JPH08287221A - Processor and method for image processing - Google Patents

Abstract

PURPOSE: To reduce the memory capacity used for pixel density conversion processing by an image processor which changes the density of input image data and outputs the image data. CONSTITUTION: An input image 100 is inputted by a stripe unit, to a stripe/block conversion buffer 101. The stripe/block conversion buffer 101 divides the inputted stripe image into blocks under the control of a block division address control part 102 and transfers the divided blocks images to a block image buffer 103. For the respective block image, pixel density conversion processing is performed thereafter by an outline processing part 107 and an SPC processing part 106. Then the two pixel density conversion results are selectively outputted to a block/stripe conversion part 109 according to the area decision result of an area decision part 104. The block/stripe conversion buffer 109 reproduces the stripe images by combining the respective block images supplied by pixel unit.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and an image processing method, and more particularly to an image processing apparatus and an image processing method for changing the density of input image data and outputting it.

[0002]

2. Description of the Related Art Conventionally, an SPC (Selective P
The rocessing conversion method or projection method has been proposed and used. Here, the SPC method is a method of performing pixel density conversion by repeatedly dot-printing each pixel of the original image according to the scaling ratio of the image according to the scaling ratio or by periodically thinning out pixels. This SPC method is based on "Matsumoto, Kobayashi: A Study on Image Quality Evaluation in Facsimile Pixel Density Conversion, The Institute of Image Electronics Engineers, Vol. 12, No5, pp, 354-362,".
1983 ". The projection method is a method of projecting an original image on a converted image plane having different linear densities and binarizing an integral value over one pixel in this plane by threshold logic to determine the pixel value of the transformed image. is there. This projection method is based on "Arai, Yasuda: A study of facsimile line density conversion,
The Institute of Image Electronics Engineers of Japan, Vol.7, No.1, pp11-18, 1978 ”.

Further, as a pixel density conversion and smoothing technique of a binary image, a smoothing method by referring to a pattern of pixels around a pixel of interest is described in "Konaka et al .: Image quality improvement by smoothing processing of facsimile received image, Proposed in the Annual Meeting of the Institute of Image Electronics Engineers, No. 18, 1991, etc., and used for facsimile machines, electronic files and various peripheral devices.

However, in the above-mentioned pixel density conversion method, when a high-magnification conversion is performed, the curved portion of the image may be stepped (jagged), and the image quality may be deteriorated. Further, the pixel density conversion and smoothing technology by referring to the surrounding pixel pattern has a drawback that it can only deal with fixed magnifications such as 2 × and 4 × of the pixel of interest in the main scanning and sub scanning directions of the image.

Therefore, in order to solve the above problems, 2
Instead of processing the value image as it is in bitmap format,
The contour of the black image (when the binary image is composed of white or black) is extracted as vector data (hereinafter referred to as contour vector data), and the subsequent processing such as scaling and smoothing is performed on the contour vector data. The outline smoothing method for obtaining a high quality image is performed in Japanese Patent Application No. 3-34.
No. 5062. Hereinafter, the outline smoothing method is referred to as outline processing.

In the outline processing, the contour vector data is given by the coordinate data of the start point and the end point, and in order to perform the scaling processing accompanying the pixel density conversion, the coordinate value of the coordinate vector is enlarged according to the designated magnification.・ Do by reducing. Further, in smoothing, among the extracted contour vectors, the target vector to be smoothed is subjected to pattern matching with the length and direction of the neighboring contour vector as elements, and the target vector is based on the result. It is realized by redefining. The smoothing process includes processes such as removal of isolated points and notches, smoothing of jaggies (jagged patterns), and the like.

By the way, the outline processing as described above is extremely effective for a character or line drawing image, but is very effective for general images including a pseudo halftone image obtained by processing a photograph or a natural image. Couldn't expect.
This is because the minute pixels or fine lines representing the pseudo halftone image are smoothed by the outline processing, and the density value of the area is changed. Therefore, the above-mentioned outline processing is provided with means for discriminating the pseudo halftone image area from the character / line drawing image area in order to deal with this problem, and the conventional SPC processed image is selected for the pseudo halftone image area. However, it also has a method of selecting an outline processed image for a character / line drawing image. Hereinafter, such a method is referred to as pseudo-halftone compatible outline processing.

Further, in the above outline processing,
Since an extremely large number of contour vectors may be extracted, the memory required for processing them becomes enormous. Therefore, saving memory for outline processing,
Alternatively, a method of dividing an input image into stripes composed of a plurality of lines and performing outline processing in stripe units in order to speed up the processing speed by making the outline processing pipelined (hereinafter referred to as stripe processing)
Is taken. The image processed in stripe units is
After that, they are combined with each other and output again as a whole image.

