JP2009159631A - Image processor, image processing method, computer program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor and an image processing method that attain speeding-up of an image processing speed with respect to the whole blocked image data while reducing a memory capacity required for image processing. <P>SOLUTION: The image processor includes a block image processing means for executing image processing in block units to image data input in block units, a conversion means for converting the image data processed in block units into a raster shape, and a pixel image processing means that uses the image data converted into the raster shape so as to execute image processing in one-pixel unit. Image processing applied to the input data is executed in two stages of the block image processing means and the pixel image processing means. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method suitable for performing image processing such as clipping processing, resolution conversion, and rotation processing on block image data.

  Conventionally, in an image processing apparatus that generates desired data by performing image processing such as clipping processing, resolution conversion processing, and rotation processing on block image data, input block-shaped image data is Each image processing was performed after rasterization.

  However, in the conventional image processing method, since the image processing is performed after rasterization, not only the time required for the entire image processing is lengthened, but also the processing memory required for the image processing is stored in the image. There was a problem that it was necessary to mount many in a processing apparatus. Also, when performing rearrangement processing such as image rotation, the conventional method requires a buffer for rearrangement processing. In addition, in order to reduce the buffer for the rearrangement process, when rearrangement is executed at the write address, the burst address cannot be used because access is not continuous, and there is a problem that access takes time. .

  The present invention has been made in consideration of such circumstances. When image processing is performed on blocked image data, the image processing speed for the entire image data is increased, and image processing is performed. An object of the present invention is to provide an image processing apparatus and an image processing method capable of reducing the required memory capacity.

In order to achieve the above object, according to the present invention, block image processing means for executing image processing in block units on image data input in block units,
Conversion means for converting the image data processed in block units into a raster shape;
Pixel image processing means for performing image processing in units of one pixel using the image data converted into a raster shape,
There is provided an image processing apparatus characterized in that image processing to be performed on input image data is performed in two stages of the block image processing means and the pixel image processing means.

  As described above, according to the present invention, a range required for processing is specified for image data in a certain block unit, image processing is performed only on the required range, and a more detailed image than the subsequent image processing is performed. Since the processing is performed, the image processing speed for the entire image data can be increased, and the memory required for the image processing can be reduced. In addition, the continuity of memory access can be improved and the memory use efficiency can be improved.

It is a block diagram for demonstrating the outline | summary of the image processing apparatus by the 1st Embodiment of this invention. It is a block diagram for demonstrating the detailed structure of the front | former stage image process part 21 in FIG. It is the schematic for demonstrating the clipping process implemented in the clipping process part 211. FIG. It is the schematic for demonstrating the rotation process performed in the rearrangement process part. It is a figure for demonstrating the specific example of the processing operation of the rearrangement process part 213 in 1st Embodiment. It is a figure for demonstrating in detail the resolution conversion process and rearrangement process after a clipping process. It is a figure which shows the interpolation coefficient in the case of each sampling point shown in FIG. 5 is a flowchart for explaining an operation procedure of the image processing apparatus according to the first embodiment. It is a flowchart for demonstrating in detail the clipping process procedure in step S82. It is a flowchart for demonstrating the detail of the resolution conversion process performed by step S83. It is a block diagram which shows the structure of the image processing apparatus which concerns on the 2nd Embodiment of this invention. It is a figure for demonstrating the specific example of the image processing by the image processing apparatus which concerns on 2nd Embodiment. 3 is a block diagram showing a detailed configuration of an MCU reduction circuit 112. FIG. 3 is a block diagram showing a specific configuration of a horizontal size reduction circuit 131 or a vertical size reduction circuit 132. FIG. It is a figure which shows the sampling point after MCU reduction concretely. It is the figure which showed a mode that the luminance signal data in the specific example of FIG. 12 was stored in a Y block buffer. It is the figure which showed a mode that the color difference signal data in the specific example of FIG. 12 was stored in Cr block buffer and Cb block buffer. It is a figure which shows the relationship between rotation mode and the reading position of a Y block buffer. It is a figure which shows a mode that the luminance signal data at the time of in-block rotation at the time of writing are stored in a Y block buffer. It is a figure which shows a mode that the color difference signal data at the time of in-block rotation at the time of writing are stored in Cr block buffer and Cb block buffer.

