JP5181916B2 - Image processing apparatus, compression method and decompression method - Google Patents

Description

  The present invention relates to an image processing apparatus, a compression method, and a decompression method.

  In a printer or the like, image data drawn by rasterization processing is temporarily stored in a memory and sequentially read from the memory when necessary. For example, the controller generates image data with a resolution of 600 dpi, but at the time of printing, it is possible to perform resolution conversion from 600 dpi to 1200 dpi to perform high-resolution printing.

Further, when saving in a memory, it is common to reduce the amount of data by applying compression processing to image data. Conventionally, a BTC (Block Truncation Coding) method has been disclosed as one of the compression methods (see, for example, Patent Document 1). This is a method of performing encoding in units of blocks composed of a plurality of pixels.
JP 11-164150 A

  On the other hand, the controller side can generate image data with a high resolution of 1200 dpi. However, if the resolution is increased, the memory capacity has to be increased, leading to higher costs. Although it is conceivable that the resolution is converted from 1200 dpi to 600 dpi and stored in the memory without changing the capacity of the memory, the image quality may be deteriorated by simply performing the resolution conversion.

It has been found that the edge portions of characters and line drawings cannot be maintained at the original resolution through compression processing or resolution conversion, and sharpness is lost.
In particular, if the sharpness of the edge portion subjected to the anti-aliasing process is lost, the screen dot interval is widened when the screen process is performed, and jaggies may occur in the edge portion, which is not preferable.

  An object of the present invention is to suppress image quality deterioration of an anti-aliased image.

According to the invention of claim 1,
When a control signal indicating that the image has been anti-aliased is input together with the multi-value image to be compressed, the image is switched to compression processing for the anti-aliased image, and the image including the anti-aliased region of the image The area includes an image compression conversion unit that performs quantization by converting the data value of the decompressed image to binary ,
In the compression processing for the anti-aliased image, the image compression conversion unit uses a single threshold value for the image region subjected to the anti-aliasing among the image regions including the anti-aliased region. the image processing apparatus is provided you quantized.

According to invention of Claim 2,
The image processing apparatus according to claim 1, wherein the image compression conversion unit converts the resolution of the image to a lower resolution than before the quantization when the image is quantized.

According to invention of Claim 3,
In the compression processing for the anti-aliased image, the image compression conversion unit processes the image for each predetermined region, and among the processing regions in which the density difference is larger than the first threshold, pixels larger than the second threshold are adjacent. The image processing apparatus according to claim 2, wherein an area to be processed or an area adjacent to pixels smaller than the second threshold is used as an anti-aliased image area , and the image area is quantized by the BTC method using the second threshold. Is done.

According to invention of Claim 4,
The image compression conversion unit processes the image for each predetermined region in the compression processing for the anti-aliased image, and among the processing regions in which the density difference is larger than the first threshold, A region where pixels smaller than two thresholds are adjacently mixed is used as an image region including an anti-aliased region, and a density pattern is created by binarizing the data value of each pixel of the image region. The image processing apparatus according to claim 2, wherein the image processing apparatus performs quantization according to the classification.

According to the invention of claim 5,
The image compression conversion unit processes the image for each predetermined region in the compression processing for the anti-aliased image, and a processing region having a density difference smaller than the first threshold includes a plurality of thresholds including a second threshold. The image processing apparatus according to any one of claims 2 to 4, wherein the image processing apparatus performs quantization using the BTC method.

According to the invention of claim 6,
The said image compression conversion part performs the quantization using these threshold values with respect to the average value which averaged the data value which each pixel of the process area | region where the said density | concentration difference is smaller than a 1st threshold value has. An image processing apparatus is provided.

According to the invention of claim 7,
An image expansion conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 3,
In the original image before quantization, the image extension conversion unit is configured to process a region where a pixel having a density difference larger than the first threshold is adjacent to a region where pixels larger than the second threshold are adjacent or pixels smaller than the second threshold are adjacent. Provides an image processing apparatus that performs decoding by the BTC method and outputs a binary decoded value.

According to the invention described in claim 8,
In the original image before quantization, the image expansion conversion unit includes a region where pixels larger than the second threshold are adjacent, or a region where pixels smaller than the second threshold are adjacent among processing regions where the density difference is larger than the first threshold. The image processing apparatus according to claim 7, wherein the resolution is converted to the same resolution as before the quantization, and a decoded value is assigned to each pixel in the region after the resolution conversion.

According to the invention of claim 9,
An image expansion conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 4,
In the original image before quantization, the image expansion conversion unit is configured to process a region in which pixels larger than the second threshold and pixels smaller than the second threshold are adjacently mixed among the processing regions where the density difference is larger than the first threshold. Provides an image processing apparatus that predicts a density pattern created at the time of quantization for the area and assigns a binary decoded value corresponding to the predicted density pattern to each pixel in the area after resolution conversion. Is done.

According to the invention of claim 10,
An image expansion / conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 5,
The image expansion conversion unit is provided with an image processing apparatus that decodes a processing region having a density difference smaller than a first threshold in an original image before quantization using a BTC method.

According to the invention of claim 11,
The image expansion conversion unit decodes a processing area in which the density difference is smaller than a first threshold, performs resolution conversion to the same resolution as before quantization, and decodes each pixel in the processing area after resolution conversion. An image processing apparatus according to claim 10, which assigns a value, is provided.

According to the invention of claim 12,
When a control signal indicating that the image has been anti-aliased is input together with the multi-value image to be compressed, the image is switched to compression processing for the anti-aliased image, and the image including the anti-aliased region of the image the region, have rows quantization to binary data values of the image after decompression,
In the compression processing for the anti-aliased image, there is a compression method for quantizing the anti-aliased image region into a binary value using a single threshold among the image regions including the anti-aliased region. Provided.

According to the invention of claim 13,
The compression method according to claim 12, wherein when the image is quantized, the resolution of the image is converted to a lower resolution than before the quantization.

According to the invention of claim 14,
In the compression processing for the anti-aliased image, the image is processed for each predetermined region, and among the processing regions in which the density difference is larger than the first threshold, pixels larger than the second threshold are adjacent to each other or from the second threshold. The compression method according to claim 13, wherein a region adjacent to a small pixel is used as an anti-aliased image region , and the image region is quantized by a BTC method using a second threshold.

According to the invention of claim 15,
In the compression processing for the anti-aliased image, the image is processed for each fixed region, and among the processing regions where the density difference is larger than the first threshold, there are pixels larger than the second threshold and pixels smaller than the second threshold. Create a density pattern that binarizes the data values of each pixel in the image area, using the adjacent mixed area as an image area including the anti-aliased area, and quantize according to the density pattern classification A compression method according to claim 13 or 14 is provided.

According to the invention of claim 16,
In the compression process for the anti-aliased image, the image is processed for each fixed region, and a processing region having a density difference smaller than the first threshold is quantized by a BTC method using a plurality of thresholds including the second threshold. A compression method according to any one of claims 13 to 15 is provided.

According to the invention of claim 17,
17. The compression method according to claim 16, wherein quantization using the plurality of threshold values is performed on an average value obtained by averaging data values of pixels in a processing region in which the density difference is smaller than a first threshold value. .

According to the invention of claim 18,
A decompression method for decoding and decompressing an image quantized by the compression method according to claim 14,
In the original image before quantization, the image extension conversion unit is configured to process a region where a pixel having a density difference larger than the first threshold is adjacent to a region where pixels larger than the second threshold are adjacent or pixels smaller than the second threshold are adjacent. Is provided with a decompression method that performs decoding by the BTC method and outputs a binary decoded value.

According to the invention of claim 19,
In the original image before quantization, out of the processing areas where the density difference is larger than the first threshold, the area where pixels larger than the second threshold are adjacent or the area where pixels smaller than the second threshold are adjacent is decoded and resolution is increased. 19. The decompression method according to claim 18, wherein conversion is performed to obtain the same resolution as before quantization, and a decoded value is assigned to each pixel in the region after resolution conversion.

According to the invention of claim 20,
A decompression method for decoding and decompressing an image quantized by the compression method according to claim 15,
In the original image before quantization, the image expansion conversion unit is configured to process a region in which pixels larger than the second threshold and pixels smaller than the second threshold are adjacently mixed among the processing regions where the density difference is larger than the first threshold. Provides a decompression method that predicts a density pattern created during quantization for the area and assigns a binary decoded value corresponding to the predicted density pattern to each pixel in the area after resolution conversion. The

According to the invention of claim 21,
A decompression method for decoding and decompressing an image quantized by the compression method according to claim 16,
The image decompression conversion unit is provided with a decompression method for decoding a processing region having a density difference smaller than the first threshold in the original image before quantization by the BTC method.

