JP2002064712A - Image processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize high compression rate, while suppressing the occurrence of image quality deterioration, even in the case of compressing/expanding image data and especially restrain image quality deterioration due, especially twisted character and mosquito noise or the like. SOLUTION: The image processing unit 10 is constituted with a character/ pattern separation means 12, that separates received image data into a character area and a pattern area, a quantization means 14b that uses one of preset quantization tables 15a-15c to quantize the image data for data compression, and a table changeover means 15d that changes over the quantization tables 15a-15c used by the quantization means 14b for the character area and the pattern area, on the basis of the result of separation by the character/pattern separation means 12.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an image processing apparatus having a function of performing data compression processing on image data.

[0002]

2. Description of the Related Art Generally, as a data compression technique for image data of a still image, JPEG (Joint Photo
Lossless compression, such as the Graphicographic Experts Group), is known. In JPEG, image data is divided into blocks of 8 × 8 pixels, and discrete cosine transform (Disc
Rete Cosine Transform (hereinafter abbreviated as “DCT”) and Huffman coding reduce redundant data and high-frequency data to reduce the amount of data.

[0003] By the way, since JPEG is irreversible compression, if its compression ratio is increased, image quality is deteriorated when compressed image data is decoded and reproduced.
In order to suppress such image quality degradation as much as possible and to increase the compression ratio, for example, Japanese Patent Application Laid-Open No. 10-285409 discloses that image data is divided into blocks in a data compression / expansion process, and each block is divided into blocks. The edge / non-edge is determined for the block, and for the block determined to be non-edge,
By performing a compression process after performing a correction process using a low-pass filter, the frequency-resolved high-frequency component is lost to increase the compression ratio. On the other hand, for a block determined to be an edge, the correction process is not performed. There has been proposed an image data processing device which prevents a component from blurring and eliminates deterioration of image quality of a decoded image.

[0004]

However, in the image data processing apparatus disclosed in Japanese Patent Laid-Open No. 10-285409, for example, edge / non-edge determination is performed for each block of 8 × 8 pixels. 1 near the separation boundary
Since the density of pixels affects the determination result of the entire block, there is a possibility that a part which is originally an edge is determined as a non-edge. For example, a block of 8 × 8 pixels is shown in FIG.
If the density value is as shown in FIG.
The 8 × 8 DCT coefficients converted by T are as shown in FIG. 3B, and the block is determined to be an “edge”. However, as shown in FIG. 4A, FIG. Is different from the case of (1), even if only one pixel value (see the hatched portion in the figure) is changed, the 8 × 8D converted by the two-dimensional DCT is used.
The CT coefficients are as shown in FIG. 4B, and the block is determined to be “non-edge”.

[0005] As described above, when a part which is originally an edge is determined to be a non-edge, the block is
Since the compression processing is performed so as to increase the compression rate by losing the high-frequency components, when the decoded image is reproduced, the loss of the high-frequency components may be, for example, as shown in FIG. May appear. In addition, by losing the high-frequency component, so-called mosquito noise may appear around the edge portion of a portion where the gradation change is large, such as a character or a line.

Accordingly, the present invention makes it possible to realize a high compression ratio while suppressing the deterioration of image quality even when performing compression / expansion of image data. Accordingly, it is an object of the present invention to provide an image processing apparatus capable of reliably eliminating image quality deterioration due to character twisting, mosquito noise, and the like.

[0007]

According to the present invention, there is provided an image processing apparatus devised to achieve the above object, wherein a pictographic character separating means for separating input image data into a character area and a picture area, Quantizing means for performing quantization for data compression on the input image data by using any of the plurality of quantization tables obtained as described above, and a character area and a pattern based on a separation result by the pictogram separating means. Table switching means for switching a quantization table used when the quantization means performs quantization between regions.

In the image processing apparatus having the above configuration, the table switching means switches a quantization table used when the quantization means performs quantization between the character area and the picture area.
For this reason, for example, for the character area, the high-frequency component is held using a quantization table with a small coefficient with priority on image quality, and the occurrence of character twisting and mosquito noise caused by loss of the high-frequency component is suppressed, while the picture area is Since the image quality degradation is not conspicuous even if the compression rate is increased, it is possible to use a quantization table having a large coefficient with priority on the compression rate.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An image processing apparatus according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram illustrating a functional configuration of an example of an image processing apparatus according to the present invention. Here, a case in which the present invention is applied to an image processing apparatus mounted on a digital copying machine will be described as an example.