  Embodiments of the present invention are shown below. Of course, the following embodiments are provided for facilitating implementation by those skilled in the art of the present invention, and are only included in the technical scope of the present invention defined by the claims. It is only an embodiment of the part. Therefore, it will be apparent to those skilled in the art that even embodiments that are not directly described in the present specification are included in the technical scope of the present invention as long as they share the same technical idea.

  In addition, although several embodiment is described below for convenience, these are not only materialized as an invention separately, but of course, it will be understood by those skilled in the art that the invention can also be achieved by appropriately combining a plurality of embodiments. Easy to understand.

[First Embodiment]
Hereinafter, an image processing apparatus according to a first embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram for explaining the outline of the image processing apparatus according to the first embodiment of the present invention. In FIG. 1, as image data input from the input unit 1, image data scanned in a block shape with a certain size (for example, 8 × 8 pixels) is input. Such data is generated when image data (encoded data) block-encoded as in JPEG is decoded (block decoding) (when the block decoding procedure is executed).

  The image processing unit 2 is connected to the input unit 1, performs various image processing on the block image data in units of processing blocks at the time of block encoding (block image processing procedure), and then rasterizes the image data (raster-like data). The former image processing unit 21 for performing conversion processing (execution of the conversion procedure), and the subsequent image processing unit 22 for performing image processing (pixel image processing procedure) in units of pixels on the raster-converted image data. Consists of The output unit 3 is connected to the image processing unit 2 and is an output device for outputting image data processed by the image processing unit 2 to the outside or storing it in a storage device or the like. The number of image processing units in the image processing unit 2 is not limited to the two described above, and may be one or three or more. That is, the present invention is characterized by including first image processing means (previous image processing unit 21) that performs image processing for each predetermined block on image data after image processing. Further, the present invention is characterized in that the second image processing means (the subsequent image processing unit 22) performs image processing on the image data in units of one pixel.

  FIG. 2 is a block diagram for explaining a detailed configuration of the pre-stage image processing unit 21 in FIG. As shown in FIG. 2, the pre-stage image processing unit 21 in this embodiment includes a clipping processing unit 211 that specifies (extracts) a processing range (specified range) from the range of input image data, and the resolution of the image data. It comprises a resolution conversion processing unit 212 for converting, a rearrangement processing unit 213 for rearranging image data, and an SDRAM 214 used as a band buffer for rasterization.

  In FIG. 2, image data input to the clipping processing unit 211, which is the first processing unit, is indicated by Data_in, and image data output from each processing unit is indicated by Data. A name representing the process is given before “Data” which is image data output for explanation. Further, Valid_in is a valid signal related to image data input to the clipping processing unit 211 as the first processing unit, and Valid is a valid signal related to image data processed by each processing unit. A name indicating the processing is also given before this Valid. The pre-stage image processing unit 21 according to the present embodiment uses the input image data Data_in and the valid signal Valid_in related to the input image data when the input image data is valid (the valid signal Valid_in = H is input). Only when the input image data is taken in, the respective processes are performed.

  First, the image data Data_in and the valid signal Valid_in are input from the input unit 1 illustrated in FIG. 1 to the clipping processing unit 211 of the pre-stage image processing unit 21. As described above, in the present embodiment, it is assumed that input image data is a blocked image. For example, image data for each block of 8 × 8 pixels used for block coding is input to the clipping processing unit 211 as image data Data_in, and at the same time, if the input image data is valid data, a valid signal Valid_in = H is input. In the case of invalid data, Valid_in = L is input.

  The clipping processing unit 211 determines whether or not the input image data is within a clipping range (clipping effective range) in units of blocks. As a result, when it is determined that the input image data is valid data and is within the clipping valid range, a valid signal (Crip_Valid = H) is output. If the image data is outside the clipping valid range or invalid data, no valid signal is output (Crip_Valid = L). Alternatively, when the image data is outside the clipping effective range, another invalid signal may be output. By this operation in the clipping processing unit 211, the resolution conversion processing unit 212 connected to the clipping processing unit 211 can perform the resolution conversion processing only for the blocks that are determined to be valid by the clipping processing unit 211.