According to the invention of claim 22,
23. The decoding of a processing region whose density difference is smaller than a first threshold, performing resolution conversion to the same resolution as before quantization, and assigning a decoding value to each pixel of the processing region after resolution conversion The described stretching method is provided.

  According to the present invention, the data value after decoding can be made binary for the anti-aliased image area. As a result, it is possible to suppress image quality deterioration such as blurring of the antialiased edge portion or jaggy when screen processing is performed.

  In the present embodiment, an example in which the present invention is applied to an MFP (Multi Function Peripheral) will be described. The MFP is a composite image forming apparatus having a plurality of functions such as a copying function and a printing function.

FIG. 1 is a diagram illustrating a functional configuration of the MFP 100 according to the present embodiment.
The MFP 100 is connected to an external PC (personal computer) 200, and can print after generating image data from PDL (Page Description Language) format data transmitted from the external PC 200 and performing image processing. .

  As illustrated in FIG. 1, the MFP 100 includes a controller 20, an image processing unit 10, a control unit 11, a reading unit 12, an operation unit 13, a display unit 14, a storage unit 15, an image memory 16, and a printing device 17. ing.

The controller 20 generates image data for each pixel by rasterization processing.
For example, document data created in the external PC 200 is converted into PDL format by the printer driver software and transmitted to the controller 20, so that the controller 20 generates image data for each pixel by rasterization processing. In the rasterizing process, PDL commands are analyzed, and image data of each color of C (cyan), M (magenta), Y (yellow), and K (black) is generated for each image unit to be drawn (this is called an object). To do. That is, a pixel is assigned to an object to be drawn, and a data value is set for each assigned pixel.

Further, the controller 20 may perform anti-aliasing processing on the generated image data as necessary. For example, when text data or the like is input. In this case, the controller 20 generates a control signal indicating that the antialiasing process has been performed, and outputs the control signal to the image processing unit 10 along with the antialiased image.
In the present embodiment, the configuration in which the controller 20 is built in the MFP 100 has been described. However, the controller 20 may be provided outside the MFP 100.

  The control unit 11 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), and the like. The control unit 11 performs various calculations in cooperation with various processing programs stored in the storage unit 15, and the MFP 100. Central control of each part.

  The reading unit 12 includes a scanner having an optical system and a CCD (Charge Coupled Device), and generates an image signal (analog signal) by optically scanning the document. The generated image signal is subjected to various correction processes in a processing unit (not shown), then digitally converted and output to the image processing unit 10.

The operation unit 13 is used to input an operator's operation instruction, and includes various keys and a touch panel configured integrally with the display unit 14. The operation unit 13 generates an operation signal corresponding to the operation and outputs the operation signal to the control unit 11.
The display unit 14 displays an operation screen or the like on the display according to the control of the control unit 11.
The storage unit 15 stores parameters and setting data necessary for processing in addition to various processing programs.
The image memory 16 is a memory for storing image data.

The printing device 17 performs printing based on the printing image input from the image processing unit 10. The image for printing is an image generated by the image processing unit 10 performing necessary image processing on the image generated by the controller 20.
The printing device 17 performs electrophotographic printing, and includes, for example, a paper feeding unit, an exposure unit, a developing unit, and a fixing unit. At the time of printing, the exposure unit irradiates the photosensitive drum with laser light based on the image data to form an electrostatic latent image. Then, when the toner image is formed by developing by the developing unit, the toner image is transferred onto the paper fed from the paper feeding unit, and fixed on the paper by the fixing unit.

Next, the image processing unit 10 will be described with reference to FIG.
The image input from the controller 20 is temporarily stored in the image memory 16, and when there is a print instruction, the image is read from the image memory 16 and output to the printing device 17.
When the image processing unit 10 stores the image in the image memory 16, the image processing unit 10 performs compression processing on the image and performs resolution conversion to a low resolution. On the other hand, the image read from the image memory 16 is subjected to expansion processing and resolution conversion to restore the original resolution. Thereafter, image processing such as image rotation, reduction / enlargement, density correction processing, and screen processing is performed, and an image for printing is generated and output to the printing apparatus 17.

  FIG. 2 is a diagram showing an image compression conversion unit 1 and an image expansion conversion unit 2 which are components that mainly function during compression processing or expansion processing. The image compression conversion unit 1 and the image expansion conversion unit 2 are composed of an image processing circuit, a line memory for holding image data, and the like.

<Compression processing>
With reference to FIG. 3, the flow of processing of the image compression conversion unit 1 during compression processing will be described. Here, as shown in FIG. 4, a processed image generated for each color of CMYK, which is compressed from 1200 dpi of 1 pixel and 8 bits of data into data of 4 bits of 1 pixel and converted to 600 dpi, is generated. An example will be described. Since compression is performed in units of 8 × 8 pixels, FIG. 4 shows an original image of 8 × 8 pixels (1200 dpi) and a processed image corresponding to 4 × 4 pixels (600 dpi) corresponding to the area. Yes.

FIG. 5 shows the data structure of the processed image (600 dpi, 4 bits, 4 × 4 pixels) in FIG.
As shown in FIG. 5, the processed image is composed of 4-bit data per pixel, and the line memory and the image memory 16 used in the image compression conversion unit 1 and the image expansion conversion unit 2 have areas for holding the processed image. Will be formed. That is, four data layers (referred to as planes) corresponding to 4 × 4 pixels are formed with 1 bit per pixel.

As shown in FIG. 5, the 2-bit data value BTC (bij) of the processed image is held in the 0th to 1st bit data layers. This 0th to 1st bit data layer is called a BTC plane.
In the second bit data layer, the 1-bit data value flag (bij) of the identification flag is held. The identification flag is identification data used for identifying a decoding method. This second bit data layer is called an identification plane.

  In the data layer of the third bit, the maximum value and the minimum value (both 8 bits) in the 8 × 8 pixel block of the original image are held. In FIG. 5, the maximum value is represented by Max (k), and the minimum value is represented by min (k) (k represents a bit position, 0 ≦ k ≦ 7). The 8-bit data value is held at a position determined by k according to the bit position in 2 × 4 pixels determined for the maximum value and the minimum value. This third bit data layer is called a difference plane.

  When the original image to be processed is input from the controller 20, as shown in FIG. 3, the image compression conversion unit 1 determines whether or not a control signal indicating that the antialiasing processing is performed together with the image (step E1). . When only an image is input (step E1; N), the image compression conversion unit 1 performs compression processing by standard compression on the image (step E2). In contrast, when an antialiased control signal is input together with the image (step E1; Y), switching to the compression processing by binarized compression that is the compression processing for the antialiased image, The input image is compressed (step E3).

First, with reference to FIG. 6, a compression process using standard compression will be described.
As shown in FIG. 6, the image compression conversion unit 1 extracts and inputs an image in units of 8 × 8 pixel blocks from the input image (1200 dpi, 8 bits) (step S1). Then, the maximum value Max and the minimum value min are acquired from the data values of the pixels in the block (step S2). Since the data value of the pixel indicates the density value after printing, Max is the maximum density value in 8 × 8 pixels, and min is the minimum density value.

  In the example shown in FIG. 4, each pixel of 8 × 8 pixels of the original image (1200 dpi, 8 bits) is represented by aij (0 ≦ i ≦ 7, 0 ≦ j ≦ 7), and 4 × 4 pixels of the processed image (600 dpi, 4 bits). Each pixel is represented by bij (0 ≦ i ≦ 3, 0 ≦ j ≦ 3). Hereinafter, the data values of the pixels aij and bij may be indicated by aij and bij. In the original image, Max is the maximum value in aij and min is the minimum value. Max and min in the original image are also Max and min in the processed image.

  Next, quantization is performed. Different quantization methods are used depending on whether each pixel is a pixel in a halftone area or a pixel in a high resolution area. A halftone area is an image area that does not need to maintain a high resolution. For example, an image portion having a halftone density or a density between adjacent pixels that is not halftone (the density difference is small). This refers to the image part. The high resolution region refers to an image portion that needs to maintain high resolution, for example, an image portion such as an edge portion of an object, a thin line structure, or an isolated point. In the high resolution area, the resolution is more important than the gradation, whereas in the halftone area, the gradation is more important than the resolution. Since the required image quality varies depending on the characteristics of the image as described above, the image is divided into a high resolution area and a halftone area, and quantization is performed by different methods.

