JPH0340179A - Image processor - Google Patents

Abstract

PURPOSE:To prevent the blur of an image due to the magnification/reduction processes and the emphasis of a period structure or the moire noises by controlling the space frequency response of a picture based on an image area signal and variable magnification. CONSTITUTION:An image area identifying means 103 identifies the attribute of an image, e.g., a character/line image, a dot image or an isolated noise based on the signal that read the picture. Then the means 103 outputs an image area signal. If the signal shows a character/line image, the high frequency component of the image is emphasized and the degree of emphasis is controlled by the variable magnification. In a magnification state of a magnification/reduction processing circuit 104, for example, the degree of emphasis is increased. Thus, a clear character/line image is obtained with no blur despite the increase of the magnification ratio. Then no moire noise, etc., are produced even in a reduc tion state for the dot image or the isolated noise. Furthermore such an indecent period structure as a dot image, etc., is never emphasized even in a magnifica tion state. Then an image of high quality is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an image processing device that corrects a decrease in the spatial frequency response of an image during scaling.

(Prior Art) Conventionally, in a device that reads an image using an image input device, processes it to enlarge or reduce it, and outputs a hard copy, etc., the response of high-frequency components is affected due to the response of the reading lens, the aperture size of the sensor, etc. This may cause the output image to become blurry. Furthermore, in addition to the reduced response of such image input devices, output images often become blurred due to the scaling algorithm. In particular, when the enlargement ratio is increased, it is difficult to suppress blur no matter what enlargement/reduction processing algorithm is used, and the edges of character images and the like become blurred, resulting in a disadvantage that the image becomes very unsightly.

Therefore, it is known that by combining scaling processing and high-frequency emphasis processing that emphasizes the high-frequency components of an image, it is possible to suppress the drop in the response of the high-frequency components and prevent the occurrence of blurring caused by scaling processing. (Unexamined Japanese Patent Publication No. 62-25727
No. 6).

However, depending on the image, performing high frequency enhancement may actually make the image unsightly. For example, when a halftone image is enlarged, the halftone structure becomes easier to see, and when high frequencies are further emphasized,
Only the halftone structure is visible and it becomes difficult to identify the original image. Further, when reducing a halftone image, moiré noise may occur.

In addition, there is a drawback that performing high frequency enhancement processing after enlarging processing has little effect, and conversely, if high frequency enhancement processing is performed and reduction processing is performed, not only will moire noise occur, but there will also be a lot of graininess. There were also some drawbacks, such as the appearance of images.

(Problems to be Solved by the Invention) As described above, with conventional image processing devices that uniformly emphasize high-frequency components by correcting the drop in high-frequency spatial frequencies caused by scaling processing, depending on the type of image, it may be unsightly. This method has drawbacks such as emphasizing periodic structures and moiré noise, and not producing significant effects.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing device that can effectively prevent the occurrence of blurring and emphasis of periodic structures, moiré noise, etc. due to scaling processing, and can significantly improve image quality.

[Objective of the Invention] (Means for Solving the Problems) This invention provides image attributes such as character/line image, halftone dot image, or isolated point noise from an image signal. image area identification means for identifying and outputting the identification result as an image area signal, scaling processing means for scaling the image, and filtering processing means for controlling the spatial frequency response of the image based on the image area signal. It is characterized by having the following.

5- The present invention also provides means for controlling the response of a spatial frequency in the vicinity of 1/2 of the Nyquist spatial frequency of the image signal based on the image area signal, and a means for controlling the response of the spatial frequency in the controlled image signal by this means. The invention is characterized by comprising means for enlarging/reducing the image.