  FIG. 3 is a schematic diagram for describing clipping processing performed by the clipping processing unit 211. As shown in FIG. 3A, the horizontal direction of the image data is X and the vertical direction is Y. At this time, the clipping range in block units is designated as follows. That is, the clipping range is set as a rectangular area (rectangular range), and the coordinates (XS, YS) are designated with the top left vertex of the area as Start_Point. Similarly, the coordinate (XE, YE) is designated with End_Point being the lower right vertex of the rectangular area. In this embodiment, a portion surrounded by a rectangular area range that can be defined as shown in FIG. 3 at these two points is clipped, and a valid signal Crip_Valid is output for the blocks in the range. Note that the clipping range designation method is not limited to the above, and the start point (or end point) and size (width, height) may be used. Further, a block unit clipping parameter may be generated from a pixel unit clipping parameter.

  Note that the block shape used in the clipping process in the present embodiment is not limited to a square as shown in FIG. 3A, but a horizontally long block as in the example shown in FIG. It may be a vertically long block like the example shown in FIG. That is, it can be applied to any block shape.

  As described above, for block image data such as a DCT block (8 × 8 pixels) unit in a JPEG image, block-unit output pixels are obtained by performing block unit clipping in the first clipping process. In addition, it is possible to perform clipping in pixel units by performing the second clipping process in the subsequent image processing. Then, the clipping processing unit 211 determines whether the image data is valid data or invalid data in units of blocks, and only the valid data is processed in the subsequent processing unit, thereby performing unnecessary processing in the subsequent processing. It is possible to reduce the load on the entire image processing.

  In particular, when the output of the pre-stage image processing unit 21 is stored in a buffer, the buffer capacity can be reduced. Since clipping is performed in units of blocks, parameters such as the number of pixels in the block are the same for all blocks, and there is no processing that increases due to clipping, including subsequent processing. Further, as described above, since the clipping process is performed by operating the valid signal, it is not necessary to change the interface (I / F) of the subsequent process. In this embodiment, the subsequent processing is terminated at an appropriate timing by propagating the clipping end signal (Crip_Finish) to the subsequent processing while compensating for the latency in each processing. With this configuration, the timing of the final pixel can be detected in the final stage processing, and wasteful processing (time) can be reduced by notifying the CPU or the like of the end of the final processing by an interrupt signal or the like.

  The resolution conversion processing unit 212 performs resolution conversion (Resize) processing on the block corresponding to the valid signal Crip_Valid. In the present embodiment, the resolution conversion processing unit 212 performs block size (Bx × By) reduction. For example, an 8 × 8 pixel block is changed to a 4 × 4 pixel block. That is, output pixels are reduced by reducing the block size. By performing the resolution conversion process within the block, simple resolution conversion is enabled. For example, when an input image is input in a block shape having a size of 8 × 8 pixels, the block size after resolution conversion is 8 × 8, 7 × 7, 6 × 6, 5 × 5, 4 × 4, 3 There are only 8 types of × 3, 2 × 2, and 1 × 1, and the coefficient generation is easy. Further, not only the same reduction circuit can be used in the horizontal direction and the vertical direction, but also a line memory becomes unnecessary. However, in the case of linear interpolation, a register corresponding to the horizontal pixel of the block is required between the horizontal reduction circuit and the vertical reduction circuit.

  In the case of reduction to 7 × 7 pixels or less, the resolution conversion processing unit can be pipelined by invalidating the valid signal output (Resize_Valid = L) at an appropriate timing. That is, in the case of pipelining, an unnecessary signal is generated at a predetermined timing, but the thinning-out process can be easily executed only by invalidating the valid signal output (Resize_Valid = L). By this operation, the control based on the reduction ratio is only required to switch the coefficient and the effective signal output, and the processing can be simplified.

Thus, by performing resolution conversion processing within a block in accordance with the block size of input image data, it is possible to easily realize resolution conversion of image data. Of course, further fine resolution conversion can be performed by the post-stage image processing unit 22 or the like. Similarly, when the resolution-converted block is valid image data, by outputting the resolution-converted image data Resize_Data_out and the valid signal Resize_valid related to the image data, the processing target of the next rearrangement process is performed. Can be further reduced. Note that resolution conversion in the enlargement direction is not performed here because the number of output pixels increases.
In the present embodiment, resolution conversion processing in units of blocks is performed independently in the horizontal direction and the vertical direction. In this embodiment, when the horizontal reduction rate in the resolution conversion process is Rx and the vertical reduction rate is Ry, the output size of the image data in the block-unit resolution conversion process is
n / Bx ≦ Rx <(n + 1) / Bx, where n is an integer from 1 to Bx,
m / By ≦ Ry <(m + 1) / By, where m is an integer from 1 to By,
N × m satisfying In the present embodiment, the horizontal and vertical resolution conversion processing switches the interpolation coefficient according to the position on the block of the input image data, and sets data other than the interpolation point as invalid data. Also, in the horizontal resolution conversion process, the pixel calculation at the right end of the block is invalid data, and in the vertical resolution conversion process, the pixel calculation at the bottom end of the block is invalid data. In the present embodiment, in the horizontal and vertical resolution conversion processing, when the interpolation coefficient is a maximum value that can be set, a pixel multiplied by the coefficient that is the maximum value is set as an interpolation output. In the present embodiment, the horizontal and vertical resolution conversion processing by the block image processing means is a pixel corresponding to a pixel to be multiplied by a coefficient of 0 and not multiplied by 0 when the interpolation coefficient is 0. This pixel is the interpolation output.