In the striping process, since the smoothing process is applied to each of the striped images, an image that should not be originally smoothed (for example, a vertical line may extend over a plurality of stripes) is striped. There is a case where the image quality is deteriorated due to the smoothing processing in the part divided into (for example, a case where unevenness occurs where it should be a straight line). Therefore, in order to prevent the image quality deterioration due to such stripe processing, it is assumed that there are margins at the upper and lower edges of the images divided into stripes, and the process of deleting the margins is performed when combining the stripe images. (Hereinafter, it is called margin processing). As a result, the image of the divided portion subjected to the extra smoothing is deleted, and a good image without image quality deterioration can be obtained.

[0010]

By using the outline processing described above, it is possible to realize pixel density conversion of a relatively high quality image at any magnification from low magnification to high magnification. Further, it is possible to reduce the work memory used for the outline processing by the stripe processing.

However, the stripe processing in the above-mentioned conventional example requires a stripe width of at least 3 lines in consideration of the margin processing, and further considering the image quality, the stripe width for the margin processing is set to the upper and lower ends. It is desirable to secure at least 4 lines in total. Therefore, it is still necessary to extract a large amount of contour vectors when performing the outline processing, and to perform processing such as smoothing and binary image reproduction on the contour vector data. As a work memory for storing the extracted contour vector, a large capacity memory is naturally required, and an economic disadvantage is unavoidable.

Since it is difficult to incorporate a large-capacity memory in the image processing LSI, the memory is
You are forced to connect to the outside of SI. Therefore, the inevitable result is that the time spent for memory access prolongs the time for outline processing.

The present invention has been made in view of the above problems, and it is an object of the present invention to reduce the memory capacity used for pixel density conversion processing in an image processing apparatus that changes the density of input image data and outputs the image data. .

[0014]

[Means for Solving the Problems] and

In order to solve the above problems, an image processing apparatus according to the present invention is an image processing apparatus for changing the density of input image data and outputting the same, and an input means for inputting image data in stripe image units. A dividing unit that divides the input stripe image into block images of a predetermined size, a contour vector extracting unit that extracts a contour vector of the divided block image, and a contour vector of each extracted block image is smoothed and set. It is characterized by comprising a smoothing / magnifying unit for changing the magnification with a magnification ratio, and an image forming unit for synthesizing each block image subjected to the magnification process to form an output image, and converting the pixel density for each block image. Execute the process.

The image processing method according to the present invention is an image processing method of changing the density of input image data and outputting the image data, which comprises an input step of inputting image data in stripe image units, and an input stripe image. A dividing step of dividing the block image into a block image of a predetermined size, a contour vector extracting step of extracting a contour vector of the divided block image,
A smoothing / magnifying process of smoothing the extracted contour vector of each block image and scaling with a set scaling ratio; and an image forming process of combining the scaled block images to form an output image. The pixel density conversion processing is executed for each block image.

[0016]

【Example】

[First Embodiment] A preferred first embodiment according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a high image quality pixel density conversion apparatus using the outline smoothing method according to this embodiment. Input image 1
00 is an image in the form of a binary bit map, which is input line by line. Taking an image that is normally used in a G3 standard facsimile device as an example, the data size of the input image 100 is about 1728 × 1144 pixels in the case of an A4 image in 200 dpi standard reading, and that of an A4 image in 200 dpi fine reading. In this case, the size is about 1728 × 2288 pixels. Hereinafter, for convenience of description, the input image size is appropriately set to 17
28 × 1144 pixels or 1728 × 2288 pixels will be described as an example.

The input image 100 is input line by line,
In the case of the above example, 1728 pixels (bi
t). Input image 1 input line by line
In 00, a predetermined number of lines set in advance are stored in the stripe / block conversion buffer 101, and an image stripe division process (excluding the margin process) is performed. In the stripe division processing of the input image 100, the number of lines captured in the stripe / block conversion buffer 101 is counted using a counter (not shown) provided in the stripe / block conversion buffer 101, and the input / output of the image is performed according to the count value. It is realized by controlling.

Stripe / block conversion buffer 101
The stripe image stored in the block image is divided into block images and stored in the block image buffer 103 under the control of the block division address control unit 102.