First, prior to description of an image processing apparatus to which the present invention is applied, a digital copier equipped with the image processing apparatus will be briefly described. As shown in FIG. 1, the digital copying machine includes an image processing system (Image Processing System).
(hereinafter referred to as “IPS”) 10 and an image reading terminal (hereinafter referred to as “IIT”) 20.
And an image forming device (Image Output Terminal;
T ”) 30).

An IIT 20 is a CCD (Charge Coupled Device) 2 for optically reading image data from a document image.
An A / D (analog / digital) converter 22 for digitizing image data read by the CCD 21;
It is provided with.

The IOT 30 prints out image data as a visible image on a medium such as paper using a well-known electrophotographic technique.

The IPS 10 uses these IITs 20 and I
When the image data read by the IIT 20 is received between the OT 30 and the IIT 20
Image processing necessary for output from the OT 30 is performed.

In the IPS 10, if necessary,
The image data received from the IT 20 is temporarily stored (spooled). for that reason,
The IPS 10 performs a data compression process on the image data from the IIT 20 according to the JPEG system to reduce the storage capacity required for the spool, and decodes the image data before sending it to the IOT 30.

Here, the functional configuration of the IPS 10 will be described in more detail. The IPS 10 includes color conversion units 11a and 11b, a TI separation processing unit 12, a filtering unit 13, a DCT unit 14a, and a quantizer 14b.
, The quantization table selector 15 and the Huffman encoder 1
4c, a memory unit 16, a Huffman decoding unit 17a,
It has an inverse quantizer 17b, an IDCT unit 17c, a filtering unit 18, and a gradation correction unit 19.

The color conversion units 11a and 11b respectively
The color component is converted for the image data received from the IT 20. Specifically, the color conversion unit 11
a converts R (red), G (green), and B (blue) color component signals into L ^* a ^* b ^* signals represented by lightness and chromaticity. The color conversion unit 11b converts the L ^* a ^* b ^* signal into an IO
The signal is converted into color component signals of Y (yellow), M (magenta), C (cyan), and K (black) suitable for print output at T30.

The TI separation processing section 12 performs a character (text) / picture (image) separation process on the image data from the IIT 20. That is, for each pixel constituting the image data, it is determined whether each pixel belongs to a character area or a picture area, and furthermore, pixels belonging to the character area represent black characters or color characters (intermediate saturation characters). Is determined, and these determination results are output as tag information (hereinafter, referred to as “TI tag”) for each pixel. However, the details of the character / pattern separation process use well-known technology, and a description thereof will be omitted.

The filtering unit 13 performs a filtering process (for example, a process for adjusting the waveform of image data) for correcting the data reading characteristics of the IIT 20. In the filtering unit 13, T
The parameters for performing the filtering process may be changed according to the result of the character / picture separation process in the I separation processing unit 12.

The DCT unit 14a converts the image data from the IIT 20 into blocks of, for example, 8 × 8 pixels.
The necessary two-dimensional DCT for compressing the data amount is performed. Since the well-known technique is used for the two-dimensional DCT, the description thereof is omitted here.

The quantizer 14b performs scalar or vector quantization on pixels or combinations thereof determined to be significant by the two-dimensional DCT in the DCT unit 14a. As for the scalar or vector quantization, a well-known technique is used, and the description thereof is omitted here. However, the quantizer 14b is configured to perform quantization using one of three preset quantization tables, as described later.

The quantization table selection unit 15 includes a quantizer 1
4b, three quantization tables (Quantzite Tables; hereinafter referred to as “Q-tables”) 15 used when performing quantization.
a to 15c are stored in advance, and which Q-table
A selection unit (hereinafter, referred to as "SEL") 15d for selecting whether to use 15a to 15c is provided. Three Q-tabs
le15a to 15c are Q suitable for quantization of black characters.
-table (black character) 15a and Q suitable for color character quantization
-table (color character) 15b and Q-table (halftone) 15c suitable for quantization of a halftone image. Also, SEL1
In 5d, any one of these three Q-tables 15a to 15c is selected based on the tag information output from the TI separation processing unit 12, and the selection result is
b and the inverse quantizer 14b.