Therefore, the image compression conversion unit 1 determines whether the pixel is a halftone area pixel or a high resolution area pixel by determining whether the halftone condition or the high resolution condition is satisfied in the original image. To do. The determination is made in units of small areas obtained by subdividing the original image aij of 8 × 8 pixels into units of 2 × 2 pixels (corresponding to 1 pixel bij of the processed image). Further, thresholds THa1 to THa3 represented by the following formula are calculated and used for the determination.
THa1 = min + (Max-min) × 1/6
THa2 = min + (Max-min) × 3/6
THa3 = min + (Max-min) × 5/6

Then, it is determined whether or not any of the following halftone conditions (D11) to (D14) is satisfied in units of 2 × 2 pixels, and if it is satisfied, aij of 2 × 2 pixels is a pixel in the halftone area. to decide.
(D11) Among the four pixels aij, when at least one pixel satisfies THa1 <aij ≦ THa3 (D12) When all four pixels aij satisfy aij ≦ THa1 (D13) All four pixels aij When aij> THa3 is satisfied (D14) When Max-min <T (0 ≦ T ≦ 255) is satisfied
T is a threshold value set to determine whether the difference between Max and min, that is, whether the density difference in the 8 × 8 pixel region is small. For example, a value such as T = 30 can be set.
According to the halftone conditions (D11) to (D14), the aij area of 2 × 2 pixels has a halftone density, or the density values are all near the maximum value or the minimum value and have the same density. Or whether the change in density is small.

On the other hand, when the following high resolution condition (B1) is satisfied, it is determined that aij of 2 × 2 pixels is a pixel in the high resolution region.
(B1) In the four pixels aij, a pixel satisfying aij ≦ THa1 and a pixel satisfying aij> THa3 are mixed. That is, the density change in the aij area of 2 × 2 pixels due to the high resolution condition (B1) It can be determined whether or not is a high-resolution area.

  As shown in FIG. 6, first, it is determined whether or not (Max-min) <T is satisfied (step S3). If it is satisfied (step S3; Y), the halftone condition (D14) is satisfied. If satisfied, the process proceeds to halftone condition processing for performing quantization according to the halftone region (step S9). Even if (Max-min) <T is not satisfied (step S3; N), the thresholds THa1 to THa3 are calculated (step S4), and aij of 2 × 2 pixels is processed (step S5). Then, it is determined whether or not the halftone conditions (D11) to (D13) are satisfied (steps S6 to S8).

  For example, when attention is paid to a00, a01, a10, a11 in the original image in FIG. 4, if any one of a00, a01, a10, a11 satisfies THa1 <aij ≦ THa3 (step S6; Y), a00 , A01, a10, a11 all satisfy aij ≦ THa1 (step S7; Y), or all a00, a01, a10, a11 satisfy aij> THa3 (step S8; Y), the above halftone condition Since (D11) to (D13) are satisfied, the process proceeds to halftone condition processing (step S9).

  On the other hand, when none of the halftone conditions (D11) to (D13) is satisfied (step S6; N, S7; N, S8; N), that is, in the above example, aij ≦ THa1 in a00, a01, a10, a11 If there is a mixture satisfying the above condition and a condition satisfying aij> THa3, it is determined that the high resolution condition (B1) is satisfied, and the process proceeds to a high resolution condition process for performing quantization according to the high resolution region (step S10).

The halftone condition process will be described with reference to FIG.
As shown in FIG. 7, 2 × 2 pixel aij satisfying the halftone condition is quantized by the BTC method. First, an average value avr (bij) obtained by averaging the data values of aij of 2 × 2 pixels is calculated (step S91). In the examples of a00, a01, a10, and a11, avr (b00) = 1/4 (a00 + a01 + a10 + a11). Next, the average value avr (bij) of 8 bits using threshold values THa1, THa2, THa3
Are quantized into 2-bit data values BTC (bij) of 00, 01, 10, and 11. The quantized data value BTC (bij) is held in the BTC plane (step S92).

The quantization is performed according to the following conditions.
When min ≦ avr (bij) <THa1, BTC (bij) = 00
When THa1 ≦ avr (bij) <THa2, BTC (bij) = 01
When THa2 ≦ avr (bij) <THa3, BTC (bij) = 10
When THa3 ≦ avr (bij) ≦ Max, BTC (bij) = 11
That is, as shown in FIG. 8, four data values are quantized according to which of the density ranges defined by Max, min, and THa1 to THa3 by the BTC compression method. This 2-bit data value BTC (bij) becomes the data value of one pixel bij of the processed image.
In addition, after averaging, quantization is performed to perform resolution conversion together with quantization.

Next, the image compression conversion unit 1 sets the data value flag (bij) of the identification flag of bij of the processed image corresponding to 2ij pixels aij of the original image to flag (bij) = 0. Then, it is held at a position corresponding to bij of the identification plane (see FIG. 5) (step S93). Next, 1 bit of the bit position corresponding to bij is stored in the position corresponding to bij of the difference plane (see FIG. 5) (step S94).
For example, for the pixel of b00, the data value of 0 is held at the position of flag (b00) of the identification plane, and the data value of the 7th bit of min is held at the position corresponding to b00 of the difference plane.
When the above process is completed, the process proceeds to step S11 in FIG.

Next, the high resolution condition processing will be described with reference to FIG.
In the high resolution condition processing, the image compression conversion unit 1 creates a density pattern using aij data values of 2 × 2 pixels, and performs quantization according to the density pattern.
As shown in FIG. 9, the image compression conversion unit 1 binarizes the 2 × 2 pixel aij data value satisfying the high resolution condition into values of 0 and 1 according to the following conditions to create a density pattern (step S101). .
When aij> THa3, aij = 1
When aij ≦ THa1, aij = 0

When the high resolution condition is satisfied, the pixel with aij = 1 is close to the maximum value Max, and the pixel with aij = 0 is close to the minimum value min. The pattern is a density pattern obtained by patterning the density of the aij area of 2 × 2 pixels.
Next, the image compression conversion unit 1 quantizes four 2-bit data values BTC (bij) of 00, 01, 10, and 11 according to the created density pattern (step S102). Specifically, data values after quantization of 00, 01, 10, and 11 are assigned in advance for each density pattern, and converted into data values corresponding to the density pattern created for 2 × 2 pixels aij. Quantize.

Here, an example will be described in which density patterns are classified into four groups H0 to H3 as shown in FIG. 10 and quantized data values 00, 01, 10, and 11 are assigned to each group.
As shown in FIG. 10, the group of density pattern H0 is a case where only one aij = 1 is included in aij of 2 × 2 pixels. In this case, quantization is performed to BTC (bij) = 00.
The groups of density patterns H1 and H2 are both patterns that include two aij = 1, but are classified into density patterns H1 and H2 depending on where aij = 1 is located as shown in FIG. . If it corresponds to the density pattern H1, it is quantized to BTC (bij) = 01, and if it corresponds to the density pattern H2, it is quantized to BTC (bij) = 10.
The group of density patterns H3 is a pattern including three aij = 1 and is quantized to BTC (bij) = 11.

At the time of decoding, the density pattern is predicted from the quantized data value BTC (bij). As described above, the density of 2 × 2 pixels is the same (the number of 0s and 1s included in the density pattern is the same). By quantizing these density patterns as the same group, even if the prediction is wrong, it can be expressed with the same density within a small region of 2 × 2 pixels. Therefore, there is an effect that even if an error occurs, it is difficult to appear visually as image quality degradation.
It should be noted that the density patterns may be classified into groups based on the arrangement positions of 0 and 1 in the density pattern instead of the density, and the quantized data value may be assigned to each group.

When quantization is performed as described above, the image compression conversion unit 1 sets the data value flag (bij) of the identification flag of bij corresponding to aij of 2 × 2 pixels to flag (bij) = 1, and FIG. (Step S103). Further, 1 bit of the bit position corresponding to bij is held among the 8-bit data values of Max or min at the position corresponding to bij of the difference plane shown in FIG. 5 (step S104).
When the above process is completed, the process proceeds to step S11 in FIG.

The description will be continued from the processing of step S11 in FIG.
In step S <b> 11, a 4 × 4 pixel bij processed image (600 dpi, 4 bits) obtained through the halftone condition process or the high resolution condition process is output to the image memory 16.
Next, it is determined whether halftone condition processing or high resolution condition processing has been completed for all of the 8 × 8 pixel aij blocks of the original image (step S12). If the process is still in progress (step S12; N), the process returns to step S5, and the processes of steps S1 to S12 are repeated for the other unprocessed 2 × 2 pixel aij in the 8 × 8 pixel aij block. .

  On the other hand, if the processing has been completed for all the 8 × 8 pixel aij blocks (step S12; Y), it is determined whether the processing has been completed up to the end of the original image (step S13). S13; N), returning to the process of step S1, the processes of steps S1 to S13 are repeated for other unprocessed 8 × 8 pixel aij blocks in the original image. Then, when the process is finished up to the end of the original image (step S13; Y), this process is finished.