FIG. 3 is a block division address control unit 102.
It is a figure which shows the address control by the typically. In FIG. 3, 300 is a stripe image, and 301 to 303 are block images obtained by dividing the stripe image 300 into a predetermined width. X is the number of pixels in the main scanning direction of the input image, and the value of X is 1728 according to the above example. Y is the stripe width of the input image, and it is necessary to set a predetermined value in advance. As a result of actual simulation, it is preferable that the value of Y is 8 or more in consideration of the image quality. Bx is a block size to be divided, and the smaller it is, the smaller the work memory required for the subsequent processing is. However, if it is too small, the boundary portion between the block images is processed.
Overall processing time increases. Therefore, the block size Bx needs to be determined by comprehensively judging the stripe width, the number of pixels on one line, the memory capacity, the required processing speed, and the like. Note that it is necessary to preset a predetermined value for Bx as well as the stripe width Y.

Stripe / block conversion buffer 101
The stripe image 300 stored in the
The data is sequentially read in pixel units from the position of T1 and stored in the block image buffer 103. The pixel reading direction is the main scanning direction, and when the pixel at the position of END1 is read and stored in the block image buffer 103, the operation is temporarily stopped. This is to match the processing speed with the subsequent outline processing and the like. Where END1
The position (address value) of is given by equation (1).

END1 = START1 + Bx−1 Equation (1) Once stopped, the storage of pixels in the block image buffer 103 starts again from the position of START2 of the next line and stops again at the position of END2. Where STAR
The position of T2 is given by equation (2).

START2 = END1 + (X−Bx) Equation (2) After that, these algorithms are repeated while changing the number of lines corresponding to the stripe width Y, so that the block image (301) is stored in the block image buffer 103. Through 30
3) is generated. Here, the positions (address values) of START and END on the n-th line of the block image 301 are given by equations (3) and (4).

STARTn = START1 + (n-1) X (Equation (3)) ENDn = STARTn + Bx-1 (Equation (4)) Further, n lines of the m-th block image from the left of the stripe image 300 The positions (address values) of START and END of the eyes are given by equations (5) and (6).

STARTnm = START1 + (n-1) X + (m-1) Bx ... Equation (5) ENDnm = STARTnm + Bx-1 ... Equation (6) The block division address control unit 102 is The storage of pixels in the block image buffer 103 is controlled using equations (5) and (6).

In this way, the block image 301 is supplied to the block image buffer 103, and after the subsequent pixel density conversion processing is completed and the block image buffer 103 becomes empty, a series of processing is performed on the next block image 301. Run. However, when the processing speed becomes a problem, the block image buffer 103 is set to the two-sided buffer format or the three-screen buffer format.
It is also possible to speed up the processing by adopting the surface buffer format and switching between the buffers.

By the way, when the number X of pixels of one line forming the stripe image 300 is not divisible by the block size Bx, the block image 303 at the right end has a block size smaller than Bx. In the example above,
If X = 1728 pixels and Bx = 10 pixels, a block image of 172 blocks can be obtained from the stripe image 300, and the block size of the final block is 8 pixels. In this case, the equation (5) can be applied as it is. However, using the formula (6), the rightmost block image 30
When calculating the address of 3, it is necessary to substitute 8 (pixels) for Bx (provided that 10 (pixels) is substituted for Bx of the equation (5) to calculate STARTnm).

As described above, the block images (301 to 303) are held in the block image buffer 103, and the pseudo halftone correspondence outline processing is applied to the block images. That is, the area determination unit 104 determines the area of the block image and holds the determination result in the area determination result storage buffer 105 on a pixel-by-pixel basis. Therefore,
The area determination result storage buffer 105 needs to have the same capacity as the block image buffer 103. There are two area determination results when the pixels forming the block image belong to the pseudo halftone image area or the pixels belonging to the character / line drawing image area.

As described above, by dividing the stripe image 300 and processing it in units of blocks, the area determination result storage buffer 105, for example, when the block size Bx is 10 bits (pixels), 80 bits (10 bits).
A storage capacity of × 8 lines = 80 bits) is sufficient. On the other hand, in the case of the conventional stripe processing, about 14 kbit (1
The area determination result storage buffer 105 of 728 bits × 8 lines = 13824 bits) is required.

Further, when the three-sided buffer format is used in order to increase the processing speed, it is conventionally about 42 kbit (1
The block image buffer 103 of 728 bits × 8 lines = 41472 bits was required, but according to the present embodiment, about 240 bits (10 bits × 8 lines × 3 = 240 bits) is sufficient from the same trial calculation. From the above, it is possible to dramatically reduce the capacity of the block image buffer 103 by dividing and processing the stripe image 300.

