JP2006313985A - Image processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus capable of minimizing an information amount required for reading of each block and decoding in reading compressed data in the unit of blocks of columnar orders each comprising M-rows and N-columns for rotating an image by 90 degrees and outputting the rotated image. <P>SOLUTION: The image processing apparatus creates compressed data with a data structure wherein a restart interval marker is inserted to just after a block of a final column of each row of blocks each comprising M-rows and N-columns and records the data to a flash ROM 30, and creates an information table wherein address information (at first, address information of restart interval marker) denoting a write address of compressed data of a concerned block associated with a block number of the block by one optional column and a cumulative value (0 at first) for calculating an absolute value of the compressed data of the concerned block are stored. When reading and decoding of the block by each column, the address information and the cumulative value are updated to attain reading and decoding of the block of a succeeding column. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus that can resize compressed data of an arbitrary size into an output image having a desired image size with a small memory capacity, and rotate the output image to output the image.

  In general, when the pixel arrangement main scanning direction of an image is different from the output main scanning direction by 90 degrees, it is necessary to rotate the image data by 90 degrees. When the image data is compressed by variable length coding represented by JPEG (Joint Photographic Experts Group) images, it is necessary to rotate the image 90 degrees after decompression, and a large-capacity memory for storing the developed image is required. Become.

  On the other hand, an image data compression / decompression apparatus that can extract and rotate a region of an image at high speed has been proposed (Patent Document 1).

In this image data compression / decompression apparatus, image data for n pixels × n pixels is divided into a plurality of rectangular blocks, each of which is a unit, and a compression code is generated for each rectangular block. The address in the compression code buffer is stored in association with the block number given to each rectangular block, and the address is read and read based on the specific block number stored in the compression code table. It has a circuit that reads and expands the compressed code stored in the compressed code buffer based on the specified address, and a circuit that outputs the decompressed data, so that only the necessary compressed code can be read and expanded. Processing is easy.
JP-A-8-317225

  However, the image data compression / decompression apparatus described in Patent Document 1 divides image data of input a × b pixels in a raster format into a plurality of rectangular blocks each having image data of n × n pixels as one unit, and the rectangle. Since the block position information indicating the position in the image data of the input image a × b is assigned to the block by the division / assignment means, if the input image is compressed data, it is temporarily expanded to image data of a × b pixels There is a problem that a large memory capacity is required.

  Further, the device described in Patent Document 1 needs to store all addresses in association with the block numbers given to the respective rectangular blocks.

  Furthermore, since the JPEG image data includes information (DC value of the frequency component) obtained by encoding the difference from the previous encoding unit (block) information, the DC value of each block needs to be stored and extracted. .

  The present invention has been made in view of such circumstances, and when reading compressed data in block order of M rows and N columns in order to rotate and output an image by 90 degrees, it is possible to read and decode each block. An object of the present invention is to provide an image processing apparatus capable of minimizing a necessary amount of information and capable of realizing image processing such as expansion / compression and image rotation using a small capacity memory.

  In order to achieve the above object, an image processing apparatus according to claim 1 divides image data for one frame divided into blocks of M rows and N columns from a block of 1 row and 1 column to a block of M rows and N columns. A compression means for creating compressed data having a data structure in which a restart interval marker is inserted immediately after a block in the last column of each row, and one frame compressed by the compression means; A memory for storing the compressed data of the block, position information indicating the writing position of the compressed data of the block in association with the block number of the block for an arbitrary column, and an absolute value from the first value of the compressed data of the block An information table for storing an accumulated value for calculation, and a lister inserted immediately after the last column block of each row in association with the block number of the first column block Information table creation means for writing position information indicating the write position of the interval marker and an accumulated value of 0 to the information table and updating the position information and accumulated value; and position information recorded in the information table Reading means for reading the compressed data of the blocks in each column based on the above, decoding means for decoding the compressed data of the read blocks using the accumulated value stored in the information table, and output of a desired image size Buffer means having a capacity smaller than the data amount of the image, and writing means for changing the output order of the image data in the block of n × n pixels decoded by the decoding means and writing to the buffer means, The information table creation means reads the compressed data of the i-th column (1 ≦ i ≦ N) block of each row by the reading means, and the decoding When the decoding by the stage is completed, the position information used for reading the compressed data of the block in the i-th column in each row of the information table is updated to the position information indicating the writing position of the compressed data in the block in the i + 1-th column in each row. At the same time, the accumulated value recorded in the information table is updated to the accumulated value of the first value of the compressed data of the blocks from the first column to the i-th column of each row and rotated by 90 degrees from the buffer means. It is characterized by enabling output.

  First, compressed data having a data structure in which a restart interval marker is inserted immediately after the last column block of each row of M rows and N columns is created, so the number of restart interval markers is minimized (M rows). The amount of compressed data for one frame stored in the memory can be prevented from becoming large. In normal JPEG compressed data, a restart interval marker is inserted every several MCUs (minimum encoding unit).

  When the compressed data of the block in the first column is read from the memory in order to rotate the output image by 90 degrees, the first data is written based on the position information indicating the writing position of the restart interval marker written in the information table first. Read the compressed data of the block in the first column. Since the blocks in the first column are blocks immediately after the restart interval marker, the DC value at the head of the compressed data is an absolute value.

  Therefore, the block in the first column can be decoded without requiring the first DC value of other blocks. When the compressed data of the block in the first column is read, position information (address, bit position) indicating the first writing position of the block in the second column can be detected. The position information of the start interval marker is updated with the detected position information.

  Further, the DC value of the block in the first column is recorded in the information table as the accumulated value of the DC value up to the block immediately before the block in the second column (the accumulated value in which the value of 0 is recorded is the first value). Update to the DC value of the block in the first column). Thereby, the block of the second column can be read based on the updated information table, and the DC value of the block of the second column (difference value from the DC value of the immediately preceding block) and the information table Is added to the updated accumulated value, the absolute value of the DC value of the block in the second column can be obtained, and the block in the second column can be decoded. Further, after reading / decoding the block in the second column, the position information and the accumulated value in the information table are updated for the block in the third column. By repeating this up to M columns, blocks can be read and decoded in order from the first column to M columns, and image data rotated by 90 degrees can be output.