  In the rearrangement process in the rearrangement processing unit 213, YCrCb signal synchronization (dot sequential), rotation, and rasterization are performed. Usually, in block coding, in order to improve coding efficiency, it is converted into luminance data Y and color difference data CrCb and coded. Therefore, the luminance data Y and the color difference data CrCb are input to the rearrangement processing unit 213 of the present embodiment in block order. Therefore, in order to convert YCrCb data to RGB data in the subsequent stage, it is necessary to synchronize the YCrCb data. In the present embodiment, the YCrCb data synchronization, intra-block rotation, and intra-block rasterization processes are simultaneously executed using the YCrCb data synchronization buffer.

  First, when Y block data is input, it is stored in a Y block buffer (not shown). When Cr block data is input, it is stored in a Cr block buffer (not shown), and Cb block data is input. Stored in a Cb block buffer (not shown). Next, the pixel positions of the Y block buffer, Cr block buffer, and Cb block buffer are simultaneously read and synchronized. If the sampling rates of the luminance data Y and the color difference data CrCb are different, the data with the lower sampling rate (usually color difference data) is interpolated and output. In the case of a plurality of blocks, the data is rasterized and output in a state where the plurality of blocks are combined.

  Rasterization of the entire image is performed when stored in a band buffer (configured on the SDRAM 214). Here, it is assumed that the YCrCb dot sequential data is converted into RGB data and then stored in the band buffer. The image data subjected to the intra-block rasterization and YCrCb synchronization (dot sequential) is stored in blocks at the corresponding positions of the band buffer. As a result, rasterized images within the clipping range are stored in the band buffer.

  In addition, when image data is rotated simultaneously with the execution of raster conversion, after the rotation within the block is executed in the block buffer, the data is banded while changing the block start address according to the rotation mode. Save it to the buffer. By executing the rotation in the block in the block buffer in advance, the difference in the write sequence to the band buffer depending on the rotation mode is concentrated only on the setting of the head address of the block. Therefore, the post-stage image processing unit 22 or the like can perform image processing using RGB point sequential image data that has undergone raster conversion and rotation processing.

  FIG. 4 is a schematic diagram for explaining the rotation processing performed in the rearrangement processing unit 213. In FIG. 4, as an example of the rotation process, four angles (rotation angles in units of 90 °) of 0 ° (0 °), 90 ° (90 °), 180 (180 °), and 270 ° (270 °) of the image data. ) Example of rotation is shown. The image data used in this example is a 4 × 3 block image with 4 horizontal components and 3 vertical components. As shown in this example, rearrangement (rotation) processing in units of pixels and rearrangement (rotation) processing in units of blocks (blocks themselves) can be performed in the same manner.

  As shown in FIG. 4, in an image 40 that is the target of the rearrangement (rotation) process, a number is assigned to each block in the image. In this number, the number on the right is the X component of the image, and the number on the left is the Y component of the image. That is, a block with X = 0 and Y = 0 is represented by “00”, and a block shifted by 1 in the X direction, that is, a block immediately adjacent to the block 00 is represented by “01”.

  When rearrangement processing with a rotation angle of 0 degrees is performed on the image 40, an image 41 having the same arrangement as the image 40 is stored in, for example, the SDRAM 214.

  When rearrangement is performed at a rotation angle of 180 degrees, as shown in the image 43, the block 00 located at the upper left in the image 40 must be moved to the lower right. In this way, all blocks from block 00 to block 23 are moved. In addition, when the rotation angle is 90 degrees and 270 degrees, the same processing is performed in units of blocks, resulting in the results shown in images 42 and 44, respectively. In the case of 90 degrees and 270 degrees, the vertical and horizontal sizes are reversed.