Furthermore, the present invention provides a two-dimensional spatial filtering means for controlling the spatial frequency response by performing two-dimensional spatial filtering processing on the image signal based on the image area identification signal in a device that processes one line of the image signal every time it is input. image area identification means for identifying attributes of the image based on the processing results of the means and outputting the identification results as the image area signal; and one-dimensional enlargement for the processing results of the two-dimensional filtering means. An enlargement/reduction processing means for performing reduction processing;
The present invention is characterized by comprising one-dimensional spatial filtering means for performing one-dimensional spatial filtering processing on the image signal enlarged/reduced by this means based on the image area signal and controlling the spatial frequency response.

6- (Function) According to the present invention, the image area identifying means identifies the attribute of the image, for example, whether it is a character/line image, a halftone image, or isolated point noise, from the signal obtained by reading the image, and identifies the image area signal. Output.

If this image area signal indicates, for example, a character/line image, the high frequency components of the image are emphasized. At this time, it is desirable that the degree of emphasis is also controlled by the scaling factor. For example, when enlarging, the degree of emphasis is increased. Furthermore, if the identification signal indicates a halftone image or an isolated point, low-pass processing is performed on the contrary.

As a result, even when the magnification ratio becomes large, text/line images can be obtained without blurring and clear images can be obtained, and halftone images or isolated point noise can be obtained without moiré noise even when being reduced, and when being enlarged. However, high-quality images can be obtained without emphasizing unsightly periodic structures such as halftone dot images.

In addition, by performing high-frequency emphasis processing on the neighboring spatial frequency that is 1/2 of the Nyquist spatial frequency, and then performing scaling processing, the effect of high-frequency emphasis processing during enlargement is enhanced, and moiré noise and noise are reduced during reduction. It becomes possible to effectively remove roughness noise.

Furthermore, if one-dimensional scaling processing is performed after two-dimensional spatial filtering processing, only t-dimensional filtering processing is sufficient for subsequent correction of the spatial frequency response. Therefore, it becomes possible to easily perform real-time processing for each line.

(Example) Hereinafter, an image processing apparatus according to an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an image processing apparatus according to a non-embodiment.

In FIG. 1, a reading device 101 is composed of a reading lens, a line sensor, etc., and reads image information from a document line by line.
It consists of a color reading device or a black and white only reading device as shown in No. B and Japanese Patent Application No. 135238/1984. Image information read by this reading device 101 is given to a two-dimensional spatial filter circuit 102.

The two-dimensional spatial filter circuit 102 emphasizes high-frequency components in order to correct a drop in high-frequency response caused by the reading lens and sensor aperture shape in the reading device 101. Furthermore, the two-dimensional spatial filter circuit 102 performs high-frequency suppression, that is, low-pass processing, depending on the image.

The image area identification circuit 103 inputs the high frequency component of the image obtained by the two-dimensional spatial filter circuit 102, and based on this input, determines the attributes of the image, that is, whether it is a character/line image, a halftone image, or an isolated point. Identify attributes such as noise. The identification results are output as image area signals to the two-dimensional spatial filter circuit 102 and the one-dimensional spatial filter circuit 105. As a result, the two-dimensional spatial filter circuit 10
The two-dimensional and one-dimensional spatial filter circuits 105 receive information on whether edges should be emphasized or blurred.

The scaling processing circuit 104 is a two-dimensional spatial filter circuit 1
One-dimensional scaling processing is performed on one image signal that has been subjected to two-dimensional filtering processing in step 02 at a set magnification.

The one-dimensional spatial filter circuit 105 has a function of performing one-dimensional filtering processing on the enlarged/reduced image signal and removing moiré noise from the image signal. That is, when the output device 1O6 at the next stage expresses halftones by performing dithering or halftone modulation, the output image has a periodic structure, and there is a gap between this periodic structure and the image data. Moiré noise may occur. Therefore, the one-dimensional spatial filter circuit 102 identifies whether the image signal is an image that is likely to cause moire noise based on the image area signal received from the image area identification circuit 103, and controls the filtering characteristics to enlarge or reduce the image. Remove periodic structures that may be moiré noise from the processed image signal.