The Huffman encoding unit 14c includes a quantizer 14
Huffman coding (code allocation) is performed on the quantized output from b so that the average code length is shortened. Since the well-known technique is used for the Huffman coding, the description thereof is omitted here.

The memory unit 16 is composed of, for example, a hard disk device. The image data whose data amount has been reduced by the DCT unit 14a, the quantizer 14b, and the Huffman encoding unit 14c (hereinafter referred to as “code data”). ) Is temporarily stored and held.

Huffman decoding unit 17a, inverse quantizer 17
b and the IDCT unit 17c are for decoding the code data in the memory unit 16 and reproducing the image data before compression. Note that the inverse quantizer 17b performs the inverse quantization by using the same Q-tables 15a to 15c used when the quantizer 14b performs the quantization while following the instruction from the SEL 15d in the quantization table selector 15. Shall be performed.

The filtering section 18 performs a predetermined filtering process on the image data reproduced by decoding. More specifically, filtering processing for compensating for data compression characteristics according to the JPEG method, enhancement filter processing for adjusting the image quality of an output image, low-pass filter processing, and the like are performed.

The tone correcting section 19 performs a tone correcting process on the image data reproduced by decoding in accordance with the image output characteristics of the IOT 30. In particular,
Hi-γ gradation correction processing, linear gradation correction processing, and the like are performed.

Each unit other than the memory unit 16 described above
1a-15, 17-19 are dedicated ASICs (Applicatio
n Specific Integrated Circuit).

Next, the IPS 10 configured as described above
An example of the processing operation in will be described. FIG. 2 is a flowchart illustrating an example of a processing operation in the image processing apparatus according to the present invention.

In the IPS 10, upon receiving the image data read by the IIT 20, after undergoing color component conversion processing in the color conversion units 11a and 11b and filtering processing in the filtering unit 13, the DCT unit 14a and the quantizer 14
b and the Huffman encoding unit 14c perform data compression processing on the image data by the JPEG method in units of 8 × 8 pixels.

However, prior to the data compression processing,
The TI separation processing unit 12 performs character / picture separation processing on the image data from the IIT 20, and outputs a TI tag for each pixel of the image data. Then, based on the TI tag, the quantization table selecting unit 15
b selects the Q-tables 15a to 15c used when performing quantization, and instructs the quantizer 14b of the selection result.

More specifically, as shown in FIG. 2, when a TI tag is input from the TI separation processing unit 12 (step 10).
1. Hereinafter, the steps are abbreviated as “S”.) The SEL 15 d of the quantization table selection unit 15 first performs TI separation into a total of 64 pixels in a block of 8 × 8 pixels to be quantized by the quantizer 14 b. The processing unit 12 counts how many pixels belong to the character area and are determined to correspond to black characters. Then, the count is compared with a preset threshold (n / 64) to determine whether the count is greater than the threshold (S102).

If the result of this determination is that the count number is greater than the threshold value, the SEL 15d determines that the block represents a black character (S103), and the three Q-tables 1
Q-table suitable for quantization of black characters from 5a to 15c
(Black character) 15a is selected (S104) and its Q-table is selected.
An instruction is given to the quantizer 14b to perform quantization using (black characters) 15a.

By following this instruction, the quantizer 14b quantizes the block determined to be a black character using a Q-table (black character) 15a suitable for quantizing the black character. Specifically, by using a Q-table (black character) 15a in which the coefficient is set to be small, the frequency-resolved high-frequency component is not lost even if quantization is performed.

If the number of pixels of a black character is equal to or less than the threshold value, the SEL 15d then determines whether the pixel belongs to the character area by the TI separation processing unit 12 and has a color of 64 pixels in a block of 8 × 8 pixels. The number of pixels determined to correspond to the character (intermediate saturation character) is counted. Then, the count number is set to a predetermined threshold value (n / 64).
Then, it is determined whether or not the count number is larger than the threshold value (S105).

If the result of this determination is that the count number is greater than the threshold, the SEL 15d determines that the block represents a color character (S106), and the three Q-tables 1
Q-table suitable for quantization of color characters from 5a to 15c
(Color character) 15b is selected (S107), and its Q-table is selected.
An instruction is given to the quantizer 14b to perform quantization using the (color character) 15b.