  Furthermore, the process in the rearrangement process part 213 is demonstrated. FIG. 5 is a diagram for explaining a specific example of the processing operation of the rearrangement processing unit 213 according to the first embodiment. The reason why the continuity of the address of the image data is increased and the effect of improving the transfer efficiency is obtained will be described with reference to FIG. Here, it is assumed that the rearrangement processing unit 213 performs processing to rotate the image 50 configured by 4 × 4 pixel blocks by 180 degrees to generate the image 51. In the conventional process, since the address corresponding to the position of the image 51 is generated in the order of the input pixels, when the image data is recorded in the SDRAM 214, the address is in a decreasing direction, so the burst mode is used. It was not possible to access the SDRAM 214 every pixel, and the transfer efficiency was very poor.

  However, in this embodiment, the image data after the rotation process is temporarily stored in the block buffer 52. As a result, a plurality of data (four data in FIG. 5) continuous in the direction of increasing addresses can be recorded in the SDRAM 214 at a time. In other words, it was possible to access four times instead of the conventional 16 times required for one block. In other words, by performing the rearrangement process in the block, the continuity of the address is increased, thereby making it possible to improve the transfer efficiency.

  FIG. 6 is a diagram for explaining in detail the resolution conversion process and the rearrangement process after the clipping process. Here, it is assumed that image data in which one block is configured by 8 × 8 pixels is input to the resolution conversion unit 212 and is converted into a block having a resolution of 4 × 4 pixels by resolution conversion. Then, it is assumed that the rearrangement processing unit 213 performs a 180-degree rotation rearrangement process.

  First, data of one block 8 × 8 pixels (0 to 63) is input to the resolution conversion unit 212 simultaneously with the valid signal Crip_Valid in order from the 0th data of the luminance signal Y. Then, resolution conversion processing is performed in the resolution conversion unit 212 and converted into image data of 1 block 4 × 4 pixels. The converted image data is subjected to the rearrangement process as described above, and is temporarily stored in the Y block buffer 60, the Cr block buffer 61, and the Cb block buffer 62, respectively. When YCrCb data for one block is stored, it is sequentially read from the heads of the Y block buffer 60, Cr block buffer 61, and Cb block buffer 62, converted into RGB data, and stored at the corresponding address of the band buffer. The sorting is completed.

  Note that the order of the processes in the image processing unit 2 as described above is arbitrary. Furthermore, any processing may be missing. Furthermore, a mode (through mode) in which an input signal is output as it is in each process may be provided so that execution of each process can be arbitrarily selected.

  In the above-described embodiment, an image using a storage device such as SDRAM is described as an example of a block image such as block coding. However, this may be performed by software processing.

  As described above, the present invention is an image processing apparatus including an image processing unit that performs image processing on image data for each predetermined block, and determines whether or not image data is to be image processed for each block. A determination unit is provided, and the image processing unit performs image processing on the image data of the block determined to execute the image processing.

  In addition, the present invention is characterized by comprising designation means (clipping processing unit 211) for designating a range for performing image processing on image data. Furthermore, the present invention is characterized by comprising resolution conversion means (resolution conversion processing unit 212) for converting image data to a predetermined resolution. Furthermore, the present invention is characterized by comprising rearrangement means (rearrangement processing unit 213) for rearranging the positions of blocks constituting image data to predetermined positions.

  Furthermore, the present invention provides JPEG decoded data before input image data is rasterized, and a block used for image processing is a block (8 × 8 pixels) used when JPEG decoding is performed. Or MCU). Furthermore, the present invention provides MPEG decoded data before the image data is rasterized, and a block used for image processing is a block (8 × 8 pixels) used when MPEG decoding is performed. Or MB).

  Next, the processing operation of the image processing apparatus configured as described above will be described. FIG. 8 is a flowchart for explaining the operation procedure of the pre-stage image processing unit 21 according to the first embodiment.

  First, image data to be processed is input from the input unit 1 of the image processing apparatus, and parameter settings necessary for the above-described image processing are performed using an operation unit (not shown) (step S81). Here, the parameters in the image processing are: designation of a range for clipping an image, resolution conversion processing size, rotation mode necessary for rearrangement processing, sampling mode (JPEG 4: 4: 4, 4: 2: 2: 4: 2: 0, etc.), an output band buffer area, and the like.