In the present embodiment, the area determination unit 104
Is arranged in the latter stage of the block image buffer 103, but it can be arranged between the stripe / block conversion buffer 101 and the block image buffer 103. In this case, the block image buffer 103
Since the number of accesses to F is reduced, further improvement in processing speed can be expected. As a matter of course, the area determination processing is executed in parallel when the stripe image 300 is divided into block images, and the area determination result is directly stored in the area determination result storage buffer 105.

Since the divided block image is subjected to pixel density conversion, the SPC processing unit 106 and the outline processing unit 10
7 is supplied. As described above, the SPC processing is a method of performing pixel density conversion by simply repeating dots to dot or periodically thinning out pixels.
On the other hand, the outline processing is a method of extracting the contour portion of the image as contour vector data and performing smoothing / magnifying processing based on the extracted contour vector data.

FIG. 2 is a block diagram showing the configuration of the outline processing unit 107. The outline processing unit 107 uses the block image output from the block image buffer 103 as an input image, and performs the following processing on the input image. That is, the contour vector extraction unit 200 extracts a contour vector from the input image and stores the extracted contour vector data in the contour vector buffer 201. The smooth scaling unit 202 performs smoothing / scaling processing on the contour vector data held in the contour vector buffer 201. The contour drawing unit 204 performs the contour drawing of the binary image as a pre-stage of reproducing the binary image using the smoothed / scaled contour vector data. The FIFO 203 is a FIFO (First In First Out) type memory for matching the processing speeds of the smooth scaling unit 202 and the contour drawing unit 204. The contour drawing buffer 205 is the contour drawing unit 20.
4 is a buffer for accumulating a binary image whose contour is drawn by 4. The filling unit 206 fills the internal area surrounded by the contour image based on the contour-drawn image,
A complete binary image is reproduced, and the result is reproduced by the pixel selection unit 108.
Output to.

The operation of the outline processing unit 107 will be described below. The contour vector extraction unit 200 extracts a contour line located at a boundary between a black pixel and a white pixel from the input block image as contour vector data. The contour vector data has a horizontal vector and a vertical vector as basic vectors, and is extracted in a positional relationship in which black pixels are viewed on the right side of the vector. Data extraction is performed by raster scanning a matrix window (for example, 3 × 3 matrix window) of a predetermined size, and contour coordinates located around the pixel of interest (center pixel of the 3 × 3 matrix window). And the connection destination information of the contour coordinates.

When the connection of the extracted contour coordinates extends to the image area where the window is not yet scanned, the connection information of the contour coordinates is temporarily provided in the connection undefined buffer (the contour vector extraction unit 200). ), And then used for searching the connection destination of the contour coordinates. Also, when the connection of the extracted contour coordinates crosses the boundary of the block image, the same processing as above is performed. The contour vector coordinate data extracted as described above is stored in the contour vector buffer 201.

The contour vector coordinate data has a feature that it always forms a closed loop for an arbitrary image. That is, the contour vector coordinate data output from the contour vector extraction unit 200 always returns to the starting point of the original contour vector data when the connection information is traced. The contour vector coordinate data group extracted from the block image is a set of a plurality of contour vector loops that do not overlap each other.

By the way, the contour vector buffer 201 is conventionally required to hold all the data extracted from the entire stripe image, and it is necessary to pipeline the contour vector extraction processing and the smoothing / magnifying processing. Since it was in the form, it had a very large memory capacity. Specifically, when the data width of the contour vector is 32 bits (including the coordinate value and the connection information), and the number of typical contour vectors extracted from one stripe image is 4000, about 256 kbit (32 b
It × 4000 × 2 = 256 kbit).

However, in this embodiment, since the block image is processed as the input image, the number of vectors extracted from the block image is reduced. For example, when the stripe image is divided into 173, the average number of vectors extracted from the block image is about 24 (4000
/173=23.1). Therefore, according to this embodiment, the contour vector buffer 201 is about 1.6 kbit.
It can be set to (32 bits × 24 × 2 = 1536 bits), and the memory capacity can be reduced to about 0.6% of the conventional memory capacity.

The smoothing / magnifying unit 202 is a processing unit that receives the contour vector coordinate data and executes the smoothing of the contour portion and the pixel density conversion.

The smoothing process can be further divided into a first smoothing process and a second smoothing process. The first smoothing process sets a target vector to be smoothed as a reference vector using a plurality of vectors to be connected before and after the target vector, based on the target vector and the length and direction of the reference vector. Matches with the selected pattern and redefines the vector of interest according to the result.
Of course, the redefinition of the vector pattern and the vector of interest is
It goes without saying that the contour portion of the binary image is defined so as to be smoothed.

The pixel density conversion process is executed on the contour vector coordinate data after the first smoothing process.
The method of pixel density conversion is the coordinate value of the contour vector data,
It is executed by performing a simple arithmetic process for multiplying the scaling factor.