Next, with reference to FIG. 11, compression processing by binarized compression will be described.
When the image is subjected to anti-aliasing processing, the image region subjected to the anti-aliasing processing includes an area having a halftone density even though it is an edge portion of a character or the like. When compression processing by compression is applied, it may be determined that the region is a halftone region and averaged by halftone condition processing. In this case, the sharpness of the edge portion is lowered, and jaggy is generated by screen processing or the like. In order to avoid such a result, when a control signal indicating that antialiasing processing has been performed is input to the image compression conversion unit 1, the image compression conversion unit 1 performs compression processing by binarized compression compression described later for the antialiased image. Switch.

  As shown in FIG. 11, in the compression processing by binarized compression, the image compression conversion unit 1 extracts an image in units of 8 × 8 pixel blocks from the image (1200 dpi, 8 bits) input from the controller 20 and inputs it. (Step T1). Then, the maximum value Max and the minimum value min among the data values of the pixels in the block are acquired (step T2), and thresholds THa1, THa2, THa3 are calculated using the Max, min (step T3). Since the calculation method is the same as the method described in the standard degeneracy compression, the description is omitted here.

Next, the image compression conversion unit 1 performs quantization using aij of 8 × 8 pixels as a processing target, but different quantization methods are used for regions that satisfy the following conditions.
Each condition will be described.
When the following condition (D12) is satisfied for the 8 × 8 pixel processing region, the processing region is quantized by the same method as the quantization method for the halftone region.
(D12) The threshold value T satisfying (Max-min) ≦ T is a first threshold value for determining whether the density difference is small or large for the aij area of 8 × 8 pixels. The threshold T can be set as appropriate. For example, T = 30 can be set.

When the following condition (B2) is satisfied, the 2 × 2 pixel processing region is quantized by the same method as the quantization method for the high resolution region.
(B2) For a processing region of 8 × 8 pixels (Max-min)> T, and among the 4 × 2 pixels aij of 8 × 8 pixels, aij satisfying aij ≦ THa1, When there are pixels that satisfy> THa1

When the following condition (C) is satisfied, the 2 × 2 pixel processing region is quantized by the BTC method using the threshold value THa1 that is the second threshold value.
(C) For a processing region of 8 × 8 pixels, (Max-min)> T, and all of four pixels aij of 2 × 2 pixels out of aij of 8 × 8 pixels are aij ≦ THa1 or aij> THa1 If any of

  In the condition (C), another threshold value THa2 or threshold value THa3 may be used as the second threshold value. For the region that satisfies the condition (C), the second threshold value is used to quantize the binary value, and the decompressed data value is decoded into binary values of min or max. Can be designed to output easily. The edges of characters and the like also include fine lines, and if there are many pixels that are decoded to min at the edge part of the fine lines, there are cases where images such as characters are lost due to further thinning, so considering this case It is preferable to use a small threshold as the second threshold.

  That is, for the region satisfying the condition (B2) and the condition (C), quantization is performed so that the decompressed data value is binarized. Since anti-aliasing is applied to the edge of an object such as a character line drawing, it is considered that an image area including such an edge satisfies the above condition (B2) or the above condition (C). Therefore, the region satisfying the condition (B2) or the region satisfying the condition (C) is defined as an image region including an anti-aliased region, and the image data is subjected to quantization with a binary data value after expansion. By doing so, edge blurring and jaggy generation are prevented.

  More specifically, if an edge is included in an aij area of 8 × 8 pixels, the Max-min density difference increases, and this is determined by the threshold value T. Further, it is considered that the edge portion basically has a relatively large density difference between adjacent pixels, and the anti-aliased edge portion has a small density difference between adjacent pixels. Therefore, paying attention to the processing area of 2 × 2 pixels out of 8 × 8 pixels, and all the densities of 2 × 2 pixels in the processing area are higher than the threshold THa1 or lower than the threshold THa1, that is, the same level. In the region where pixels with the same density are adjacent, include the antialiased edge portion, and when pixels with higher density than the threshold THa1 and pixels with lower density than the threshold THa1 coexist, Since these are considered to be present, the quantization is performed on these areas so that the decompressed data value is binary.

  As shown in FIG. 11, it is determined whether or not (Max-min)> T is satisfied as shown in FIG. 11 (step T4). If not (ie, (Max-min) ≦ T) (step T4; N), halftone condition processing is executed for the 8 × 8 pixel aij in the same manner as the halftone region, and compression and resolution conversion are performed (step T5). The contents of the halftone condition process are the same as in the case of the standard degeneration compression described with reference to FIG.

On the other hand, if (Max-min)> T is satisfied (step T4; Y), it is determined whether or not aij ≦ THa1 or aij> THa1 is satisfied for all four pixels aij of 2 × 2 pixels (step T6). If satisfied (step T6; Y), the processing region of 2 × 2 pixels is set as an image region including the antialiased region, and aij of 2 × 2 pixels is converted into 00 using only one threshold THa1 according to the BTC method. , 11 are quantized into two 2-bit data values BTC (bij). Specifically, when all four pixels satisfy aij ≦ THa1, the data value BTC (bij) of one pixel corresponding to aij of four pixels is set to 00, and all four pixels satisfy aij> THa1 11 Thereby, quantization and resolution conversion are performed.
That is, as shown in FIG. 12, the region satisfying the condition (D12) is quantized into four values using three threshold values THa1 to THa3 as in the halftone region, whereas the condition (C) is satisfied. The region is quantized into a binary value using one threshold THa1. The quantized data value BTC (bij) is held at a position corresponding to bij of the BTC plane (see FIG. 5) (step T7).

Next, the image compression conversion unit 1 sets the identification flag flag (bij) to flag (bij) = 0 similarly to the case of the halftone region. This is because the identification flag flag (bij) is used to determine the decoding method according to the quantization method, and the region satisfying the condition (C) is decoded by the decoding method according to the BTC method similarly to the halftone region. It is. Here, such a configuration is used for convenience, but a configuration may be adopted in which a data value of an identification flag flag (bij) unique to an area satisfying the condition (C) is set and a decoding method corresponding to the data value is taken.
The data value of the set identification flag flag (bij) is held at a position corresponding to bij on the identification plane (see FIG. 5) (step T8).

  Next, the image compression conversion unit 1 holds 1 bit of the bit position corresponding to bij among the data values of Max or min at the position corresponding to bij of the difference plane (see FIG. 5) (step T9). When it ends, the process proceeds to step T11.

Next, in step T6, when all the 4 pixels aij of 2 × 2 pixels do not satisfy aij ≦ THa1 or aij> THa1, that is, pixels where aij ≦ THa1 and aij> THa1 are mixed (step T6; N) Will be described.
In this case, since the condition (B2) is satisfied, the image compression conversion unit 1 executes the high-resolution condition processing for aij of 2 × 2 pixels that satisfies the condition (B2), similarly to the high-resolution area, and performs compression and resolution conversion. Perform (step T10). The content of the high resolution condition process is the same as that in the case of the standard degeneration compression described with reference to FIG. After finishing the high resolution condition process, the process proceeds to step T11.

  In step T11, it is determined whether or not the processing has been completed for all aij of 8 × 8 pixels that are processing regions (step T11). If the process is in progress (step T11; N), the process returns to step T4, and the processes of steps T4 to T11 are repeated for the unprocessed area.

  If the process has been completed for all aij of 8 × 8 pixels (step T11; Y), it is determined whether the process has been completed up to the end of the original image (step T12). If the process has not been completed (step T12; N), Returning to step T1, the processes of steps T1 to T12 are repeated for the unprocessed area. When the process is completed up to the end of the original image (step T12; Y), this process ends.

<Extension processing>
Next, processing of the image extension conversion unit 2 will be described with reference to FIG.
As shown in FIG. 14, the image expansion conversion unit 2 decodes the processed image input from the image memory 16, converts each pixel 4 bits into 8 bits data, and expands the data. The decompression process is performed in units of 4 × 4 pixel (bij) blocks corresponding to 8 × 8 pixels (aij), which is a processing unit during compression. At the time of decoding, the resolution is converted from 600 dpi to 1200 dpi.

  As shown in FIG. 13, first, when a processed image is input in units of bij processing of 4 × 4 pixels (step P1), the image expansion conversion unit 2 calculates Max (k), min (k) from the difference plane of the processed image. Are obtained and arranged in the bit order to restore the data of Max and min (step P2). Next, in order to decode the original image for each pixel, attention is paid to one pixel bij in the processing region of 4 × 4 pixels (step P3).

  The image extension conversion unit 2 acquires the identification flag flag (bij) set in the target pixel bij from the identification plane. If flag (bij) = 0 (step P4; Y), halftone decoding processing is performed. Execute (Step P5). On the other hand, if flag (bij) = 1 (step P4; N), high-resolution decoding processing is executed (step P6).