  According to a second aspect of the present invention, in the image processing apparatus according to the first aspect, the interface means for continuously receiving the compressed data of the image data of an arbitrary size for one frame, and the compressed data in units of blocks of n × n pixels. Decompression means for decompressing, decompression data buffer means for temporarily storing image data of n × n pixels decompressed by the decompression means, and image data of n × n pixels stored in the decompression data buffer means to an arbitrary size Resolution conversion means for performing conversion processing and outputting to the buffer means, and the compression means sequentially holds image data in units that can be processed in the buffer means in units of blocks of n × n pixels. It is characterized by creating compressed data having a compressed data structure.

  That is, compressed data of image data of an arbitrary size is continuously received from the interface means for one frame, and the compressed data is decompressed while receiving the compressed data, resolution conversion, and compression processing of the image data after resolution conversion. To save to memory. Here, at the time of resolution conversion, the received compressed data is expanded in units of n × n pixels, and the expanded image data of n × n pixels is temporarily stored in the expanded data buffer means. The n × n pixel image data in units of blocks held in the decompressed data buffer means is converted into a resolution of an arbitrary magnification by the resolution conversion means and temporarily held in the resolution conversion output buffer means. Since expansion and resolution conversion are performed in units of n × n pixel blocks, a large memory capacity is not required as a buffer unit. When the resolution-converted image data is stored in the buffer means as image data in a unit that can be processed, it is compressed in block units of n × n pixels.

  According to a third aspect of the present invention, in the image processing apparatus according to the second aspect, the resolution conversion means is used for the interpolation calculation in accordance with a magnification a / b (a, b: natural number) for reducing or enlarging the image size. An interpolation table in which interpolation coefficients are stored, the interpolation table having a combination of interpolation coefficients corresponding to the number of numerators, where the number of pixels of the denominator b of the magnification a / b is one cycle, and an image of the image data for one frame A magnification a / b for reducing or enlarging the image size is calculated based on the size and the image size of the desired output image, and an interpolation coefficient is stored in the interpolation table based on the calculated magnification a / b, or plural A table setting means for setting an interpolation table corresponding to the calculated magnification a / b from the interpolation table as an interpolation table to be used for the interpolation calculation, and the decompressed data buffer means The image data of 憶 been n × n pixels by interpolation operation on the basis of the interpolation coefficient of the interpolation table is characterized in that the resolution conversion. Thereby, resolution conversion can be performed in units of blocks of n × n pixels, and a memory for holding all image data for one frame at the time of resolution conversion is not required.

  4. The image processing apparatus according to claim 3, wherein the resolution conversion means includes a plurality of blocks of i rows and j columns obtained by dividing the image data of an arbitrary size into blocks of M rows and N columns. At the time of horizontal resolution conversion of the image data of n × n pixels of the block of (1 ≦ i ≦ M, 1 ≦ j ≦ N), the image data of the n columns of the block of i rows and j columns and the image data of the n columns Is used for the resolution conversion of the image data of the block of i row j + 1 column by holding the position of one cycle of the interpolation table corresponding to 1 to the resolution conversion of the image data of the block of i row j + 1 column. At the time of resolution conversion, the image data of n rows of the block of i rows and j columns and the position in one cycle of the interpolation table corresponding to the image data of n rows are held until the resolution conversion of the image data of the blocks of i + 1 rows and j columns. I + 1 It is characterized by using the resolution conversion of the image data blocks of the j-th column. Thereby, even if resolution conversion is performed in units of n × n pixel blocks, it is possible to perform resolution conversion with continuity in the horizontal and vertical directions of an image for one frame.

  According to the present invention, when the compressed data in units of blocks of M rows and N columns is read in the column order in order to output the image after being rotated by 90 degrees, the amount of information necessary for reading and decoding each block is minimized. be able to. In addition, a minimum memory capacity (ASIC (application-specific integrated circuit) is required for performing image processing such as decompression, resolution conversion (resizing), compression, and 90-degree rotation of compressed data of arbitrary size image data received continuously. ) With a small memory capacity).

Hereinafter, preferred embodiments of an image processing apparatus according to the present invention will be described in detail with reference to the accompanying drawings.
[Overall configuration of image processing apparatus]
FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention.

  This image processing apparatus performs image processing on image data input from the outside and outputs it to a printing head of a printer. The image processing apparatus includes an ASIC 10 and a flash ROM 30, and print data image-processed by the ASIC 10 is printed on the printing head 32. Is output.

  The ASIC 10 mainly includes an interface unit 12, a central processing unit (CPU) 14, a work memory 16, a JPEG expansion block 18, an expansion data buffer 20, an H resize buffer 22, a V resize buffer 24, and a resize block 26. And a print control unit 28.

  The interface unit 12 may be a wireless or wired interface such as IrDA, Bluetooth, wireless LAN, UWB (Ultra Wide Band), USB (Universal Serial Bus), wireless USB, IEEE 1394, SCSI, and RS232C.

  The CPU 14 performs overall control of each circuit, and performs analysis of data and images, setting of necessary information based on the analysis result to each circuit, JPEG compression processing by software, data read / write control, and the like.

  The work memory 16 includes a memory having a memory capacity smaller than the uncompressed data amount of an output image having a desired image size (in this embodiment, VGA (image size of 640 × 480 pixels)). And a storage area for temporarily storing a reduced image described later.

  The JPEG decompression block 18 includes a Huffman decoding unit, an inverse quantization unit (IQ unit), and an inverse discrete cosine transform unit (IDCT unit), and the main scanning / sub-scanning direction resolution and the number of color components sent from the CPU 14. Based on each color sampling factor (SF), each color quantization table (IQNT), and each color Huffman code definition table (DHT), JPEG-compressed compressed data is expanded in a minimum coding unit (MCU). That is, the input compressed data is Huffman decoded to generate a quantized DCT coefficient, and the generated quantized DCT coefficient is inversely quantized by an inverse quantization unit to generate a DCT coefficient. The DCT coefficient thus obtained is subjected to IDCT processing in an IDCT unit to generate image data.

  The decompression data buffer 20 is composed of a double buffer for each color having a data capacity of image data for one block of n × n pixels (8 × 8 pixels in this embodiment). Stores the decompressed data. The 8 × 8 pixels stored in the decompressed data buffer 20 are transferred to the resize block 26.

  The H resize buffer 22 is a buffer that temporarily holds image data obtained by converting the resolution of 8 × 8 pixel image data in the horizontal direction (H direction), and the V resize buffer 24 is provided in the H direction and the vertical direction ( The image data resized in the V direction) is temporarily stored, and the output image for printing (image data for a plurality of lines) is temporarily stored. The memory capacity is smaller than the uncompressed data amount of the output image. It consists of memory. In this embodiment, the V resize buffer 24 can hold image data for 640 W × 32 lines for each color.