On the other hand, the second smoothing process is executed on the contour vector coordinate data after scaling by a weighted average calculation of the coordinate values. The weighted average calculation is a general mathematical method, and the coordinate data subjected to this processing is redefined so that the coordinate values are regarded as weights and smoothed.

The contour vector coordinate data that has passed through the smoothing / magnifying section 202 is temporarily stored in the processing speed adjusting FIFO 203, and then supplied to the contour drawing section 204. The smoothed and scaled contour data is reproduced as a contour image of a binary bitmap by passing through the contour drawing unit 204. In contour drawing, the inclination of the contour vector to be drawn is obtained in advance from the start point and end point coordinates of the vector, and for example, a process is performed in which the pixels in the main scanning direction are sequentially shifted in pixel units and the dots are drawn (pixel drawing) at the displacement position in the sub scanning direction. Can be performed using a method for all contour vectors.

The contour drawing data is stored in the contour drawing buffer 205.
To be played on. Therefore, the contour drawing buffer 205 needs to have a capacity corresponding to the image size reproduced at one time, that is, the image size after the pixel density conversion of the block image. Conventionally, since processing is performed in stripe image units, assuming that the pixel density conversion magnification is 3 times in both the main scanning direction and the sub scanning direction and the stripe width is 8 lines (the margin width is 4 lines), the contour drawing buffer 205
Is 63 kbit (1728 bit × 3 × 4 lines × 3 =
It requires a capacity of 62208 bits).

However, if the stripe image is divided into block images and processed as in this embodiment, the block size B
If x is 10 pixels, then about 360 bits (10 bi
A contour drawing buffer 205 having a capacity of t × 3 × 4 lines × 3 = 360 bits) is sufficient.

2 reproduced on the contour drawing buffer 205
The value contour image outputs the final outline processed image by filling the inside of the contour with black pixels. The filling processing inside the contour is performed by the filling unit 206.

Specifically, while the contour image stored in the contour drawing buffer 205 is raster-scanned in the main scanning direction, the pixel values of two adjacent pixels (the pixel to be scanned and the pixel previously scanned) are exclusive. Performs a logical sum operation,
By replacing the pixel value of the scan target with the result, the inside of the contour image can be filled. Therefore, the filling unit 206 is composed of a latch circuit for holding one pixel value (pixel value of the previous scan) of two adjacent pixels, and an exclusive OR circuit.

For example, if white pixels (pixel values are 0) are adjacent to each other, the output is a white pixel, and as a result, pixel value replacement is not performed. When a black pixel (pixel value is 1) is scanned next to a white pixel, the output becomes a black pixel, and in this case also, the pixel value is not replaced. On the other hand, when a white pixel is scanned next to a black pixel, the output is a black pixel, so the white pixel is replaced by the black pixel,
Even if the next pixel is a white pixel, a black pixel is output again.
Therefore, when the black pixel is scanned after the white pixel, as long as the pixel scanned after that is the white pixel, the pixel is replaced by the black pixel, and the replacement by the black pixel is performed until the black pixel is scanned again. On the other hand, when there is only one black pixel in one line of the block image, the pixel value after replacement at the right end of the block image is held until the next block image is processed, so that matching between block images is performed. Take

This filling algorithm is used when a part of the contour drawing image is a horizontal line, and when one pixel is projected (dent).
In case of, it is not applicable and special processing is required for each. The pattern that is the target of the special processing is detected by the pattern matching of the contour shape, and the special pattern is applied. The filling unit 206 of this embodiment also includes these processing mechanisms.

The block image outline-processed as described above is output in the form of a block image and supplied to the pixel selection unit 108 as a smoothing-processed and image density conversion-processed image.

The block images subjected to the pixel density conversion processing by the SPC processing unit 106 and the outline processing unit 107 are supplied to the pixel selection unit 108 on a pixel-by-pixel basis. The pixel selection unit 108 refers to the result of the area determination performed by the area determination unit 104 in advance, and refers to the SPC processing unit 10
6 or the pixel by the outline processing unit 107 is selected and output. That is, if the region determination result is a pixel belonging to the pseudo halftone region, the output pixel from the SPC processing unit 106 is selected, while if it is a pixel belonging to the character / line drawing region, the outline processing unit 107 is selected.
Select the output pixel from.