By following this instruction, the quantizer 14b performs quantization on the block determined to be a color character by using a Q-table (color character) 15b suitable for color character quantization. Become. More specifically, since the deterioration of the image quality (such as twisting of characters) is less noticeable in color characters than in black characters, the coefficient Q is set to be slightly larger than in black characters.
Using -table (color character) 15b to achieve both image quality and performance (compression ratio).

If the number of pixels of the color character is equal to or less than the threshold value, the SEL 15d determines that the block does not represent a character but a halftone image, that is, belongs to a picture area (S108), and the three Q-tables 15a Q-table (halftone) 15c suitable for quantizing halftone images from
Is selected (S109), and the Q-table (halftone) 15c is selected.
Is given to the quantizer 14b so as to perform quantization by using.

By following this instruction, the quantizer 14b quantizes the block determined to be a halftone image using a Q-table (halftone) 15c suitable for quantizing the halftone image. Will be. More specifically, since the image quality degradation of the halftone image is not conspicuous even if the compression ratio is increased, a Q-table (halftone) 15c having a large coefficient with a priority on the compression ratio is used.

As described above, when the DCT unit 14a, the quantizer 14b, and the Huffman encoding unit 14c perform the data compression processing in units of 8 × 8 pixels, the IPS 10
Then, the memory unit 16 temporarily stores the code data after the data compression processing.

Thereafter, the code data in the memory section 16 is
When transmitting to the OT 30, the Huffman decoding unit 17
a, the inverse quantizer 17b and the IDCT unit 17c decode the coded data extracted from the memory unit 16 in block units to reproduce the image data before compression. At this time,
The inverse quantizer 17b includes a quantizer 14b for each block.
Is the same Q-table 15a as used when performing the quantization.
Inverse quantization is performed by .about.15c.

Huffman decoding section 17a, inverse quantizer 17
b and the IDCT unit 17c reproduce the image data,
Subsequently, in the IPS 10, the filtering unit 18 performs a filtering process. For example, for a block determined to be a character area by the TI separation processing unit 12, an enhancement filter process is performed to enhance an edge portion of an image to enhance sharpness. In addition, for the block determined as a picture area by the TI separation processing unit 12, low-pass filter processing is performed to blur high-frequency components and increase continuity of each pixel.

These filtering processes are performed by the filtering unit 18 after the compression and expansion of the image data. Therefore, even if these filtering processes are performed, there is no possibility that the image quality is deteriorated due to mosquito noise or the like.

Thereafter, in the IPS 10, the gradation correction section 19 performs a gradation correction process. For example, for a block determined to be a character area by the TI separation processing unit 12, gradation correction processing is performed using the Hi-γ parameter.
In addition, for the block determined to be a picture area by the TI separation processing unit 12, gradation correction processing is performed using linear parameters.

The gradation correction processing is performed by the gradation correction unit 19.
Is performed immediately before sending the image data to the IOT 30, that is, at the final stage in the IPS 10. Therefore, the previous processing (data compression processing, filtering processing, and the like) can be performed without being conscious of the image output characteristics of the IOT 30, so that the degree of freedom in configuring the IPS 10 can be increased.

The image data after the gradation correction processing is
The data is sent from the IPS 10 to the IOT 30, where the IOT 30 prints out a visible image on a medium such as paper.

As described above, the IPS in this embodiment
Reference numeral 10 denotes SEL1 of the quantization table selection unit 15 in accordance with the result of the character / picture separation processing by the TI separation processing unit 12.
5d selects the optimal Q-tables 15a to 15c, and uses the selected Q-tables 15a to 15c to
b performs quantization. Therefore, for example, for a black character portion of a character area, high-frequency components are held using a Q-table (black character) 15a having a small coefficient with priority on image quality, and character twisting and mosquito noise due to loss of the high-frequency component are generated. On the other hand, in the halftone image portion of the picture area, even if the compression ratio is increased, the image quality degradation is not conspicuous,
(Halftone) 15c can be used. In other words, according to the IPS 10 of the present embodiment, even when performing compression / expansion of image data, prevention of image quality deterioration and high compression ratio can be realized without depending on edge / non-edge determination. The three-dimensional trade-off that achieves a high compression rate during data compression while preventing the character twisting phenomenon due to the lack of high-frequency components due to data compression, and prevents image quality degradation due to mosquito noise, etc. The realization of the item is possible.