First, the halftone decoding process will be described with reference to FIG.
As shown in FIG. 15, the image expansion conversion unit 2 decodes 2-bit BTC (bij) data by the BTC expansion method using the restored Max and min data, and expands the data to 8-bit data (step P51). . At the time of decoding, one pixel bij is divided into 2 × 2 pixels aij, resolution conversion is performed, and 8-bit data obtained by decompression is assigned to each pixel aij. That is, the data values of 2 × 2 pixels aij after decoding are all the same (step P52).

For example, when converting the resolution of one pixel of b00 into four pixels of a00, a01, a10, and a11, expansion is performed according to the following conditions.
When BTC (b00) = 00, a00 = a01 = a10 = a11 = min
When BTC (b00) = 01, a00 = a01 = a10 = a11 = min + (Max-min) × 1/3
When BTC (b00) = 10, a00 = a01 = a10 = a11 = min + (Max-min) × 2/3
When BTC (b00) = 11, a00 = a01 = a10 = a11 = Max
That is, as shown in FIG. 16, the density range determined by Max and min is decoded into density values that are equally divided into three.

  That is, a region satisfying the halftone conditions (D11) to (D14) in the standard degeneration compression and a region satisfying the condition (D12) in the binary degeneration compression are quantized into four values using a plurality of thresholds THa1 to THa3. Therefore, the decoded value is also four values. On the other hand, since the region satisfying the condition (C) in the binarized compression is quantized to a binary value of 00 or 11 using a single threshold THa1, the obtained decoded value is a binary value of min or Max. It is. When the decoded value obtained for bij is assigned to each aij of 2 × 2 pixels after resolution conversion, the process proceeds to step P7 in FIG.

Next, the high resolution decoding process will be described with reference to FIGS. 17 and 18.
For the high-resolution area that satisfies the high-resolution condition (B1) in the standard degeneration compression, the area that satisfies the condition (B2) in the binary degeneration compression is quantized using a density pattern in which binary values of 1 and 0 are arranged. Yes. At the time of quantization, since the values (00, 01, 10, 11) of BTC (bij) are assigned for each of the plurality of density patterns, when decoding them into the original 8-bit data, As shown in FIG. 19, several density patterns are conceivable depending on the data value of BTC (bij). In the high-resolution decoding process, decoding is performed while predicting what density pattern was created at the time of quantization.

The density pattern is predicted using a template.
FIG. 20 is a diagram illustrating a relationship between a template used when BTC (bij) = 00 and a density pattern predicted using the template. Each template is given an identification number (the number on the upper left of the template).

For C defined on each template, the pixel at the position of C satisfies the halftone condition, and the density difference between the C pixel and the target pixel bij | C _den -bij _Max | is | C _den -bij _Max | < _Tc indicates that the template is determined to match. C _den is a value when the quantized data (2 bits) of the pixel at the position C is decoded into an 8-bit data value under the conditions shown in FIG. That is, if the quantized data of the C pixel is BTC (bij) = 00, C _den = min, if BTC (bij) = 01, C _den = min + (Max-min) × 1/3, BTC (Bij ) = 10, C _den = min + (Max-min) × 2/3, and BTC (bij) = 11 is Max. bij _Max is the maximum density value Max in the processing region of 4 × 4 pixels to which the target pixel bij belongs.

M defined on each template is such that the pixel at the M position satisfies the high resolution condition and the density difference | M _Max -bij _Max | between the pixel at the M position and the target pixel bij is | M _Max -bij _Max | if the <T _M, have shown that it is determined that match the template. M _Max is the maximum density value Max in the processing region of 4 × 4 pixels to which M pixels belong. When the M pixel and the target pixel bij belong to the same processing region, M _Max = bij _Max = Max, so the density difference is zero.

Note that T _c and T _M are thresholds for determining whether the density difference is small and can be set as appropriate. For example, T _c = 30, T _M = 35, etc. can be set. . T _c and T _M may be different values or the same value. By comparing with T _c and T _M , a density pattern is predicted in which the density difference is small, that is, the pixel at the position C or M and the target pixel bij have the same density.

Similarly, FIGS. 21 and 22 are diagrams showing a relationship between a template used when BTC (bij) = 01 and a predicted density pattern, and FIGS. 23 and 24 show BTC (bij) = 10. FIG. 25 is a diagram showing a relationship between a template and a density pattern used when BTC (bij) = 11.
21 to 25, M1 indicates that the pixel at the position of M1 determines that the pixel matches the template when the M condition is satisfied and the density pattern H1 is satisfied. That is, the condition is that the pixel of M1 is BTC (bij) = 01.
M2 indicates that it is determined that the template matches when the pixel at the position of M2 satisfies the condition of M and corresponds to the density pattern H2. That is, the condition is that the pixel of M2 is BTC (bij) = 10.
Q indicates that it is determined that the pixel at the position of Q matches when any of the conditions C, M, M1, and M2 is not satisfied.

Each template is classified into three groups of X1, X2, and X3. This is because prediction is performed in three stages.
The template of the X1 group is a condition for determining that all the conditions such as C and M defined in the template match the template. On the other hand, the templates X2 and X3 are evaluated not to check whether all the matching conditions are satisfied, but to what extent the matching conditions are satisfied, and determine that they match based on the evaluation result. For example, in the case of the X2 group, collation is performed once for all the template groups of the X2 group, and the number of pixels satisfying the matching condition such as C and M is counted for each template, and this is used as the evaluation value. Then, it is determined that the obtained evaluation value matches the maximum template.

These templates are used for predicting the density pattern of the target pixel bij from the shape of the edge, the thin line structure, etc. included in the original image. Since the edge shape can be specified from the density pattern of the peripheral pixels of the target pixel bij, the template defines the conditions of the peripheral pixels when forming the edge shape as the above-described conditions such as C and M.
In particular, in the X1 group, a template is designed so that a density pattern can be predicted when the pixel of interest bij is a pixel with a thin line structure that needs to maintain a high resolution, and the X2 and X3 groups are widely used as looser conditions than X1. It is designed so that the edge shape can be predicted.

  For example, as shown in FIG. 26A, when an original image of a00 to a77 includes a one-dot-width oblique image, four pixels a44, a45, a54, and a55 are subjected to the high resolution condition ( Since B1) is satisfied and corresponds to the density pattern H1, the pixel b22 of the processed image corresponding to these four pixels should be quantized to BTC (b22) = 01. Then, at the time of decoding, from the density pattern of the peripheral pixels b13 and b31 (upper right and lower left of the target pixel b22), the dot of the b22 pixel is 1 dot wide and connected to the dots formed by the pixels b13 and b31. It can be predicted that the density of these dots is approximately the same. Therefore, in order to predict such a density pattern, it is only necessary to prepare a template 8 that defines the condition of M1 in the peripheral pixels as shown in FIG. 26A. The template 8 is the template 8 when BTC (bij)) = 01 shown in FIG.

  Further, as shown in FIG. 26B, in the case of an original image including an edge of an image having a certain density, a44, a45, a54, and a55 constituting this edge portion correspond to the density pattern H1. In order to predict the density pattern in such an edge shape at the time of decoding, as shown in FIG. 26B, a template 20 is prepared in which the C condition is defined in the peripheral pixels of the pixel b22 of the processed image corresponding to a44, a45, a54, and a55. do it. The template 20 (the template 20 when BTC (bij) = 01 shown in FIG. 21) is an X2 group template. In the original image, (a24, a25, a34, a35) corresponding to the pixel immediately above the target pixel b22 satisfies the high resolution condition (B1), and therefore the template 20 does not satisfy the condition C immediately above b22. However, (a22, a32, a23, a33), (a24, a34, a25, a35) and (a26, a36, a27, a37) corresponding to the left three pixels of b22 satisfy the condition of C. Become. The evaluation value will be high, and the possibility that it will be determined that it matches the template 20 should be large.

  In addition, in order to perform weighting evaluation, it is good also as setting a weighting coefficient in the template of X2 and X3 group. For example, in the case of the original image shown in FIG. 26B, if all three pixels located on the left side of the target pixel b22 satisfy the condition C, the target pixel b22 has a value of 1 for the two left pixels out of 2 × 2 pixels. There is a high possibility that it is a density pattern. Accordingly, when the matching condition C set for the three pixels located on the left side of the target pixel b22 of the template 20 is set with a weighting coefficient such as double, for example, and the condition C is satisfied at the three pixel positions. May be a value obtained by multiplying the evaluation value by a weighting coefficient. Thereby, a matching rate with a template can be adjusted.

  Note that the templates shown in FIGS. 20 to 25 are examples. What is necessary is just to design suitably according to the edge shape etc. which are considered to be included in the original image.