The area determination result is stored in the area determination result storage buffer 105 for the entire block image and is referred to for each pixel. At this time, the output image of the SPC processing unit 106 or the outline processing unit 107 has been subjected to the pixel density conversion, while the determination result stored in the region determination result storage buffer 105 corresponds to the block image before the pixel density conversion. , The pixel densities of the target pixels on which the determination result is reflected are different. Therefore, when the area determination result is applied to the pixel selection unit 108, the pixel density of the determination result is also converted by using the SPC processing unit 112 to match the target pixel density. The SPC processing unit 112 is an SP
A configuration similar to that of the C processing unit 106 may be used.

The pixel selection processing unit 108 selectively outputs a pixel related to the output of the SPC processing unit 106 or the outline processing processing unit 107, and the pixel thus selected and output is the block / stripe conversion buffer 109.
Stored in. The block image processed in the block unit in this way is the block / stripe conversion buffer 1.
Sequentially stored in 09, the block images are combined and reproduced again as a stripe image.

The block image combination processing is performed along with the processing of storing the selected pixel in the block / stripe conversion buffer 109, and the control is performed by the block combination address control processing unit 110. That is, the block image after pixel density conversion, which is supplied to the block / stripe conversion buffer 109 on a pixel-by-pixel basis, is stored at the position (address) corresponding to the coordinate value of each pixel. In addition, block /
The memory area of the stripe conversion buffer 109 has a capacity corresponding to the size of the stripe image after pixel density conversion. Further, the same value is held in the block combination address control unit 110 as the block size Bx set when the stripe image is divided into blocks, and the block combination address control unit 110 determines the block size Bx.
Then, the block size after the pixel density conversion (block size on the block / stripe conversion buffer 109) can be calculated from the image scaling ratio.

The first pixel output from the pixel selection unit 108 is the first pixel of the block / stripe conversion buffer 109.
It is stored at the position of START1 on the line. Hereinafter, the output selected pixels are sequentially stored in the block / stripe conversion buffer 109. The pixel accumulation direction at this time is the main scanning direction of the stripe image. Then, the number of selected pixels corresponding to the block size after the pixel density conversion is stored, and when the pixel storage position reaches the position of END1, the storage in the block / stripe conversion buffer 109 is temporarily stopped. This is to adjust the processing speed with the outline processing unit 107 and the like. The important thing to note here is
START and END in block image combination processing
The relative position of is the same as START and END when the stripe image 300 is divided into block images, but their address values are different. This is because the block image combining process is performed on the image after the pixel density conversion.

Hereinafter, the block image 30 is obtained by repeatedly executing the same processing for the second and subsequent lines.
The block image after pixel density conversion corresponding to 1 is reproduced. The stripe image 300 is reproduced by repeating the same processing for the subsequent block images 302 and thereafter.

The block images processed in block image units as described above are combined again in the block / stripe conversion buffer 109 as a binary stripe image after pixel density conversion. Then, the reproduced stripe image is output according to the specifications of the image output 111. The image output device is, for example, a laser beam printer (LB).
P) and ink discharge type printers. In the case of LBP, image output is output line by line. On the other hand, in the case of an ink discharge type printer, since a plurality of lines are printed at once, the image output from the block / stripe conversion buffer 109 is to output a plurality of lines in parallel.

As typified by the block division address control unit 102 and the block combination address control unit 110, it is necessary to control the memory address in a centralized manner. This is a CPU having a memory storing a control program.
It may be realized by, or may be configured as hardware.

As described above, the stripe image is divided into block images having a smaller amount of data, and the pixel density conversion processing is executed in block image units,
The memory used for processing can be dramatically reduced.

In addition, accompanying this, it becomes easy to integrate the memory required for the pixel density conversion processing with the image processing unit to be built in one chip. As a side effect of this one-chip implementation, it is possible to prevent a decrease in processing speed due to access to the external memory, and it is possible to increase the overall processing speed. Second Embodiment In the present embodiment, the stripe image is divided into block images, the pixel density conversion processing is performed on the block image, and the processing until the stripe image is reproduced again is performed more simply and at high speed. It provides a method. Hereinafter, a second preferred embodiment according to the present invention will be described with reference to the drawings.

FIG. 4 is a block diagram showing the arrangement of a high image quality pixel density conversion apparatus using the outline smoothing method according to this embodiment. The blocks having the same functions as those in the first embodiment are designated by the same reference numerals. The input image 100 is stored in the stripe / block conversion buffer 401 line by line, and stripe image division is performed when outputting the image. A predetermined value is set in advance for the stripe width and the size of the block image to be divided. As described above, in stripe processing, it is preferable to secure a stripe width of about 8 lines in consideration of good image quality, and a block size of 10 pixels in consideration of image quality and the capacity of work memory. It is desirable to set it to a degree.