In addition, the CPU 107 and the control panel t08 are
It is provided to set the magnification ratio and control the feed speed of the reading device 101 and the output device IO6 based on this setting.

0- Next, the operation of the image processing apparatus configured as described above will be explained.

In this embodiment, the following processing is performed every time the image signal is read by the reading device 101 for one line. First, the image signal read line by line by the reading device 101 is shading corrected and output. In this embodiment, high-frequency emphasis is performed to correct the deterioration of response characteristics due to optical systems such as lenses and deterioration of response characteristics due to scaling processing. Therefore, if there is shading noise (containing a lot of high-frequency component noise) caused by uneven sensitivity of the sensor, etc., this will be emphasized and it will become very unsightly. Therefore, remove such noise in advance.

The shading-corrected image signal is input to a two-dimensional spatial filter circuit 102. In this embodiment, the enlarging/reducing process is performed in the line sensing direction (hereinafter referred to as the main scanning direction) by signal processing (interpolation processing) in the enlarging/reducing processing circuit 104. Paper feeding direction (hereinafter referred to as sub-scanning direction)
This is done by controlling the mechanical feed speed and changing the number of sampling data in the sub-scanning direction. For this reason,
In the main scanning direction, the image is blurred due to interpolation processing during enlargement, and in the sub-scanning direction, the aperture size of the reading sensor is constant, so even if the number of samplings is increased, the high frequency response will still be blurred. Occur. For this reason, the two-dimensional spatial filter circuit 102 controls the filtering characteristics in both the main and sub-scanning directions.

The two-dimensional spatial filter circuit 102 includes, for example, a 2×2 low-pass filter 201, a 3×3 low-pass filter 202, a Laplacian calculation unit 203, and a high-frequency emphasis unit 205, as shown in FIG.

First, a 2×2 low-pass filter 2 is connected to the input terminal 200.
The image signal input to 01 is added to the image signal delayed by one line by a delay line 206 in an adder 207. Next, the output of the adder 207 is added to the image signal delayed by one pixel by the one-pixel delay circuit 208 in the adder 209. This addition result is averaged to 174 by means not shown. Thereby, 2×2 low-pass processing is performed on the image signal.

Next, the 3×3 low-pass filter 202 performs addition on every other line and every other pixel. That is, the input image signal is added in an adder 212 to an image signal delayed by two lines in the line direction by delay lines 210 and 211. Next, the output of the adder 212 is sent to the 1-pixel delay circuit 213.
, 214 and the image signal delayed by two pixels and the adder 215
It is added at . This addition result is also 1/4 as above.
averaged to As a result, a convolution operation between the 2×2 low-pass filter 201 and the 3×3 low-pass filter 202 is performed.

FIG. 3(a) shows the kernel of the 2×2 low-pass filter 201, and FIG. 3(b), (c), and (d) show the kernel of the 2×2 low-pass filter 201, and the 3×3 low-pass filter 202. The kernels of 13 and the kernels of the results of these convolution operations are shown, respectively.

The output of the 2×2 low-pass filter 201 and the convolution calculation output are subtracted by the adder 216 of the Laplacian calculation unit 203, and then subtracted by the overflow processing unit 217.
will be processed. With this, the Laplacian can be found.

The obtained Laplacian signal 218 is sent to the image area identification circuit 1
Sent to 03. In addition, the Laplacian signal 218 and 2×
The output signals of the second low-pass filter 201 are each sent to a delay line section 204 in order to synchronize with the identification signal of the image area identification circuit 103. The delay line section 204 transfers these signals to delay lines 219 and 220.
delay each line by one line.

The spatial frequency of this delayed output is controlled by the high frequency emphasizing section 205. That is, the convolution calculation output is
In the ROM table 221, it is multiplied by a coefficient (K) determined by the identification signal and the magnification rate. The convolution calculation output multiplied by the coefficient is passed through a 2×2 low-pass filter 20.
1 and the adder 222. The result of this calculation is subjected to overflow processing in the overflow processing unit 223, and then output to the output terminal 224 as a secondary spatial filter output.
is output from.