  The resizing block 26 performs resizing in the H direction and the V direction in units of 8 × 8 pixel blocks transferred from the decompressed data buffer 20. That is, the data held in the decompressed data buffer 20 is transferred to the resize block 26, where the resize in the H direction is first performed. The data resized in the H direction is temporarily held in the H resize buffer 22. When data of 8 dots or more is accumulated in the H resize buffer 22 in the H direction, it is transferred again to the resize block 26 in units of 8 dots in the H direction. The transferred data is resized in the V direction using the same resize circuit as in the H direction, and then transferred to the V resize buffer 24. Note that the resizing in the H direction and the V direction may be performed by separate resizing circuits, and the sequential processing may be performed.

  The print control unit 28 has a 3D-LUT (three-dimensional lookup table) and inputs YCbCr color system luminance data Y and color difference data Cb and Cr for each column in the sub-scanning direction of the printer from the V resize buffer 24. Then, the luminance data Y and the color difference data Cb and Cr are converted into RGB color system R, G and B data by a 3D-LUT, and image processing is performed using image processing parameters created by the CPU 14. The processed R, G, B data is sent to the print head 32.

On the other hand, the flash ROM 30 has a free capacity capable of storing a microcomputer program and storing compressed data of an output image having a desired image size (VGA) for one frame.
[Print section]
FIG. 2 is a perspective view showing a print medium (instant film) 50 on which an image is printed by the print head 32.

  As shown in FIG. 2, the instant film 50 is located above the photosensitive sheet 51, the image receiving sheet 52 opposite to the photosensitive sheet 51, and the photosensitive sheet 51 and the image receiving sheet 52, and encloses the developer. The developer pod 53 and a trap portion 54 that is located below the photosensitive sheet 51 and the image receiving sheet 52 and absorbs the excess developer. Then, after a latent image is formed photochemically by exposing the photosensitive sheet 51, the photosensitive sheet 51 and the image receiving sheet 52 are superposed, and both sheets are pressurized while developing a developing solution therebetween. As a result, the positive image is transferred to the image receiving sheet 52.

  FIG. 3 is a perspective view showing a main part of the printing unit.

  In the drawing, a print head 32 is disposed in the print unit, and a pair of developing rollers 60 are disposed on the upstream side of the print head 32. Further, the film pack 56 capable of accommodating a plurality of the instant films 50 is set at a predetermined position of the print unit. As shown in FIG. 3, the print head 32 has a longitudinal direction (main scanning direction) perpendicular to the transport direction of the instant film 50 (sub-scanning direction indicated by the arrow A). It is fixedly arranged near the upper end of the exposure aperture.

  At the time of printing on the instant film 50, the uppermost instant film 50 in the film pack 56 is fed at a constant speed in the sub-scanning direction (direction of arrow A) by a feeding mechanism (not shown), and one row is printed by the printing head 32. R, G, B light exposure is performed every time.

When the instant film 50 is fed out by a certain amount, the tip of the instant film 50 (tip on the image liquid pod 53 side) is sandwiched between the pair of developing rollers 60. The instant film 50 is pulled out at a constant speed while an appropriate pressing force is applied by the developing roller 60. As a result, the developer in the image liquid pod 53 is spread uniformly.

  FIG. 4 is a sectional view of the printing head 32. As shown in the figure, the printing head 32 is mainly composed of R, G, and B light emitting diodes 33R, 33G, and 33B, a light guide 34, a reflector 35, a liquid crystal shutter 36, and a minute lens array 37. It is configured.

  The R, G, and B light emitting diodes 33R, 33G, and 33B are controlled to be turned on in order when the instant film 50 is exposed to one line of image light. The R, G, or B light emitted from each light emitting diode is guided to the light guide 34 extending in the main scanning direction, and is incident on the liquid crystal shutter 36 by the reflector 35.

  The liquid crystal shutter 36 is configured by arranging minute liquid crystal segments corresponding to pixels, and the transmittance of each liquid crystal segment is controlled in accordance with one line of R, G, and B image data. In other words, the liquid crystal segments of the liquid crystal shutter 36 control the transmittance in a line-sequential manner according to the sequential light of R, G, and B, and control the transmittance for each pixel three times per line. The light emitted from the liquid crystal shutter 36 is imaged on the exposure surface of the instant film 37 via the micro lens array 37.

The R, G, B data output from the print control unit 28 to the print head 32 is used to control the transmittance of the liquid crystal shutter 36 of the print head 32 for each of R, G, B.
[Image processing]
The image processing apparatus having the above configuration includes data reception for receiving compressed data (JPEG data) of arbitrary size image data from the interface unit 10, data processing for analyzing the received image and generating image processing parameters, Based on the image processing parameters, the image processing corresponding to the time of printing that processes the output image and outputs it to the print head 32 is sequentially performed.
<< A. When receiving data
(1) Header Information Analysis FIG. 5 is a flowchart showing the CPU operation at the time of header information analysis by the image processing apparatus.

In FIG. 5, compressed data (JPEG data) of image data of an arbitrary size is continuously received for one frame via the interface unit 10 (see FIG. 1) (step S10).
<Data structure of JPEG data>
Here, the data structure of JPEG data will be described. As shown in FIG. 9, the header portion at the beginning of JPEG data includes main scanning / sub-scanning direction resolution (image size), number of color components, each color sampling factor (SF), each color quantization table (IQNT), and each color Huffman. A code definition table (DHT) is recorded, and then the image data of the main body of JPEG data is compressed and sequentially recorded for each minimum coding unit (MCU).
That is, as shown in FIG. 10A, image data for one frame is scanned every MCU from the upper left to the lower right of one frame, and is compressed and recorded sequentially for each MCU.

  As shown in FIG. 10 (B), the MCU uses 16 × 16 pixels as the minimum unit for encoding and the size of the discrete cosine transform (DCT) is 8 × 8. The luminance data Y is divided into four blocks Y1, Y2, Y3, and Y4, and the color difference data Cr and Cb are divided into two blocks Cr1, Cr2, and Cb1 and Cb2, respectively.

  A block of 8 × 8 pixels is subjected to two-dimensional DCT, and the DCT coefficient is divided by each element of the quantization table, and the remainder is rounded. The DCT coefficients quantized in this way are encoded by the Huffman code definition table (codes are assigned according to the appearance probability of the codes).