  As shown in FIG. 17, the image extension conversion unit 2 refers to BTC (bij) as shown in FIG. 17, and when BTC (bij) = 00 (step P61; Y), the process proceeds to a density pattern H0 prediction process. (Step P62). Similarly, if BTC (bij) = 01 (step P61; N, S62; Y), the process proceeds to the density pattern H1 prediction process (step P64). If BTC (bij) = 10 (step P61; N, P63; N, P65; Y), the process proceeds to the prediction process of the density pattern H2 (step P66). If BTC (bij) = 11 (step P61; N, P63; N, P65; N), the process proceeds to the prediction process of the density pattern H3 (step P67).

The process of predicting the density patterns H0 to H3 is basically the same except for the template used. Therefore, here, the prediction process of the density pattern H0 will be described as a representative with reference to FIG.
As shown in FIG. 18, the image extension conversion unit 2 collates with one of the templates of the X1 group with the pixel of interest bij at the center. When it is determined that the template matches the collated template (step P621; Y), the pixel of interest bij is decoded according to the predicted density pattern set in the matched template, and the decoded image, that is, 2 × 2 pixels The image of aij is output (step P628).

Decoding is performed by replacing the data value of 1 in the predicted density pattern with Max and replacing the data value of 0 with min. That is, decoding is performed by assigning binary decoded values Max and min corresponding to the predicted density pattern to the 2 × 2 pixel aij after resolution conversion, and performing resolution conversion together with decoding. When patterning into a density pattern in the compression process, the value close to Max is binarized to 1, and the value close to min is binarized to 0. Therefore, the pixel aij having a data value of 1 is Max and the data value of 0 is min It can be considered that the density can be restored to the same level as the original image even if it is replaced with.
For example, if BTC (bij) = 00 and the template 1 matches, the predicted density pattern is a density pattern in which the upper left pixel is 1 and the others are 0, as shown in FIG. . In this density pattern, a 2 × 2 pixel aij image in which the value of 1 is replaced with Max (8 bits) and the value of 0 is replaced with min (8 bits) is output as a decoded image (1200 dpi, 8 bits).

  If it does not match the collated template (step P621; N), it is determined whether or not collation with all templates in the X1 group has been completed (step P622). If all the collations have not been completed (step P622; N), the process returns to step P621, and collation with other templates belonging to the same X1 group is repeated until it matches any template in the X group.

  When all the templates in the X1 group have been collated but none of them matches (step P622; Y), all the templates belonging to the X2 group are collated, and an evaluation value is calculated (step P623). When the maximum value among the evaluation values calculated for each template exceeds 0 (step P624; Y), it is determined that the evaluation value matches the template of the maximum value (step P627). Then, the target pixel bij is decoded based on the predicted density pattern defined in the matched template, and the decoded image aij is output (step P628).

  On the other hand, if none of the conditions defined by the template of the X2 group is satisfied and the maximum value of the evaluation value is 0 (step P624; N), all the templates belonging to the X3 group are collated, and the evaluation value is Calculate (step P625). If the maximum evaluation value for each template exceeds 0 (step P626; Y), it is determined that the evaluation value matches the template with the maximum evaluation value (step P627), and the template is determined to match. Then, the target pixel bij is decoded based on the predicted density pattern, and the decoded image aij is output (step P628).

  For the X3 group, none of the conditions defined in the template is satisfied and the maximum value of the evaluation value is 0 (step P626; N), a pixel portion having a data value of 1 in aij of 2 × 2 pixels May form an image of isolated points. In this case, since it is difficult to predict the density pattern even with reference to surrounding pixels, a decoded image using the average pattern is constructed and output (step P629). The average pattern is obtained by assigning an average value to each pixel aij of 2 × 2 pixels as shown in FIG.

  For example, in the case of the density pattern H0, there is one pixel in aij of 2 × 2 pixels. That is, the maximum density value is output with four pixels. Therefore, the average pattern is obtained by assigning an average value 1 / 4Max to all aij of 2 × 2 pixels. Similarly, the density patterns H1 and H2 output 2Max with 4 pixels and 3Max with 4 pixels in the density pattern H3, so the average values 1 / 2Max and 3 / 4Max are 2 × 2 pixels aij. Assigned to.

When the decoded 2 × 2 pixel aij image is output as described above, the process proceeds to step P7 in FIG.
In step P7, it is determined whether or not the decoding and resolution conversion processing has been completed for all the 4 × 4 pixel bij processing regions of the processed image (step P7). If the process is still in progress, the process returns to step P3, and the processes of steps P3 to P7 are repeated for the other one pixel bij within the 4 × 4 pixel bij processing area.

  When the decoding and resolution conversion processing is completed for all of the 4 × 4 pixel bij processing region (step P7; Y), the image expansion conversion unit 2 sets each pixel bij in the 4 × 4 pixel bij processing region. The 8 × 8 pixel aij image output for is output (step P8). Next, it is determined whether or not the processing has been completed up to the end of the processed image (step P9). If the processing has not been completed (step P9; N), the process returns to step P1 to process the next 4 × 4 pixel bij of the processed image. The processes in steps P1 to P9 are repeated for the area. When the processing image is finished up to the end (step P9; Y), this processing ends.

27 and 28 show an example in which compression and expansion are performed by standard degeneration conversion.
FIG. 27 shows an image of a “reversible” character (black 100%), an image of slanted line 1 (black 100%, thin line of 1 dot width), a slanted line 2 (thick line; magenta 100%, 6 dot width, thin line; magenta 30). Comparative Examples 1 to 3 are shown together with Example 1 in which compression and decompression by standard degeneration conversion are performed on an image of (%, 2 dot width).
FIG. 28 shows Comparative Examples 1 to 3 for an image of the letter “g.” (4 colors of CMYK), an image of a human face photo 1 (yellow), and an image of a face photo 2 (4 colors of CMYK). The first embodiment to which compression by standard degeneration conversion is applied is shown.

The image processing methods in Comparative Examples 1 to 3 and Example 1 are as follows.
Comparative Example 1 An image rasterized at 600 dpi and 8 bits was copied to 1200 dpi by copying the data value of 1 pixel of 600 dpi to 4 pixels of 1200 dpi.
Comparative Example 2: An image rasterized at 1200 dpi and 8 bits is averaged to convert the resolution to 600 dpi (the value obtained by averaging the data values of four pixels of 1200 dpi is assigned to one pixel of 600 dpi), and then converted to the original 1200 dpi Resolution conversion (simply dividing the pixel into four and assigning the same value).
Comparative Example 3: Rasterized at 1200 dpi and 8 bits. The image quality of this comparative example 3 is the target image quality.
Example 1: An image rasterized at 1200 dpi and 8 bits is compressed to 600 dpi and 4 bits by the compression method of the standard conversion reduction described above, and the resolution is converted. Then, the image is expanded to 1200 dpi and 8 bits by the expansion method according to the above-described embodiment. The resolution was converted.

  In FIGS. 27 and 28, the right side of the first embodiment is shown in FIG. 27 and FIG. 28 in order to make it easy to understand the image portion determined as the halftone area and the high resolution area in the image according to the first embodiment. The pattern is formed by adding different patterns to the resolution area.

  As can be seen from FIGS. 27 and 28, in the method of Comparative Example 2 in which the resolution conversion is simply performed, data is lost in the process of compression and resolution conversion, so that the reproducibility of the edge portions of characters and line drawings is poor. The sharpness of the edge is lacking. As a result, the overall character is blurred and rough.

  On the other hand, in Example 1, even if it is a thin line with a width of 1 dot, although the reproducibility is slightly lacking in detail, the fine line in Comparative Example 3 rasterized at 1200 dpi and 8 bits has been successfully reproduced almost accurately. . In addition, the reproducibility of the edge portion is high, and the sharpness of characters and line drawings is almost the same as in Comparative Example 3.

On the other hand, FIG. 29 is a diagram showing an application example of compression and expansion by binarized compression.
Here, Comparative Examples 4 to 6 are shown together with Example 2 in which the compression and expansion of the above-described binary reduction compression is applied to images of the characters “W” and “k” (black 100%). . The image processing methods of Example 2 and Comparative Examples 4 to 6 are as follows.
Example 2: An image that has been rasterized at 1200 dpi and 8 bits and anti-aliased is compressed to 600 dpi and 4 bits by the above-described binary conversion compression method and converted to a resolution, and then the expansion method according to the present embodiment described above. The resolution was converted to 1200 dpi, 8 bits.
Comparative Example 4: Rasterized at 1200 dpi and 8 bits and anti-aliased.
Comparative Example 5: An image rasterized at 1200 dpi and 8 bits is subjected to compression and resolution conversion to 600 dpi and 4 bits by the standard conversion compression method described above without performing anti-aliasing processing, and then the expansion method according to the above-described embodiment Was expanded to 1200 dpi and 8 bits, and the resolution was converted.
Comparative Example 6: The image rasterized at 1200 dpi and 8 bits and anti-aliased is compressed to 600 dpi and 4 bits by the compression method of the standard conversion reduction described above, and the resolution is converted, and then is converted to 1200 dpi by the expansion method according to the present embodiment described above. Expanded to 8bit and converted resolution.