When a line image having a preset stripe width Y is input, a block image having a predetermined block size Bx is then transferred from the right end of the stripe image 300 to a parallel image transfer bus having a bit width corresponding to the stripe width Y. Four
It is output in parallel via 20 and is transferred and stored in the block image buffer 403. The parallel image transfer bus 4
20 comprises an 8-bit bus according to the above example.
The stripe / block conversion buffer 401 performs serial / parallel conversion on the stripe image 300 and sequentially outputs the block images in parallel from the right end of the stripe image 300.

Stripe / block conversion buffer 401
The block division address control unit 402 performs the timing of ending the line transfer of the image to and the series of timing control for dividing the image into a predetermined block size and outputting the images in parallel.

Stripe / block conversion buffer 40
1, the block division address control unit 402, and the block image buffer 403 are stripes / pixels in the first embodiment.
The block conversion buffer 101, the block division address control unit 102, and the block image buffer 103 respectively correspond, but their respective functions are different from those of the first embodiment. Each block image stored in the block image buffer 403 is
Outline processing unit 10 described in the first embodiment
7, through the pixel density conversion unit 400 such as the SPC processing unit 106 and the region determination unit 104, the pixel selection unit 108 outputs the data in pixel units, and sequentially to the block / stripe conversion buffer 409 under the control of the block combination address control unit 410. Is stored. Block / stripe conversion buffer 4
The block image stored in 09 in pixel units is combined with another block image, and the stripe image 300 is reproduced.

FIG. 5 is a diagram schematically showing the flow of parallel transfer of block images and reproduction of stripe images after pixel density conversion processing.

The input image 100 input line by line is stored in the stripe / block conversion buffer 401. The stripe image 300 is obtained by inputting an image for eight lines, is divided into block images (301 to 303), and the parallel image transfer bus 42 is sequentially arranged from the block image 303 at the right end of the stripe image 300.
It is transferred in parallel to the block image buffer 403 via 0.

After that, the same processing as that in the first embodiment is performed in the pixel density conversion unit 400, and then transferred to the block / stripe conversion buffer 409 in pixel units, and again in the block / stripe conversion buffer 409. 300a is reproduced. In the stripe / block conversion buffer 401, the divided block images (301 to 303) are sequentially transferred in parallel from the block image 303 at the right end of the stripe image 300. Therefore, in the block / stripe conversion buffer 409, the block image at the right end is transferred. Playback starts from 303a.
Note that reference numerals 300a, 301a to 303a indicate the stripe image and the block image after being processed by the pixel density conversion unit 400. As described above, by transferring the block images in parallel, the transfer speed of the block images can be increased, and as a result, the overall processing speed can be increased.

A block image buffer (not shown) for converting pixel-by-pixel serial image data into parallel image data is arranged between the pixel selection unit 108 and the block / stripe conversion buffer 409. To block / stripe conversion buffer 40
The transfer of the block images (301a to 303a) to the image processing apparatus 9 may be performed in the parallel data format.

Next, specific examples of block division and parallel conversion in the stripe / block conversion buffer 401 will be described with reference to FIGS. 6 to 8.

FIG. 6 is a diagram showing an example in which the stripe / block conversion buffer 401 is constructed using shift registers. The shift registers 601 to 608 are shift registers corresponding to the number of pixels for one line. Further, the number of shift registers (601 to 608) corresponds to the stripe width, and in the case of the above example, it may be set to eight. In this case, the shift register 601 corresponds to the first line, the shift register 602 corresponds to the second line, and the shift register 608 corresponds to the eighth line.

The shift clocks SC1 to SC8 supplied from the block division address controller 402 are input to the shift clock (SC) inputs of the shift clocks SC1 to SC8.

FIG. 7 is a timing chart showing the control of the shift clocks SC1 to SC8 when the input image 100 is input. The shift clocks SC1 to SC8 are
While inputting the input image 100, the image data is shifted by supplying the input image 100 corresponding to each line only during the input period. That is, during the input of the first line of the stripe image 300, only the shift clock SC1 is supplied, and the other shift clocks SC2 to S2.
By stopping C8, only the shift register 601 is operated. Further, while inputting the second line of the stripe image 300, only the shift clock SC2 is supplied,
Only the shift register 602 is operated by stopping the other shift clocks SC1 and SC3 to SC8. Similarly, for the third and subsequent lines, the stripe image 3 is supplied by supplying any one of SC3 to SC8 based on the input line.
Take in 00.