In particular, in the IPS 10 of the present embodiment, the quantizer 14b performs quantization in block units of 8 × 8 pixels, and the SEL 15d of the quantization table selection unit 15 selects the Q-tables 15a to 15c in block units. Is switched. Therefore, the IP of the present embodiment
S10 is very suitable for complying with the JPEG method widely used as a data compression technique for still images.

In addition, the IPS 10 of the present embodiment has a TI
Based on the TI tag output for each pixel from the separation processing unit 12, the SEL 15d of the quantization table selection unit 15 counts the number of black character pixels in the block to be quantized, and calculates Q-tab according to the comparison result
The switching of the selection of the le 15a to 15c is controlled. Therefore, Q-table 15a
Appropriate switching can be performed even in the case of switching the selection of ~ 15c.

In this embodiment, the JPEG data compression processing has been described as an example. However, the present invention is not limited to this. For example, a fixed length compression method that is not a block unit is used. Even if it is a thing, it can be applied exactly the same.

In the present embodiment, the case where the present invention is applied to an image processing apparatus mounted on a digital copying machine has been described, but the present invention is not limited to this. That is, image data may be quantized to reduce the amount of data, and may be applied to those constituting another system as long as it has a character / pattern separation processing function.

[0051]

As described above, in the image processing apparatus according to the present invention, the table switching means switches the quantization table used when the quantization means performs quantization between the character area and the picture area. It has become. For this reason, for example, for the character area, the high-frequency component is held using a quantization table with a small coefficient with priority on image quality, and the occurrence of character twisting and mosquito noise caused by loss of the high-frequency component is suppressed, while the picture area is Since the image quality degradation is not conspicuous even if the compression rate is increased, it is possible to use a quantization table having a large coefficient with priority on the compression rate. In other words, even in the case of compressing / expanding image data, prevention of image quality deterioration and high compression ratio are realized regardless of edge / non-edge determination. While improving the character twisting phenomenon due to missing components, achieving a high compression rate at the time of data compression, and preventing image quality deterioration due to mosquito noise etc.
It is possible to realize a three-dimensional trade-off item.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a functional configuration of an example of an image processing apparatus according to the present invention.

FIG. 2 is a flowchart illustrating an example of a processing operation in the image processing apparatus according to the present invention.

3A and 3B are explanatory diagrams showing a specific example of a conventional edge / non-edge determination when an edge is determined, where FIG. 3A is a diagram of density values of a certain block, and FIG. It is a figure of the 8x8 DCT coefficient converted.

FIGS. 4A and 4B are explanatory diagrams showing a specific example in a case where a conventional edge / non-edge determination is made as a non-edge, where FIG. 4A is a diagram of density values of a certain block, and FIG. FIG. 8 is a diagram of 8 × 8 DCT coefficients converted by (1).

FIG. 5 is an explanatory diagram illustrating an example of image twist due to loss of a high-frequency component.

[Explanation of symbols]

10 IPS (image processing device), 12 TI separation processing unit, 14a DCT unit, 14b quantizer, 14c Huffman encoding unit, 15 quantization table selection unit, 15a
... Q-table (black letters), 15b ... Q-table (color letters),
15c: Q-table (halftone), 15d: SEL, 16 ...
Memory unit, 17a Huffman decoding unit, 17b inverse quantizer, 17c IDCT unit, 18 filtering unit
19: gradation correction unit

Claims (3)

[Claims] 1. A pictographic character separating means for separating input image data into a character area and a picture area, and using any one of a plurality of quantization tables set in advance for data of the input image data. Quantizing means for performing quantization for compression, and table switching means for switching a quantization table used when the quantizing means performs quantization between a character area and a picture area based on a separation result obtained by the pictorial character separating means. An image processing apparatus comprising: 2. The image processing apparatus according to claim 1, wherein the quantization unit performs quantization on the image data in a unit of a block including a plurality of pixels, and the table switching unit performs switching of the quantization table in a unit of the block. The image processing apparatus according to claim 1, wherein: 3. The table switching means controls switching of the quantization table on the basis of the number of pixels in a block to be processed by the quantization means, which is a character area or a picture area. The image processing apparatus according to claim 2, wherein: