequency characteristics. A filter selection procedure of selecting a filter for an edge pixel for a pixel and a filter for a non-edge pixel for a pixel that is not an edge pixel, and in the edge attention determination procedure, according to the determination of whether the edge pixel is an attention area, A filter control procedure for controlling the frequency characteristics of the edge pixel filter preset in the filter selection procedure; A recording medium on which a band limiting program for causing a computer to execute a filtering process for performing a filtering process on an input image is recorded. 10. A band limiting program for limiting a band of an image, comprising: an edge detecting procedure for detecting an edge of an input image; an edge information accumulating procedure for accumulating edge information of each pixel; An edge density measurement procedure for measuring pixel densities in different ranges; an edge attention determination procedure for determining whether an edge pixel is a pixel of an attention area based on a plurality of ranges of edge densities; and a preset frequency characteristic. Among a plurality of filters different from each other, an edge pixel filter is selected for an edge pixel, and a non-edge pixel filter is selected for a pixel that is not an edge pixel. Is a region of interest, the frequency of the edge pixel filter preset in the filter selection procedure is determined. A recording medium recording a band limiting program for causing a computer to execute a filter control procedure for controlling numerical characteristics and a filter processing procedure for performing a filtering process on an input image. 11. A band limiting program for limiting a band of an image, comprising: an edge detecting procedure for detecting an edge of an input image; an edge information accumulating procedure for accumulating edge information of each pixel; An edge density measurement procedure for measuring pixel density, an edge attention determination procedure for determining whether an edge pixel is a pixel in an attention area based on the edge density, and a pixel determined to be not an attention area by the edge attention determination procedure A pixel-of-interest measurement procedure for measuring the density of pixels determined not to be the region of interest in the vicinity of the pixel; and determining that the pixel determined to be not the region of interest is not the region of interest in the vicinity. A pixel-of-interest correction procedure for determining whether or not to re-classify a pixel into a region of interest based on the pixel density, and a difference in a preset frequency characteristic. A filter selection step of selecting an edge pixel filter as an edge pixel from among a plurality of filters, and a non-edge pixel filter as a non-edge pixel pixel; The frequency characteristics of the edge pixel filter preset in the filter selection procedure are controlled in accordance with the determination as to whether or not the pixel is the area of interest or whether or not to re-classify the pixel into the area of interest in the pixel-of-interest correction procedure. A recording medium storing a band limiting program for causing a computer to execute a filter control procedure and a filter processing procedure for performing a filter process on an input image. 12. A band limiting program for limiting a band of an image, comprising: an edge detecting procedure for detecting an edge of an input image; an edge information accumulating procedure for accumulating edge information of each pixel; A plurality of edge density measurement procedures for measuring pixel densities in different ranges; an edge attention determination procedure for determining whether an edge pixel is a pixel of an attention area based on the edge densities of a plurality of ranges; A pixel-of-interest correction procedure for determining whether to re-classify the pixel into a region of interest based on the density of the pixel determined not to be the region of interest surrounding the pixel for the determined pixel; From among a plurality of filters having different frequency characteristics, a filter for edge pixels is selected for edge pixels, and a filter for non-edge pixels is selected for pixels that are not edge pixels. According to a filter selection procedure for selecting a filter, in the edge attention determination procedure, it is determined whether or not an edge pixel is an attention area, or in the attention pixel correction procedure, in accordance with a determination as to whether or not to re-classify a pixel as an attention area. A recording medium on which a band control program for causing a computer to execute a filter control procedure for controlling a frequency characteristic of an edge pixel filter preset in a filter selection procedure and a filter processing procedure for performing a filter process on an input image is recorded.