Here, when u==2πf/fs (fs is the sampling frequency), the spatial frequency characteristic H(u, k) of the output of the two-dimensional spatial filter is as shown in the following equation.

H(u,k) In this equation, the spatial frequency characteristic H(u,k) when k is changed is shown in FIG. 4. As is clear from this figure, the response is O at the Nyquist frequency fs/2, and the response can be increased or decreased near the Nyquist frequency of 172 depending on the value of the coefficient (K).

For example, if the character-likeness of the identification signal is strong, k will be made thicker L (k=2), and in the case of a halftone image, 1 will be made smaller (k=
o) Do. Furthermore, in an output device that uses a periodic structure such as a dither method, it is also effective to make k even smaller (kneel). For example, when using a 4 x 4 multilevel dither method, by setting k = -1, the response to the periodic acid component (f s / 4) generated in the output device is set to O, which causes moirenois etc. It becomes possible to remove. In this way, table P is used for the coefficients that determine the response characteristics.
225. The table P225 includes an image area identification signal 226 and a signal 22 indicating the magnification rate from the CPU 107.
The parameter k is output based on 7. Details of this table P225 will be described later.

Since the spatial frequency characteristic H(u, k) of such a two-dimensional spatial filter circuit 102 is as shown in FIG. 4, there are the following advantages.

In other words, during reduction, the resampling frequency fSl will decrease as shown in Figure 4, but when the magnification is 100% to 50%, the component at the resampling frequency fs is not emphasized so much that the image quality is free from graininess. is obtained. In addition, since the enlargement process is performed after high-frequency emphasis is performed, setting the manual sampling frequency fs to 16+p/mm effectively increases the sharpness of the characters by emphasizing the vicinity of 4] p/mm. becomes possible. On the other hand, if high-frequency emphasis is performed after performing enlargement processing, spatial frequency components (containing many noise components) considerably higher than the vicinity of N 41p/mm will be emphasized, making it impossible to increase sharpness.

Next, the operation of the image area identification circuit 103 will be explained.

This circuit basically identifies the attributes of the image area from a combination of binarized patterns of the Laplacian signal of the image, as described in Japanese Patent Application Laid-Open No. 204177. The configuration is as shown in .

That is, the Laplacian signal 218 output from the Laplacian calculation unit 203 in FIG. 2 is inputted via the input terminal 500, compared with a predetermined threshold value by the comparator 501, and binarized. The binarized signal is converted into a 4x4 Laplacian by 1-bit delay lines 502, 503, and 504 and 16 La17-edges 505 of 11 pixels per t pitch arranged in a 4x4 matrix. The signal is extracted as a binary pattern. These cut out 1
The 6-bit signal is input to the determination table 506. The determination table 50Ei is created by statistically examining the properties of character images and halftone dot images in advance. That is, if the image is a character/line image, the pattern will be a combination of consecutive dots. Furthermore, if the image is a halftone image, it will be a combination of patterns in which dots are dispersed. Therefore, a value of 7 is assigned to a 16-bit pattern that looks like a character image, a circle is assigned to a pattern that looks like a halftone dot image, and values between 0 and 7 are assigned depending on the degree to which the pattern resembles a text image or a halftone dot image. It should be noted that efficient identification will be possible if the statistical properties of character images and halftone dot images are reflected in the method of allocating this value. For example, first, images consisting of characters and line images of various sizes and halftone dot images with different numbers of lines are prepared as training images. For each image (character image and halftone dot image), a binarized pattern of the 4×4 Laplacian signal is determined, and a histogram of the frequency of occurrence is determined for each pattern.

Next, for each binarized pattern of the 4×4 Laplacian signal, the ratio of the frequency of occurrence in the character image and the halftone dot image is determined. A scale of character-likeness or halftone image-likeness is set based on the magnitude of this ratio, and is stored in the determination table 506. In this way, the determination table 506 is created as a criterion for determination, that is, a table on which a determination signal is based. When a binarized pattern of a 4×4 Laplacian signal is input, the judgment table 506 selects a large value (7) if the pattern is likely to occur in characters, etc., according to the contents of a table created in advance based on statistical properties. If the pattern is likely to occur in a halftone image, a small numerical value (0) is output as the determination result. This determination result is output from the output terminal 507, and is output from the terminal 22 in FIG.
6.

The following table shows an example of coefficients that change the response characteristics based on this identification signal and the magnification rate instructed from the control panel.

9- As mentioned above, when the magnification is high and the identification signal indicates high character-likeness, the high-frequency components are sufficiently emphasized (
For example, when the magnification is 400%), moire noise is likely to occur during reduction, so if there is a certain degree of dot-likeness, the response of the high frequency component is lowered to suppress the occurrence of moire noise (for example, when the magnification is 50%). In addition, even if the magnification is large, if it is a halftone image, it is possible to lower the response of the high frequency component and make the halftone structure less noticeable. dog)]
. Conversely, if it is an image that looks like text, it will strongly emphasize the high-frequency components and convert it into an easy-to-read image, regardless of the magnification ratio.
When identification signal = 7 (character-likeness dog)].

In this way, by varying the high frequency enhancement coefficient according to the magnification ratio and the discrimination result of the character recognition section, moire noise does not occur even when the image is reduced, and the characters are easy to see without becoming blurred even when the image is enlarged. Image quality is obtained. In this embodiment, the high frequency enhancement coefficient is controlled by referring to both the magnification ratio and the image area identification result, but the high frequency enhancement coefficient is controlled by referring only to the image area identification result. However, it is possible to obtain a certain degree of effect (techniques related to this are described in Japanese Patent Application Laid-open Nos. SHO GO-204+77 and JP-A No. 61-25372).

The image signal subjected to two-dimensional spatial filtering processing based on the image area identification result and the scaling factor is then sent to the scaling processing circuit 1.
04.

For example, as shown in FIG. 6, the enlargement/reduction processing circuit 104 includes a line memory 601, an address control section 602, an enlargement/reduction operation section 603, and a coefficient operation section 604. First, the image signal is stored in the line memory 601, where the time delay associated with the conversion of the number of pixels in the enlargement calculation unit/reduction calculation unit 603 is matched, and reference pixels necessary for interpolation processing are extracted. line memory 60
The read address of 1 is given from the address control unit 602. The image signal output from the line memory 601 is input to an enlargement/reduction operation section 603 and is subjected to enlargement/reduction processing. In this embodiment, the enlargement processing is performed by linear interpolation processing, and the reduction processing is performed by processing using a projection method, which are different algorithms. In addition, the coefficient calculation unit 6
04 is supplied with a variable magnification signal 605 from the CPU 107 in accordance with the enlargement ratio designated from the control panel 103, and is also supplied with an address signal from the address control unit 602. The coefficient calculation unit 602 generates inter-part coefficients necessary for the expansion/reduction processing based on these signals, and supplies them to the expansion calculation unit◆reduction calculation unit f303.

Through such processing, scaling processing in one-dimensional direction is executed. Note that, as described above, the enlargement/reduction processing in the sub-scanning direction is performed by controlling the difference in feeding time between the reading device 101 and the output device 10Ei.

Next, the signal that has been subjected to the one-dimensional scaling process as described above is input to the one-dimensional spatial filter circuit 105. That is, as explained above, when an image having a periodic structure such as a halftone image is input, the two-dimensional spatial filter circuit 102 performs low-pass processing to eliminate the periodic acid 3 minutes that may cause moiré noise. However, when the scaling process is performed in the scaling processing circuit 104, the spatial frequency input to the output device 106 changes by the magnification bone, so moiré noise may occur. In particular, in the reduction process, the spatial frequency of the input changes to a higher side, so even if low-pass processing is performed by the two-dimensional spatial filter circuit 102, the spatial frequency may change to a value where moiré noise may occur. Therefore, the one-dimensional spatial filter circuit 105 functions to remove the spatial frequency at which this moiré noise occurs. Moiré noise is more likely to occur when a spatial frequency close to the periodic structure used in the output device 106 is output. For example 4
When using the ×4 pixel multi-value dither method, as shown in FIG. 6, the signal that has been scaled up and down by the scale-up/down processing circuit 104 is delayed by 1 pixel delay 606, f307. f30
8, and each is input to an adder 809 and its moving average is taken. The value of this moving average is
It is input to one input terminal of 10. The selector 610 selects the moving average processed signal only when it is determined based on the image area identification signal 611 that a halftone image has been input. This selectively removes only the spatial frequencies that are likely to cause moiré noise.

In this way, when enlarging/reducing processing is performed in the t-dimensional direction, the spatial frequency in the sub-scanning direction does not change at all, so the subsequent processing also suffices with t-dimensional spatial filtering processing, which makes it possible to remove moiré noise. . Note that the spatial frequencies in the sub-scanning direction that may cause moiré noise have already been removed by the two-dimensional spatial filter circuit 102 described above.

Subsequently, the signal processed as described above is input to the γ circuit 6-2. This γ circuit 812 is provided to increase the sharpness of the character image when the input image is extremely character-like. That is, if the image appears to be a character, the two-dimensional spatial filter circuit 102 performs high-frequency emphasis, but especially when the magnification is increased, sharpness may tend to be insufficiently emphasized.

However, if the high frequency emphasis is made too strong, noise components may be emphasized and the image may become unsightly. In such a case, the image quality will be higher if the γ characteristic is made steeper after high-frequency emphasis is performed to make the image closer to a binary image.

Therefore, when the image area identification signal 611 indicates that it is extremely likely to be a character, the γ characteristic of the γ circuit 612 is made steep.

In this embodiment, the image area identification signal 611 is directly used as a switching signal for the selector 61O and the γ circuit 612, but the output of the table P225 in FIG. 2 may be used as the switching signal. In this case, an appropriate switching signal can be obtained depending on the magnification ratio. Further, a new image area identification signal 611 and magnification rate table may be provided independently of the table P225 in FIG. 2, so that an appropriate switching signal can be obtained depending on the magnification rate.

In addition, in the above embodiment, processing suitable for an output device using a multilevel dither method using 4×4 pixels was explained, but 3
It is also effective for an output device using the x3 pixel multi-value dither method. In this case, it is preferable to change the kernel of the two-dimensional spatial filter so that it has the characteristics of a two-dimensional spatial filter that can obtain an O response at one-third the sampling frequency. In addition, the one-dimensional spatial filter also uses low-pass processing using a moving average of three pixels to reduce the output at 1/3 the sampling frequency.
It is better to make it a response.

Although the second embodiment describes an output device having a periodic structure, it is not necessary to use such an output device to be effective. That is, even if the output device is capable of displaying halftones with one pixel, moiré noise is likely to occur during the reduction process. Further, when the halftone dot image is enlarged, the halftone dot structure is also enlarged, and this enlarged halftone dot structure is clearly visible due to the response characteristics of the eye, making it very unsightly.

On the other hand, it would be better if the letters had a sharp feel even when enlarged.

Therefore, even in such a case, it is effective to make the spatial frequency characteristics and γ characteristics variable depending on the magnification ratio and image attributes (character images, halftone images, etc.).

Furthermore, this processing can also be applied to color images.

In this case, a color image reading device may be used as the reading device, and after performing color conversion processing to convert the ink amount signal into an ink amount signal within the reading device, the signal may be output. The rest can be carried out in the same manner as in the previous embodiment. In short, in a color reading device, three systems of R, G, and B are processed at the same time, but after color conversion processing, it is processed as a single color ink amount signal, and if the processing is repeated for the number of ink colors in the color output device, there is no difference. This can be done in the same manner as in the embodiment. In addition, in the case of an output device that can perform simultaneous color output, such as an inkjet printer, if processing is performed by switching color signals sequentially for each pixel, there is no need for multiple color processing circuits, and one color The circuit is fine. However, the delay circuits used in filter circuits and the like are required for each color channel.

[Effects of the Invention] As explained above, according to the present invention, an image can be extracted by means of identifying an image attribute from an image signal, for example, whether it is a character/line image, a halftone image, or isolated point noise. A region identification signal is obtained, and this image region identification signal is used to identify, for example, characters/characters.
If it is a line image, high-frequency components are emphasized, and if it is a halftone dot image or isolated point 8 noise, low-pass processing is performed. Therefore, even if the magnification ratio becomes large for character/line images, it is possible to output them clearly without blurring. In addition, even when reducing halftone dot images, moiré noise does not occur, and even when enlarging halftone dot images, unsightly periodic structures and isolated point noise are not emphasized, resulting in high-quality images. It will be done.

In addition, by performing high-frequency enhancement processing on the neighboring spatial frequency that is 1/2 of the Nyquist spatial frequency, and then performing scaling processing, it has the effect of increasing the sharpness of characters and line images when enlarged, and when reducing It is possible to effectively remove moiré noise and roughness noise.

Furthermore, if one-dimensional scaling processing is performed after two-dimensional spatial filtering processing, one-dimensional filtering processing is sufficient for subsequent correction of the spatial frequency response.
Real-time processing for each line becomes possible.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an image processing device according to an embodiment of the present invention, Fig. 2 is a detailed block diagram of a two-dimensional spatial filter circuit in the same device, and Fig. 3 shows the operation of the two-dimensional spatial filter circuit. 4 is a characteristic diagram showing the response characteristics of the same two-dimensional spatial filter circuit, and FIG. 5 is a detailed block diagram of the image area identification circuit in the apparatus of FIG. 1.
FIG. 6 is a detailed block diagram of the scaling circuit and one-dimensional spatial filter circuit in the same device. 101... Reading device, 102... Two-dimensional spatial filter circuit, 103... Image area identification circuit, 104... Enlargement/reduction processing circuit, 105... One-dimensional spatial filter circuit,
106... Output device, 107... CPU1108.
··Control panel.

Claims (5)

[Claims] (1) Image area identification means for identifying image attributes from image signals and outputting the identification results as image area signals; scaling processing means for scaling the image; and scaling processing with the image area signal. and filtering processing means for controlling a spatial frequency response of the image based on a scaling factor. (2) The image processing apparatus according to claim 1, wherein the image area identifying means identifies attributes of the image based on a Laplacian pattern of the image. The image processing apparatus according to claim 1, further comprising: (3) means for controlling gamma characteristics of the filtered image based on the image area signal obtained from the image area identifying means. (4) image area identification means for identifying image attributes from the image signal and outputting the identification result as an image area signal;
1. An image processing apparatus comprising: means for controlling a spatial frequency response in the vicinity of No. 2, and means for enlarging/reducing an image signal whose spatial frequency response has been controlled by said means. (5) two-dimensional spatial filtering means for controlling the spatial frequency response of the image signal by performing two-dimensional spatial filtering processing on the image signal based on the image area signal; image area identification means for identifying attributes of the image area and outputting the identification result as the image area signal; and scaling processing means for performing one-dimensional scaling processing on the image signal processed by the two-dimensional spatial filtering means. , one-dimensional spatial filtering means for performing one-dimensional spatial filtering processing on the image signal processed by the enlargement/reduction processing means based on the image area signal and controlling the spatial frequency response, An image processing device characterized in that it performs these processes at each time.