  As shown in Comparative Example 6 in FIG. 29, when compression and resolution conversion are performed on an anti-aliased image by standard compression, a rough image with a blurred edge portion is obtained. Compared to Comparative Example 5, even if the same character is reproduced, there is a difference in reproducibility after expansion depending on the presence or absence of anti-aliasing.

  On the other hand, when compression and resolution conversion are performed by binarized reduction conversion on an antialiased image as shown in the second embodiment, the same as in Comparative Example 5 in which standard reduction compression is applied to an image not subjected to antialiasing processing. Result. Thus, by reproducing the edge portion in a state where the anti-aliasing processing is not performed, it is possible to avoid a problem when the screen processing is performed, for example, a jaggy problem.

As described above, the difference between the standard reduction compression and the binarization reduction compression occurs in the anti-aliased image region because the edge portion includes pixels having a halftone density.
For example, consider an image A1 (min = 0, Max = 255) as shown in FIG. The image A1 is a diagram showing a part of an aij image area of 8 × 8 pixels including an edge portion. The numbers shown in each pixel are data values.
It is the image A2 that the image A1 has been anti-aliased. In the image A2, pixels of 40 or 140 halftone data values appear at the edge portion by anti-aliasing. When compression processing by binarized compression is performed on the image A2, the image A2 includes an edge portion composed of data values of min = 0 and Max = 255. Therefore, in the processing region of 8 × 8 pixels, (Max -min)> T. Here, if the threshold THa1 = 43 and attention is paid to a processing area of 2 × 2 pixels (indicated by a frame surrounded by a thick line), an area with all data values 0, an area with only 0 and 40, an area with only 140 and 255, All of the 255 areas satisfy the condition (C). On the other hand, the region having only the data values of 40 and 140 satisfies the condition (B2). As a result, the region satisfying the condition (C) is quantized to a binary value by the threshold THa1 = 48, and the region satisfying the condition (B2) is quantized by the binarized density pattern.

  In the decompression process, in the region satisfying the condition (C), if the quantized data is 00, all 2 × 2 pixels are min = 0, and if the quantized data is 11, all 2 × 2 pixels are Max = Decrypted to 255. In the region satisfying the condition (B2), 2 values of min or Max are assigned to 2 × 2 pixels by the prediction of the density pattern and decoded, and as a result, an image A31 is obtained.

  On the other hand, when standard degeneracy compression is applied to the same image A2, assuming that the thresholds THa1 = 43, THa2 = 128, and THa3 = 210, all are determined to be halftone areas as shown in FIG. Of the 2 × 2 pixel processing areas, all areas with data value 0, areas with only 0 and 40 satisfy the halftone condition (D12), areas with only 40 and 140 satisfy the halftone condition (D11), and all This is because the region of 255 satisfies the halftone condition (D14). For the halftone area, the average value of the data values is quantized into four values by three threshold values THa1 to THa3.

  In the decompression process, the halftone area is decoded into data values corresponding to the four values, and an image A32 shown in FIG. 31 is obtained. In the image A32, since the data value is the same for each processing region of 2 × 2 pixels, a region having a halftone density is generated in the edge portion, and as a result, the edge portion is blurred.

  As described above, according to the present embodiment, when the control signal indicating that the antialiasing process is performed together with the image to be compressed, the image compression conversion unit 1 is subjected to the antialiasing process from the compression process using the standard degeneration compression. Switch to compression processing by binarized compression for images. In the binarized compression, a region satisfying the condition (C) or the condition (B2) is set as an image region including an anti-aliased region, and a binary decoded value of min or Max is obtained for the image region. Quantize to The image extension conversion unit 2 decodes the image area satisfying the condition (C) into a data value of either min or Max by the BTC method, and the image area satisfying the condition (B2) is the data value of the original image. Is decoded into binary data of min or max by prediction of the density pattern created by binarizing the image, so that it becomes possible to binarize the decoded value for the image area including the anti-aliased area. . The binarization can suppress image quality deterioration such as blurring of the edge portion or jaggy when screen processing is performed.

  That is, an image is processed for each fixed region of 8 × 8 pixels, and a processing region whose density difference (Max-min) is larger than the threshold T is processed for each 2 × 2 pixels, and one obtained from Max and min. An area of 2 × 2 pixels adjacent to a pixel larger than the threshold THa1 or an area of 2 × 2 pixels adjacent to a pixel equal to or lower than one threshold THa1 is defined as an image area including an antialiased area. Quantization is made into binary values 00 and 11 by the BTC method using one threshold THa1. At the time of decoding, the image expansion conversion unit 2 outputs decoded values min and Max corresponding to the binary quanta 00 and 11. As a result, the image area including the area subjected to the anti-aliasing processing as described above can be binarized when decoded.

  Further, among the processing regions in which the density difference (Max-min) is larger than the threshold value T, processing is performed for each fixed region of 8 × 8 pixels, and pixels that are less than the threshold value THa1 and less than the threshold value THa1. As for the 2 × 2 pixel region in which the two are adjacent to each other, the density value is created by binarizing the data value of 2 × 2 pixels as in the high resolution region, and the quantized data corresponding to the classification of the density pattern Quantize to a value. At the time of decoding, the image expansion conversion unit 2 predicts the density pattern from the decoded values of the surrounding pixels, and outputs a binary decoded value corresponding to the density pattern. As a result, the image area including the area subjected to the anti-aliasing processing as described above can be binarized when decoded.

  Further, processing is performed for each fixed region of 8 × 8 pixels, and a processing region whose density difference (Max-min) is smaller than the threshold value T is quantized in the same manner as the halftone region. That is, the average value obtained by averaging the data values of 8 × 8 pixels and the three threshold values THa1 to THa3 obtained from Max and min are used to quantize the four values 00, 01, 10, and 11 by the BTC method. . At the time of decoding, the image expansion conversion unit 2 outputs decoded values corresponding to the four values 00, 01, 10, and 11. Thereby, it is possible to maintain gradation.

  Further, the image compression conversion unit 1 converts the resolution of the image to a lower resolution than before the quantization at the time of quantization, and the image expansion conversion unit 2 converts the resolution to the original resolution at the time of decoding. That is, in the region satisfying the condition (D12) and the condition (C), all 2 × 2 pixels aij are quantized to the same quantized data value, and the resolution is changed from 1 pixel bij to 2 × 2 pixels aij at the time of decoding. The same decoded value is assigned to aij of 2 × 2 pixels. On the other hand, for the region satisfying the condition (B2), the decoded values min and Max corresponding to the binary (0, 1) of the density pattern having the same resolution as the resolution of the original image are output. Thereby, the memory capacity for storing the compressed image can be reduced.

  Further, the image compression conversion unit 1 applies compression processing by standard degeneracy compression when no control signal indicating that antialiasing processing has been performed is input. In compression processing by standard degeneracy compression, processing is performed for each fixed area of 8 × 8 pixels, and when the four halftone conditions (D11) to (D14) are satisfied, it is determined as a halftone area and a plurality of threshold values THa1 to THa3 are set. Use to quantize. In addition, when the high resolution condition (B1) is satisfied, it is determined as a high resolution region, and quantization using a density pattern is performed. Thereby, it is possible to maintain the gradation in the halftone area and maintain the resolution in the high resolution area.

In addition, the said embodiment is a suitable example of this invention, and is not limited to this.
For example, in the above-described embodiment, for each pixel, resolution conversion is performed to a high resolution before quantization for all image areas at the time of decoding, but pixels in areas that satisfy halftone areas and conditions (A2) and (C) With regard to the above, it is possible to perform decoding while maintaining the low resolution without converting the resolution to the original high resolution at the time of decompression.

  The difference data is not limited to Max and min as long as it can be used for decoding. For example, data values of THa2 and min that are intermediate values may be held as difference data, and other data values Max, THa1, THa3, and the like necessary for decoding may be calculated from the data values of THa2 and min.

  In the above-described embodiment, the compression / expansion processing method has been described by using an example of an image generated from data input from the external PC 200. However, an image read by the reading unit 12 is input to the controller 20 and processed in the same manner. It is good to do.

In addition to the MFP, the present invention can be applied to a computer apparatus that performs image processing. Further, the above-described compression and decompression processing may be programmed and image processing by software may be performed using the program. In this case, as a computer-readable medium for the program, a non-volatile memory such as a ROM and a flash memory, and a portable recording medium such as a CD-ROM can be applied.
Further, a carrier wave (carrier wave) is also applied to the present invention as a medium for providing program data according to the present invention via a communication line.