  Then, as shown in FIG. 9, an AC value (AC value) is recorded following the head value of each block (DC value (DC value) indicating the lightness level of the entire block). Since the DC values of adjacent blocks have close values, a difference value from the DC component of the immediately preceding block is usually recorded.

  Further, as shown in FIG. 9, a restart interval marker (RIT) is recorded for each of a plurality of MCUs. The absolute value of the DC value of the block immediately after the restart interval marker is recorded.

  In the image format shown in FIG. 10C, the data amounts of the color difference data Cr and Cb constituting the pixel are each halved in the horizontal direction, and Y = 4, Cr = 2, and Cb = 2. The ratio is 4: 2: 2. This utilizes the fact that the human eye is insensitive to color, so that even if the color information is halved, no deterioration in image quality is perceived. In addition to the 4: 2: 2 format shown in FIG. 10C, there are 4: 2: 0 (4: 1: 1) and 4: 4: 4 (1: 1: 1) formats. The invention can be applied to any format.

Returning to FIG. 5, a part of the JPEG data received from the interface unit 12 is temporarily stored in the work memory 16. When the header part is held in the work memory 16, the CPU 14 analyzes the header information (step S12), and at least the main scanning / sub-scanning direction resolution, the number of color components, each color sampling factor (SF), and each color quantization table. (IQNT), each color Huffman code definition table (DHT) is set in the JPEG decompression block 18, and at least each color sampling factor (SF) and resizing rate are set in the resizing block 26 (step S14). The resizing rate can be obtained from the ratio between the image size of the JPEG data input via the interface unit 12 and the image size of the output image output to the print unit.
When the analysis of the data temporarily stored in the work memory 16 is completed, the CPU 14 sequentially deletes the data from the work memory 16 and then stores the data in the work memory 16. When the processing for the header information is completed (step S16), the process proceeds to the next step.
(2) JPEG Decompression Processing and Resizing Processing FIGS. 6 and 7 are flowcharts showing CPU operations in JPEG decompression processing and resizing processing by the image processing apparatus according to the present invention, respectively. FIG. 8 is a schematic diagram mainly showing the exchange of signals between the CPU 14, the JPEG decompression block 18, and the resize block 26.
<JPEG expansion processing>
Scan data is transferred from the interface section 12 following the information in the header section (see FIG. 9). Similarly, the scan data is temporarily stored in the work memory 16, and the CPU 14 transfers the data from the work memory 16 to the JPEG decompression block 18. At this time, DMA1 (Direct Memory Access) may be installed to reduce the load on the CPU 14 (see FIG. 1), and the DMA1 may transfer data.

  As shown in FIG. 6, the JPEG decompression block 18 receives the JPEG decompression start signal a from the CPU 14 and starts decompressing (step S20). The CPU 14 receives the JPEG expansion acknowledgment signal b from the JPEG expansion block 18 and transfers the data temporarily stored in the work memory 16 by the work memory load signal i (step S22). The CPU 14 transfers data stored in the flash ROM 30 by a flash load signal k during data processing and printing described later.

The JPEG decompression block 18 performs JPEG based on the main scanning / sub-scanning direction resolution, the number of color components, each color sampling factor (SF), each color quantization table (IQNT), and each color Huffman code definition table (DHT) sent from the CPU 14. The compressed data that has been compressed is expanded (step S24).
The JPEG decompression block 18 continues the decompression until the decompression of the MCU unit is completed, and when determining that the decompression of the MCU unit is completed, the JPEG decompression block 18 sends an MCU decompression completion signal c to the CPU 14 (step 26S). When receiving the MCU expansion completion signal c, the CPU 14 outputs the expansion standby signal d to the JPEG expansion block 18 and stops the operation of the JPEG expansion block 18 (step S28).
The CPU 14 determines whether or not the decompression of the image for one frame has been completed. If not, the CPU 14 returns to step S20 and starts decompressing the next MCU (step S30).

  As will be described later, the flash ROM 30 stores compressed data for one frame divided into MCUs (blocks) of M rows and N columns, but at the time of printing, the compressed data of each block can be read in an arbitrary order. It is like that.

Therefore, when decompressing the compressed data stored in the flash ROM 30 at the time of printing, the compressed data is transferred by the flash load signal k based on the address information indicating the memory area where the compressed data of the next decompressed block is stored. Read and transfer (step S32).
<Resize processing>
The 8 × 8 pixel data expanded by the JPEG expansion block 18 is stored in the expanded data buffer 20 for each color component. In FIG. 7, when receiving the resize preparation signal e (ON signal) from the decompressed data buffer 20 (step S40), the CPU 14 outputs a resize start signal f to the resize block 26 to start the resizing process (step S42).

  The resizing block 26 performs resizing in the H direction and the V direction in units of 8 × 8 pixels transferred from the decompressed data buffer 20 in accordance with the resizing rate set by the CPU 14 (step S44).

  As shown in FIG. 11A, the received compressed data (bit string) is decompressed by the JPEG decompression block 18 and stored in the decompressed data buffer 20. Each time one block of 8 × 8 pixels is stored in the decompressed data buffer 20, it is transferred to the resize block 26, and first, resizing in the H direction is performed (FIG. 11B).

  The data resized in the H direction by the resize block 26 is written into the H resize buffer 22 (FIG. 11C). Note that FIG. 11 shows the case of resizing to 3/4, and an 8 × 8 pixel block is resized to 8 × 6 pixels in the H direction. In this embodiment, resizing of 1/12 times to 3 times is possible, and the H resizing buffer 22 is considered in consideration of a case where one block of 8 × 8 pixels is resized 3 times in the H direction. The size of the has been determined.

  When data of 8 dots or more is accumulated in the H resize buffer 22 in the H direction, one block of 8 × 8 pixels is read from the H resize buffer 22 and transferred again to the resize block 26 (FIG. 11C )).

The data transferred from the H resize buffer 22 is resized in the V direction by the resize block 26 and then written to the V resize buffer 24 (FIG. 11D).
Returning to FIG. 7, the CPU 14 determines whether or not a predetermined amount of data necessary for compression (data of each color component of 8 dots or more in the H direction and 8 dots or more in the V direction) is stored in the V resize buffer 24. The determination is made based on the signal g from the resize buffer 24 (step S46). When the signal g (ON signal) indicating the specified amount accumulation is received, the CPU 14 writes the data of each color component of 8 × 8 pixels from the V resize buffer 24 to the work memory 16 with the work memory write signal h (step S48). ). In order to reduce the load on the CPU 14, the DMA 2 may perform transfer (see FIG. 1).
The CPU 14 JPEG-compresses the 8 × 8 pixel data for each color component written in the work memory 16 by software (step S50), and writes the compressed data in the flash ROM 30 using the flash write signal j (step S50). S52).

  In parallel with the JPEG compression process and the compressed data writing process to the flash ROM 30, the resize process is performed until the image resize process for one frame is completed (step S56).

The image data of any size received in this way is resized to VGA size, and the resized image data is sequentially JPEG-compressed for each 8 × 8 pixel block for each color component (YCbCr) and stored in the flash ROM 30. (See FIGS. 11E and 11F).
<Resize processing in block units>
Next, even if resizing (resolution conversion) processing is performed in units of blocks of n × n pixels (8 × 8 pixels in this embodiment), there is continuity in the H and V directions of the image for one frame. An image process that can be resized will be described.
The resizing block 26 has an interpolation table in which an interpolation coefficient used for the interpolation calculation is set according to the resizing rate m (= a / b).
On the other hand, the CPU 14 obtains a resizing rate from the ratio between the image size recorded in the header portion of the JPEG data and the image size of the desired output image, and stores the required interpolation coefficient in the interpolation table based on this resizing rate. Let
Here, the resizing rate m that can be resized by the resizing block 26 is set in a range of 1/12 ≦ m ≦ 3, and the values of a and b that determine the resizing rate m are 1 ≦ a ≦ 64, 1 respectively. It is an integer of ≦ b ≦ 64.

  Next, the interpolation table will be described in detail.

  The interpolation table is a table having combinations of interpolation coefficients corresponding to the number of numerators a with the number of denominators b (number of pixels) of a / b determining the resizing rate m as one period. That is, in the case of the resizing rate m (= a / b), the number of pixels of the numerator a is created by interpolation based on the number of pixels of the denominator b, and this is repeated with the number of pixels of the denominator b as a cycle.

  For example, when the resizing ratio m = 2/3, two pixels are created from three pixels by interpolation calculation. The period in this case is 3, and an interpolation coefficient for performing interpolation calculation of two pixels is repeatedly used for every three pixels.

  FIG. 12A shows an interpolation table in the case of the resizing rate m = 2/3. This interpolation table has the same denominator 3 (numbers 0, 1, 2) as one period, and the number of interpolation coefficients corresponding to the number of numerators 2. ((0.75, 0.25), (0.25, 0.75)).

  FIG. 12B shows a case where data in block units is resized to 2/3 in the H direction. As shown in the figure, two pixels {(a1, a2), (a3, a4) for every three pixels {(A1 to A3), (A4 to A6), (A7, A8, B1),. , (A5, b1),...} Are obtained by interpolation using the interpolation coefficients of the interpolation table shown in FIG.

  Incidentally, the pixel a5 can be calculated based on the pixels A7 and A8 in the same block, but the pixel b1 needs to be calculated based on the pixels A8 and B1 of two adjacent blocks. Therefore, when resizing in the H direction, the pixel data of the last column of each block (the eighth column in the case of an 8 × 8 pixel block) is held until the next block is resized. Further, the last position on one cycle in the H direction of each block (in this example, the number 1 of the cycle of 012, 012,...) Is held. Thus, the pixel data having the interpolation position between the pixels of the adjacent blocks on the left and right is calculated based on the pixel data of the last column held and the position in one cycle when the next block is resized in the H direction. can do.

  Even when resizing each block in the V direction, the resizing can be performed using the same interpolation coefficient of the interpolation table as when resizing in the H direction. When resizing in the V direction, the pixel data of the last row of each block (the 8th row in the case of an 8 × 8 pixel block) is moved to the bottom adjacent to each other in the same manner as in the resizing in the H direction. The block is held until the side block is resized, and the last position in one cycle in the V direction of each block is held. As a result, a pixel having an interpolation position between the pixels of the upper and lower adjacent blocks based on the pixel data of the last row held at the time of resizing in the V direction of the lower blocks adjacent to the upper and lower sides and the position on one cycle. Can be calculated.

In this embodiment, the CPU 14 for calculating the resizing rate stores the interpolation coefficient corresponding to the resizing rate in the interpolation table and creates the interpolation table used for the interpolation calculation. However, the present invention is not limited to this. First, a plurality of interpolation tables that store interpolation coefficients corresponding to the resizing ratio are prepared in advance, and the CPU 14 that calculates the resizing ratio uses the interpolation table that stores the interpolation coefficients corresponding to the resizing ratio as an interpolation calculation. You may make it set as an interpolation table to be used.
(3) JPEG compression processing and pointer storage As described above, the CPU 14 JPEG-compresses 8 × 8 pixel data for each color component written in the work memory 16 by software, and sequentially writes the compressed data in the flash ROM 30. .
That is, 8 × 8 pixel luminance data Y, 8 × 8 pixel color difference data Cb, and 8 × 8 pixel color difference data Cr are JPEG compressed as 1 MCU. The image format at this time is a 1: 1: 1 format. The MCU at this time is also referred to as a block hereinafter.

When writing compressed data in units of 1 MCU (1 block) to the flash ROM 30 (including the time of creating compressed data), the CPU 14 corresponds to the order of writing each block to the flash ROM 30 as shown in FIG. An information table is created in which the address pointer of each block, bit information, and the first value (DC value) of the compressed data are stored in association with the block number (0, 1, 2,... X). As the DC value, three DC values for each of luminance data Y, color difference data Cb, and Cr are stored.
Based on the address pointer on the information table created in this manner, a block at an arbitrary position can be read from the flash ROM 30 with respect to the block of variable length data, and the compression read based on the DC value on the information table The block can be JPEG decompressed.

In this embodiment, the CPU 14 performs JPEG compression on a block basis by software. However, the present invention is not limited to this, and a JPEG compression block 29 is installed as shown by the dotted line in FIG. You may make it compress. When JPEG compression is performed by hardware, the compressed data may be written to the flash ROM 30 by the DMA 3 instead of the CPU 14.
<< B. During data processing
(1) Data loading and JPEG decompression processing / resizing processing At the time of data processing in which an output image stored in the flash ROM 30 is analyzed to generate image processing parameters, the CPU 14 loads scan data from the flash ROM 30 in the address order according to the pointer. The data is sequentially transferred to the JPEG decompression block 18. The JPEG decompression process and the resizing process are described in << A. This is performed in the same manner as in the case of data reception (see FIGS. 6 and 7), but the resizing rate m during the resizing process is fixed at Y / X (X and Y are integers and constants).

  X and Y are values depending on the size of the work memory 16 (image size of a storable uncompressed image). In this embodiment, Y / X = 1/4. Note that since the image size of the output image stored in the flash ROM 30 is 640 × 480 pixels, the image size of the analysis image stored in the work memory 16 for image analysis is 160 × 120 pixels.

The resized image data is transferred to the work memory 16 by the CPU 14.
(2) Determination of Image Processing Parameter The CPU 14 performs image analysis based on a reduced image of the output image stored in the work memory 16 and determines an image processing parameter.

For example, the luminance data Y and the color difference data Cb, Cr of the analysis target image are converted into R, G, B data, and a histogram indicating the luminance distribution for each of R, G, B is created as shown in FIG. When this histogram is analyzed and, for example, it is determined that the white balance is bad because the luminance of B is low, a white balance correction value for uniformly increasing the luminance of B is obtained and used as an image processing parameter. Also, it is possible to determine image processing parameters for performing brightness correction, gamma correction, tone curve setting, and the like by judging the brightness and contrast level of the entire image from the histogram. Furthermore, the CPU 14 can recognize the face and determine the vertical direction of the output image from the vertical direction of the face.
<< C. During printing
(1) Data loading and JPEG decompression / resizing processing When printing an output image stored in the flash ROM 30 based on the image processing parameters and outputting it to the print head 32, << B. As in the case of data processing, the CPU 14 loads the scan data from the flash ROM 30 in the address order according to the pointer, and sequentially transfers it to the JPEG decompression block 18.
Here, the output image can be output by being rotated 90 degrees, 180 degrees, or 270 degrees in accordance with the reading order of each block from the flash ROM 30.
<First Embodiment for Rotating Output Image>
(90 degree rotation)
When an output image of P × Q pixels is divided into blocks of n × n pixels, it is divided into blocks of (P / n) × (Q / n). In this embodiment, since the output image is 640 × 480 pixels and one block is 8 × 8 pixels, it is divided into 80 × 60 blocks.
If the output image is divided into blocks of M rows and N columns, the flash ROM 30 has 1 row and 1 column block, 1 row and 2 column block,..., 1 row and N column block,. Scan data is recorded in the order of N columns of blocks. When the output image is rotated 90 degrees and output, blocks of M rows and N columns are sequentially read from the flash ROM 30 in the column direction from the first column.
FIG. 13 shows the reading order of each block from the flash ROM 30 when the output image is rotated 90 degrees and output (when the main scanning direction and the sub-scanning direction are converted). As shown in FIG. 13B, the block numbers corresponding to the blocks in the column direction are 0, N, 2N,... In the first column, 1, N + 1, 2N + 1,. Accordingly, the blocks in the M rows and N columns from the flash ROM 30 in the column direction sequentially from the first column based on the address and bit information corresponding to the block numbers in the column direction stored in the information table shown in FIG. Can be read. FIG. 13C shows a memory map of the flash ROM 30 showing a state in which a block in the first column is allocated.
The blocks read in the column direction in this way are sequentially transferred to the JPEG decompression block 18.

The JPEG decompression process by the JPEG decompression block 18 is the above-described << A. This is performed in the same manner as in the case of data reception, but when decoding the compressed data of the decompression target block (block of i rows and j columns (1 ≦ i ≦ M, 1 ≦ j ≦ N)), i rows j When the DC value of the compressed data of the column block is a differential value, the compression of each block from the block nearest to the i row j column block having the absolute value as the DC value of the compressed data to the i row j column block The DC value of the data is read from the information table shown in FIG. 13A, accumulated, and decoded using the accumulated value.
Further, the resizing rate m at the time of resizing processing by the resizing block 26 is fixed to 1 time (equal size). Even when printing without resizing, a special switching means for bypassing the resizing block 26 can be omitted by making the resizing ratio 1 and always passing through the resizing block 26.

  The 8 × 8 pixel data of each color component block that has been JPEG decompressed as described above and resized to 1 × is temporarily stored in the V resize buffer 24 and then sent to the print controller 28 line by line. . Then, R, G, and B data subjected to appropriate image processing by the print control unit 28 are sequentially output to the print head 32.

  FIG. 15 is a conceptual diagram showing a data flow during printing from the flash ROM to the printing head. As shown in FIG. 15 (A), each block is read from the flash ROM 30 by the head pointer of the block row, and the JPEG decompression process and the data subjected to the resizing process at the same magnification are as shown in FIG. 15 (B). The data is sequentially written to the V resize buffer 24 in units of 8 × 8 pixel blocks. At the time of writing to the V resizing buffer 24, the output order of the image data in the 8 × 8 pixel block is changed according to the rotation of the image (90 degrees, 180 degrees, 270 degrees).

Then, when block data for one column is written in the V resize buffer 24, the data held in the V resize buffer 24 is sent line by line to the print head via the print controller 28 and the print buffer (FIG. 15). (C)).
(180 degree rotation)
FIG. 16 shows the reading order of each block from the flash ROM 30 when the output image is rotated 180 degrees and output. In this case, on the information table shown in FIG. 16A, the blocks corresponding to the order of the smallest block number X-1, X-2,..., 2, 1, 0 are read from the largest block number X (FIG. 16). (B) and (C)).

  Similarly, the output image can be output after being rotated by 270 degrees, and can also be output after being mirror-inverted.

It should be noted that the image is rotated 90 times and output to the print head 32 as described above at the time of printing. The print head 32 is arranged in parallel with the short side of the rectangular instant film 50 as shown in FIG. Because. Further, the reason for rotating the output image by 180 degrees or 270 degrees is that the instant film 50 shown in FIG. 2 has a frame image printed in advance in the surrounding margin, and the top and bottom direction of the frame image is output. This is for printing so that the vertical direction of the image matches. Note that the vertical direction of the output image can be detected by the CPU 14 performing image analysis.
<Second Embodiment for Rotating Output Image>
The CPU 14 JPEG-compresses the 8 × 8 pixel data for each color component written in the work memory 16 in units of 1 MCU (1 block) by software, but inserts a restart interval marker (RIT) for each MCU ( (See FIG. 9). As a result, the first DC value of the compressed data of the block is recorded as an absolute value.
As shown in FIG. 17C, when a restart interval marker (RIT) is inserted, if there is an empty bit in the last byte of the previous block, dummy data (“1”) D After that, a 2-byte restart interval marker (RIT) is recorded. Each restart interval marker (RIT) specifies a recording position on the flash ROM 30 only with address information (without bit information).
Therefore, when writing the compressed data of each block to the flash ROM 30 (including when the compressed data is created), the restart interval marker immediately before each block is associated with the block number corresponding to the order of writing to the flash ROM 30 of each block. By creating an information table storing (RIT) address pointers, blocks can be read out from the flash ROM 30 and transferred to the JPEG decompression block 18 in any order as in the first embodiment. Accordingly, the output image can be output after being rotated by 90 degrees, 180 degrees, and 270 degrees.
Also, since the head DC value of the compressed data of each block is an absolute value, it is not necessary to use the head DC value of another block, and the DC value is added to the information table as in the first embodiment. There is no need to store it.
<Modification of Second Embodiment>
In the second embodiment, a restart interval marker (RIT) is inserted for each MCU. However, in the modification of the second embodiment, a restart interval marker (RIT) is not inserted and Compressed data in which the first DC value of the compressed data is recorded as an absolute value is created.

Further, as shown in FIG. 13A, information that stores the address pointer and bit information of each block in association with the block number (0, 1, 2,... X) corresponding to the writing order of each block to the flash ROM 30. Create a table.
Based on the address pointer on the information table created in this way, a block at an arbitrary location can be read from the flash ROM 30 and transferred to the JPEG decompression block 18.
On the other hand, in the JPEG decompression block 18, a dummy restart interval marker (RIT) is input immediately before fetching a block from the flash ROM 30. Accordingly, the JPEG decompression block 18 determines that the block fetched from the flash ROM 30 is a block immediately after the restart interval marker (RIT), and decompresses the compressed data of the block using the DC value of the compressed data of the block as an absolute value. I do.
The modification of the second embodiment can save the memory capacity of the flash ROM 30 as much as the restart interval marker (RIT) is not recorded in the flash ROM 30 as compared to the second embodiment.
<Third Embodiment for Rotating Output Image>
The CPU 14 JPEG compresses the 8 × 8 pixel data for each color component written in the work memory 16 in units of 1 MCU (1 block) by software, and inserts a restart interval marker only immediately after the block in the last column To do. As a result, the first DC value of the compressed data of the block in the first column is recorded as an absolute value, and the first DC value of the compressed data of the block in the other column is a difference value from the DC value of the immediately preceding block. Will be recorded.
As shown in FIG. 17A, when the compressed data of each block is written to the flash ROM 30, the block immediately after the block in the last column of each row is associated with the block numbers 0, N, 2N,. Create an information table that stores the address pointer of the restart interval marker (RIT). In this case, as shown in FIGS. 17A and 17C, the address pointer of the restart interval marker (RIT) is only address information and the bit position is set to 0.
Therefore, the compressed data of each block in the first column can be read from the flash ROM 30 in accordance with the address pointer shown in FIG. The block read in this way is transferred to the JPEG decompression block 18 and decoded.
On the other hand, when the compressed data of each block of the first column is read, the bit position and address of the information table can be detected because the head bit position and address information of the compressed data of each block of the second column can be detected. Information is updated to the detected bit position and address position. Accordingly, the compressed data of each block in the second column can be sequentially read based on the information table in which the bit position and the address position are updated. Similarly, the compressed data of each block in the third column, fourth column,... Can be sequentially read.
On the other hand, the DC value necessary for decoding each block in the second column is the first DC value (absolute value) of each block in the first column and the first DC value (difference value) of each block in the second column. Therefore, the first DC value of each block in the first column is held until each block in the second column is decoded. Similarly, the DC value necessary for decoding each block in the third column is the DC value necessary for decoding each block in the second column (the accumulated value of the DC value in the first column and the DC value in the second column). And the DC value (difference value) at the beginning of each block in the third column can be obtained by accumulation, and the accumulated value is held until the decoding of each block in the third column.
That is, since the DC value required for decoding the block in the i-th column (1 ≦ i ≦ N) is an accumulated value of the DC value of the compressed data of each block from the first column to the i-th column, The calculated value is held until the block of the (i + 1) -th column is decoded.

In this embodiment, until the decompression target block is decompressed, the DC value of the first block in the row of the decompression target block to the DC value of the immediately preceding block is accumulated, and the accumulated value is held. However, the present invention is not limited to this, and at the time of decompression of the decompression target block, an accumulated value from the DC value of the first block in the row of the decompression target block to the DC value of the immediately preceding block may be calculated each time. .
(2) YCbCr / RGB conversion and image processing When the data of each color component of the number of dots (480) × 8 dots of the print head is accumulated in the V resize buffer 24, the accumulated data is a print control unit line by line. 28 (see FIG. 15).

  As described above, the print control unit 28 includes a 3D-LUT. As shown in FIG. 18A, this 3D-LUT has a basic table for converting YCbCr input data to RGB output data and performing gradation conversion for improving color reproducibility and gradation reproducibility. Has been.

Further, image processing parameters corresponding to the image analysis result are added from the CPU 14 to the print control unit 28, and the print control unit 28 customizes the basic LUT based on the image processing parameters. That is, the basic LUT1 shown in FIG. 18A is customized according to the image analysis result, and becomes the LUT2 shown in FIG.
For example, when the white balance is bad because the luminance of B is low, the CPU 14 adds an image processing parameter for uniformly increasing the luminance of B to the print control unit 28. The print control unit 28 To reconstruct LUT2 in which the luminance of B is uniformly increased (× 1.3). In addition, a plurality of image processing parameters can be reflected in a single LUT.

  The YCbCr input data sent from the V resizing buffer 24 line by line to the print control unit 28 is converted into RGB output data by the 3D-LUT in the print control unit 28, as well as white balance correction and gradation conversion. Etc., image processing is performed.

  In this way, R, G, B data subjected to image processing line by line from the print control unit 28 is output to the print head 32. The R, G, B data exposes the instant film by controlling the transmittance of the liquid crystal shutter 36 of the print head 32 for each of R, G, B.