In FIG. 8, the input image 100 corresponding to the stripe image 300 is divided into blocks, and the pixel density conversion section 40 is divided.
6 is a timing chart showing control of shift clocks SC1 to SC8 when supplying 0 to 0. When supplying the input stripe image 300 to the block image buffer 403, the block size Bx of the block image to be divided
By simultaneously supplying a number of shift clocks SC1 to SC8 corresponding to the parallel image data via the parallel image transfer bus 420.
Supply to.

As described above, with the configuration using the shift register, block division and parallel conversion of image data can be realized by a simple method.

The block division and the parallel conversion of the image data are not limited to the configuration of the shift register described above, but can be configured using a general RAM. In this case, it is necessary to perform address conversion when writing and reading data, which is performed by the block division address control unit 4
Execute at 02.

The block division and the parallel conversion of the image data are provided with a write per bit function and a memory having a bit width corresponding to the stripe width, and the write operation is performed according to the line of the input stripe image 300. The stripe image 300 may be controlled to be written while controlling the per-bit data (masking bits other than the bits corresponding to the corresponding line). As a result, the stripe image 300 can be written in the memory on a pixel-by-pixel basis, and during a read operation, parallel image data having a bit width corresponding to the stripe width can be read out by a single access. Therefore, by reading out the number of times corresponding to the block size while giving predetermined address information to the memory, the reading (block division) in block image units can be performed at high speed.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. Further, it goes without saying that the present invention can be applied to the case where it is achieved by supplying a program to a system or an apparatus.

[0077]

As described above, according to the present invention,
It is possible to reduce the memory capacity used for the pixel density conversion processing.

[0078]

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a high image quality pixel density conversion apparatus using an outline smoothing method according to a first embodiment.

FIG. 2 is a block diagram showing an internal configuration of an outline processing unit 107.

FIG. 3 is a diagram schematically showing address control by a block division address control unit 102.

FIG. 4 is a block diagram showing a configuration of a high image quality pixel density conversion apparatus using an outline smoothing method according to a second embodiment.

FIG. 5 is a diagram schematically showing the flow of parallel transfer of block images and reproduction of stripe images after pixel density conversion processing.

FIG. 6 is a diagram showing an example in which a stripe / block conversion buffer 401 is configured using a shift register.

7 is a timing chart showing control of shift clocks SC1 to SC8 when an input image 100 is input. FIG.

FIG. 8 is a shift clock SC1 when the stripe image 300 is divided into blocks and supplied to the pixel density conversion unit 400.
8 is a timing chart showing control of SC8 to SC8.

Claims (8)

[Claims] 1. An image processing apparatus for changing the density of input image data and outputting the image data, the input means inputting image data in stripe image units, and dividing the input stripe image into block images of a predetermined size. Dividing means, contour vector extracting means for extracting the contour vector of the divided block image, smoothing / magnifying means for smoothing the extracted contour vector of each block image, and changing the magnification at a set magnification, An image forming device comprising: an image forming unit that forms an output image by synthesizing each block image that has been doubled. 2. The dividing means comprises: storage means for temporarily storing the input stripe image; and reading means for reading the stored stripe image in block units. Image processing device. 3. The image processing apparatus according to claim 2, wherein the reading means is a means for reading in parallel the line images forming the input stripe image. 4. A simple scaling means for simply scaling the contour vector of the extracted image of each block, and an area for determining whether or not the pixels forming the divided block image belong to a pseudo halftone area. Determination means, and each block image smoothed / scaled by the smoothing / scaling means based on the determination result of the area determination means, or each block image simply scaled by the simple scaling means The image processing apparatus according to any one of claims 1 to 3, further comprising: a selecting unit that selects and supplies the image forming unit to the image forming unit. 5. An image processing method for changing and outputting the density of input image data, comprising an input step of inputting image data in stripe image units, and dividing the input stripe image into block images of a predetermined size. The division process, the contour vector extraction process for extracting the contour vector of the divided block image, the smoothing / magnification process for smoothing the extracted contour vector of each block image, and scaling with the set scaling factor. And an image forming step of forming an output image by synthesizing each block image subjected to the double processing. 6. The dividing step comprises: a storing step of temporarily storing an input stripe image; and a reading step of reading the stored stripe image in block units. Image processing method. 7. The image processing method according to claim 6, wherein the reading step is means for reading in parallel the line images forming the input stripe image. 8. A simple scaling step of simply scaling the contour vector of the extracted image of each block, and an area for determining whether or not the pixels forming the divided block image belong to a pseudo halftone area. Determination step, each block image smoothed / scaled by the smoothing / scaling step based on the determination result of the area determination step, or each block image simply scaled by the simple scaling step The image processing method according to any one of claims 5 to 7, further comprising: a selection step of selecting and supplying the image to the image forming step.