FIG. 2 is a diagram illustrating a functional configuration of an MFP according to the present embodiment. It is a figure which shows the component which mainly functions at the time of the process of compression and expansion | extension among the image processing parts of FIG. It is a flowchart which shows the process at the time of compression. These are the original image and the processed image before and after the compression process. It is a data structure of a process image. It is a flowchart which shows the compression process by standard compression compression. It is a flowchart which shows a halftone condition process. It is a figure which shows the conditions of the quantization applied to a halftone area | region in the compression process of standard degeneration compression. It is a flowchart which shows a high resolution condition process. It is a figure which shows the correspondence of a density pattern and the quantum to quantize. It is a flowchart of the compression process by the binarization compression compression. It is a figure which shows the conditions of the quantization applied by the area | region in the compression process of binarization compression. It is a flowchart which shows an expansion | extension process. The processed image and the original image before and after the decompression process. It is a flowchart which shows a halftone decoding process. It is a figure which shows the conditions of the decoding applied to the halftone area | region and the image area | region by which the anti-aliasing process was carried out in the expansion | extension process. It is a flowchart which shows a high resolution decoding process. It is a flowchart which shows the prediction process of the density pattern H0. It is a figure which shows the density | concentration pattern estimated and decoding conditions. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the correspondence of the template used for prediction, and the density | concentration pattern estimated. It is a figure which shows the relationship between the template used for an original image and prediction. It is a figure which shows the relationship between the template used for an original image and prediction. It is a figure which shows the processing result of the compression and expansion | extension by standard compression compression. It is a figure which shows the processing result of the compression and expansion | extension by standard compression compression. It is a figure which shows the process result of the compression and expansion | extension by binary reduction compression. It is a figure which shows the processing content of a compression and expansion | extension by the binarization compression compression, and a process result with respect to the image by which the anti-aliasing process was carried out. It is a figure which shows the processing content of the compression and expansion | extension by standard compression compression, and a process result with respect to the image by which the anti-aliasing process was carried out.

Explanation of symbols

100 MFP
DESCRIPTION OF SYMBOLS 10 Image processing part 1 Image compression conversion part 2 Image expansion conversion part 11 Control part 12 Reading part 13 Operation part 14 Display part 15 Storage part 16 Image memory 17 Printing apparatus

Claims (22)

When a control signal indicating that the image has been anti-aliased is input together with the multi-value image to be compressed, the image is switched to compression processing for the anti-aliased image, and the image including the anti-aliased region of the image The area includes an image compression conversion unit that performs quantization by converting the data value of the decompressed image to binary ,
In the compression processing for the anti-aliased image, the image compression conversion unit uses a single threshold value for the image region subjected to the anti-aliasing among the image regions including the anti-aliased region. the image processing apparatus you quantized.   The image processing apparatus according to claim 1, wherein the image compression conversion unit converts the resolution of the image to a lower resolution than before the quantization when the image is quantized. In the compression processing for the anti-aliased image, the image compression conversion unit processes the image for each predetermined region, and among the processing regions in which the density difference is larger than the first threshold, pixels larger than the second threshold are adjacent. The image processing apparatus according to claim 2, wherein an area to be processed or an area adjacent to pixels smaller than the second threshold is used as an anti-aliased image area , and the image area is quantized by the BTC method using the second threshold.   The image compression conversion unit processes the image for each predetermined region in the compression processing for the anti-aliased image, and among the processing regions in which the density difference is larger than the first threshold, A region where pixels smaller than two thresholds are adjacently mixed is used as an image region including an anti-aliased region, and a density pattern is created by binarizing the data value of each pixel of the image region. The image processing apparatus according to claim 2, wherein the image processing apparatus performs quantization according to the classification.   The image compression conversion unit processes the image for each predetermined region in the compression processing for the anti-aliased image, and a processing region having a density difference smaller than the first threshold includes a plurality of thresholds including a second threshold. The image processing apparatus according to claim 2, wherein the image is quantized using a BTC method.   The said image compression conversion part performs the quantization using these threshold values with respect to the average value which averaged the data value which each pixel of the process area | region where the said density | concentration difference is smaller than a 1st threshold value has. Image processing apparatus. An image expansion conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 3,
In the original image before quantization, the image extension conversion unit is configured to process a region where a pixel having a density difference larger than the first threshold is adjacent to a region where pixels larger than the second threshold are adjacent or pixels smaller than the second threshold are adjacent. Is an image processing apparatus that performs decoding by the BTC method and outputs a binary decoded value.   In the original image before quantization, the image expansion conversion unit includes a region where pixels larger than the second threshold are adjacent, or a region where pixels smaller than the second threshold are adjacent among processing regions where the density difference is larger than the first threshold. The image processing apparatus according to claim 7, wherein resolution is converted to the same resolution as before quantization, and a decoded value is assigned to each pixel in the region after resolution conversion. An image expansion conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 4,
In the original image before quantization, the image expansion conversion unit is configured to process a region in which pixels larger than the second threshold and pixels smaller than the second threshold are adjacently mixed among the processing regions where the density difference is larger than the first threshold. Is an image processing apparatus that predicts a density pattern created at the time of quantization for the area and assigns a binary decoded value corresponding to the predicted density pattern to each pixel in the area after resolution conversion. An image expansion / conversion unit that decodes and expands an image quantized by the image processing apparatus according to claim 5,
The image expansion conversion unit is an image processing device that decodes a processing region having a density difference smaller than a first threshold in an original image before quantization using a BTC method.   The image expansion conversion unit decodes a processing area in which the density difference is smaller than a first threshold, performs resolution conversion to the same resolution as before quantization, and decodes each pixel in the processing area after resolution conversion. The image processing apparatus according to claim 10, wherein a value is assigned. When a control signal indicating that the image has been anti-aliased is input together with the multi-value image to be compressed, the image is switched to compression processing for the anti-aliased image, and the image including the anti-aliased region of the image the region, have rows quantization to binary data values of the image after decompression,
In the compression processing for an anti-aliased image, a compression method of quantizing the anti-aliased image region into a binary value using one threshold value among image regions including the anti-aliased region .   The compression method according to claim 12, wherein when the image is quantized, the resolution of the image is converted to a lower resolution than before the quantization. In the compression processing for the anti-aliased image, the image is processed for each predetermined region, and among the processing regions in which the density difference is larger than the first threshold, pixels larger than the second threshold are adjacent to each other or from the second threshold. 14. The compression method according to claim 13, wherein an area adjacent to a small pixel is used as an anti-aliased image area , and the image area is quantized by the BTC method using a second threshold.   In the compression processing for the anti-aliased image, the image is processed for each fixed region, and among the processing regions where the density difference is larger than the first threshold, there are pixels larger than the second threshold and pixels smaller than the second threshold. Create a density pattern that binarizes the data values of each pixel in the image area, using the adjacent mixed area as an image area including the anti-aliased area, and quantize according to the density pattern classification The compression method according to claim 13 or 14.   In the compression process for the anti-aliased image, the image is processed for each fixed region, and a processing region having a density difference smaller than the first threshold is quantized by a BTC method using a plurality of thresholds including the second threshold. The compression method according to any one of claims 13 to 15.   The compression method according to claim 16, wherein quantization using the plurality of threshold values is performed on an average value obtained by averaging data values of pixels in a processing region in which the density difference is smaller than a first threshold value. A decompression method for decoding and decompressing an image quantized by the compression method according to claim 14,
In the original image before quantization, the image extension conversion unit is configured to process a region where a pixel having a density difference larger than the first threshold is adjacent to a region where pixels larger than the second threshold are adjacent or pixels smaller than the second threshold are adjacent. Is a decompression method that performs decoding by the BTC method and outputs a binary decoded value.   In the original image before quantization, out of the processing areas where the density difference is larger than the first threshold, the area where pixels larger than the second threshold are adjacent or the area where pixels smaller than the second threshold are adjacent is decoded and resolution is increased. 19. The expansion method according to claim 18, wherein conversion is performed to obtain the same resolution as before quantization, and a decoded value is assigned to each pixel in the region after resolution conversion. A decompression method for decoding and decompressing an image quantized by the compression method according to claim 15,
In the original image before quantization, the image expansion conversion unit is configured to process a region in which pixels larger than the second threshold and pixels smaller than the second threshold are adjacently mixed among the processing regions where the density difference is larger than the first threshold. Is a decompression method that predicts a density pattern created at the time of quantization for the area and assigns a binary decoded value corresponding to the predicted density pattern to each pixel in the area after resolution conversion. A decompression method for decoding and decompressing an image quantized by the compression method according to claim 16,
The image decompression conversion unit is a decompression method that decodes a processing region having a density difference smaller than a first threshold in an original image before quantization by a BTC method.   23. The decoding of a processing region whose density difference is smaller than a first threshold, performing resolution conversion to the same resolution as before quantization, and assigning a decoding value to each pixel of the processing region after resolution conversion The described stretching method.