  In this embodiment, the compressed output image is stored in the volatile memory (flash ROM 30). However, the present invention is not limited to this, and the compressed output image may be stored in the volatile memory. Further, the compressed data is sequentially written in the memory every time the block unit compression is completed. However, the present invention is not limited to this, and the compressed data may be written in the memory collectively after the compression of all the data is completed. Good. Also, analysis of header information and the like can be performed using a cache memory inside the CPU, and in this case, the work memory 16 can be omitted. Furthermore, although the case where the output image is output to the printer unit has been described, the present invention can also be applied to the case where the output image is output to the image display device.

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention. FIG. 2 is a perspective view showing an instant film on which an image is printed by the printing head. FIG. 3 is a perspective view showing a main part of the printing unit. FIG. 4 is a sectional view of the printing head. FIG. 5 is a flowchart showing the CPU operation at the time of header information analysis by the image processing apparatus. FIG. 6 is a flowchart showing the CPU operation in the JPEG expansion processing by the image processing apparatus according to the present invention. FIG. 7 is a flowchart showing the CPU operation in the resizing process by the image processing apparatus according to the present invention. FIG. 8 is a schematic diagram mainly showing transmission / reception of signals between the CPU and the JPEG decompression block and resize block. FIG. 9 is a diagram used for explaining the data structure of JPEG data. FIG. 10 is a diagram used for explaining the minimum coding unit (MCU) of the received compressed data. FIG. 11 is a diagram illustrating a data flow during JPEG decompression, resizing, and JPEG compression during data reception. FIG. 12 is a diagram used for explaining details of the resizing process in units of blocks. FIG. 13 is a diagram illustrating a first embodiment of image processing in which an output image is rotated 90 degrees and output. FIG. 14 is a diagram illustrating a state in which a histogram of R, G, and B data is created and image analysis is performed. FIG. 15 is a conceptual diagram showing a data flow during printing from the flash ROM to the printing head. FIG. 16 is a diagram illustrating a first embodiment of image processing in which an output image is rotated 180 degrees and output. FIG. 17 is a diagram illustrating another embodiment of image processing when an output image is rotated 90 degrees and output. FIG. 18 is a diagram showing a YCbCr / RGB conversion and image processing method by 3D-LUT.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... ASIC, 12 ... Interface part, 14 ... CPU, 16 ... Work memory, 18 ... JPEG decompression block, 20 ... Decompression data buffer, 22 ... H resize buffer, 24 ... V resize buffer, 26 ... Resize block, 28 ... Print Control unit, 30 ... flash ROM, 32 ... printing head

Claims (4)

Image data for one frame divided into blocks of M rows and N columns is sequentially compressed in blocks of n × n pixels from blocks of 1 row and 1 column to blocks of M rows and N columns, and the final column of each row Compression means for creating compressed data having a data structure in which a restart interval marker is inserted immediately after the block of
A memory for storing compressed data for one frame compressed by the compression means;
Stores position information indicating the write position of the compressed data of the block in association with the block number of the block for an arbitrary column, and an accumulated value for calculating the absolute value from the first value of the compressed data of the block An information table to
The position information indicating the writing position of the restart interval marker inserted immediately after the last column block of each row in association with the block number of the first column block and the accumulated value of 0 are written to the information table, and Information table creation means for updating the position information and the accumulated value;
Reading means for reading out the compressed data of the blocks in each column based on the position information recorded in the information table;
Decoding means for decoding the compressed data of the read block using an accumulated value stored in the information table;
Buffer means having a capacity smaller than the data amount of the output image of the desired image size;
Writing means for changing the output order of the image data in the block of n × n pixels decoded by the decoding means and writing to the buffer means;
The information table creating means reads the compressed data of the i-th column (1 ≦ i ≦ N) block of each row by the reading means, and when the decoding by the decoding means is completed, the information table creating means completes the i-th row of the information table. The position information used for reading the compressed data of the block in the column is updated to position information indicating the writing position of the compressed data in the block of the (i + 1) th column in each row, and the accumulated value recorded in the information table is updated in each row. Update the accumulated value of the first value of the compressed data of the blocks from the first column to the i-th column,
An image processing apparatus capable of outputting image data rotated by 90 degrees from the buffer means.   Interface means for continuously receiving compressed data of image data of an arbitrary size for one frame, decompression means for decompressing compressed data in block units of n × n pixels, and n × n pixel decompressed by the decompression means A decompression data buffer means for temporarily storing image data; and a resolution conversion means for converting the n × n pixel image data stored in the decompression data buffer means into an arbitrary size and outputting the converted data to the buffer means. 2. The compression unit according to claim 1, wherein when the image data in a unit that can be processed is held in the buffer unit, the compression unit creates compressed data having a data structure that is sequentially compressed in units of blocks of n × n pixels. An image processing apparatus according to 1.   The resolution conversion means is an interpolation table storing interpolation coefficients used for interpolation calculation in accordance with a magnification a / b (a, b: natural number) for reducing or enlarging the image size, and a denominator of the magnification a / b The image size is reduced or reduced based on an interpolation table having combinations of interpolation coefficients corresponding to the number of numerators a with the number of pixels b as one cycle, and the image size of the image data for one frame and the image size of a desired output image. Magnification magnification a / b is calculated, and an interpolation coefficient is stored in the interpolation table based on the calculated magnification a / b, or an interpolation table corresponding to the calculated magnification a / b is interpolated from a plurality of interpolation tables. Table setting means for setting as an interpolation table used for calculation, and n × n pixel image data stored in the decompressed data buffer means The image processing apparatus according to claim 2, characterized in that the resolution conversion by interpolation calculation based on.   The resolution converting means is an n × x block of i rows and j columns (1 ≦ i ≦ M, 1 ≦ j ≦ N) of a plurality of blocks obtained by dividing the image data of an arbitrary size into blocks of M rows and N columns. At the time of converting the resolution of the image data of n pixels in the horizontal direction, the image data of n columns of the block of i rows and j columns and the position in one cycle of the interpolation table corresponding to the image data of the n columns are i rows and j + 1 columns. Up to the resolution conversion of the image data of the i-th block, and used for the resolution conversion of the image data of the i-row, j + 1-column block, and the n-row image data of the i-row / j-column block and the n The position of one cycle of the interpolation table corresponding to the image data of the row is held until the resolution conversion of the image data of the block of i + 1 row j column, and used for the resolution conversion of the image data of the block of i + 1 row j column. Specially The image processing apparatus according to claim 3,.

Images