  Next, clipping processing is performed on the input image data in the clipping processing unit 211 (step S82). FIG. 9 is a flowchart for explaining in detail the clipping processing procedure in step S82.

  Here, as shown in FIG. 3, a case where the clipping range start point is designated as (XS, YS) and the end point is designated as (XE, YE) is taken as an example. As described above, in the present invention, clipping is performed in units of a certain block. It should be noted that finer clipping can be performed by the subsequent processing unit. For example, in the case of JPEG data, this block unit may be an 8 × 8 pixel unit that is a DCT block, or an MCU that combines Y, Cr, and Cb into one unit. By performing such block-based clipping processing, the amount of data stored in the band buffer is reduced. This not only reduces the load required for processing in the subsequent image processing unit 22, but also increases the processing speed of the entire image processing apparatus.

  In step S91, it is determined whether or not the clipping process has been completed. That is, when the clipping range is designated as a rectangular area indicated by the start point (XS, YS) and the end point (XE, YE), the coordinates X, Y of the input image block exceed both the XE and YE ranges ( YES), since it indicates that the range to be clipped has already been exceeded, an end process is performed here (step S93). By this termination processing, the subsequent image processing unit 22 receives an instruction indicating that the subsequent image processing unit 22 has terminated (thereby, the subsequent image processing unit 22 is activated), or the completion signal (Crip_Finish) is propagated. It is also possible to end the process.

  On the other hand, if the clipping process has not ended in step S91 (NO side), it is determined whether the designated range is within the clipping range (step S92). As a result, if it is within the clipping range (YES side), effective data processing is performed to indicate that the data is valid for the subsequent image processing unit (here, resolution conversion processing step S83) (step S94). On the other hand, if it is determined in step S92 that it is not in the valid range (NO side), it indicates invalid data (step S95). It should be noted that invalid data processing can be performed by not outputting a signal or the like to the subsequent image processing unit. When the valid data processing step S94 and the invalid data processing step S95 for the input block are completed, the process returns to the termination determination step S91 to determine the next block, and the above operation is repeated.

  A resolution conversion process is performed on the block determined to be valid in the clipping process in step S82 (step S83). FIG. 10 is a flowchart for explaining details of the resolution conversion processing performed in step S83. The resolution conversion processing unit 212 first determines whether or not there is resolution conversion (step S101). As a result, when it is determined that there is no resolution conversion (NO side), the process ends without performing the resolution conversion process, and the image data input to the subsequent image processing unit is passed as it is. On the other hand, when it is determined that there is resolution conversion (YES side), resolution conversion parameter setting is performed (step S102). Then, resolution conversion processing is performed according to the set parameters, and processing such as thinning processing and linear interpolation processing is performed so as to obtain a desired size (step S103). Note that the resolution conversion process determination step S101 can be performed based on the size after the resolution conversion process set in the parameter setting step S81 of FIG. 8, and the parameter setting step S102 can be omitted. For example, when the horizontal size of the block of the input image is 8 and the horizontal size after resolution conversion is less than 8, resolution conversion is performed. At this time, the horizontal size after resolution conversion is a parameter for resolution conversion, and if it is set to 4, for example, a block of 8 × 8 pixels is reduced to a 4 × 4 block by resolution conversion.

  After the resolution conversion processing is performed in step S83, the rearrangement processing unit 213 performs rearrangement (rotation, etc.) processing within the block (step S84). The processed image data is stored in the SDRAM 214 (step S85). In step S84, the rearrangement process is executed, so that not only address generation is simplified at the time of storage, but also storage of a plurality of units (burst write) is possible instead of conventional pixel-by-pixel storage (single write). Thus, the access speed per pixel can be increased.

  That is, when the storage destination is SDRAM, it is possible to increase the continuity of the write address to the SDRAM and increase the transfer efficiency. Further, in the case where access from many bus masters occurs, such as the main memory of the embedded device, the storage destination is not only the image processing of the post-stage image processing unit 22 but also other processing by increasing the access efficiency. The influence of can be reduced.

  Further, a JPEG image will be described as an example. In clipping processing, rough clipping is performed in units of blocks as in the present embodiment, and then resolution conversion is performed on the extracted blocks, and rasterized RGB points. By storing the data sequentially, it is possible to more easily perform processing such as pixel-by-pixel clipping and more detailed resolution conversion in the subsequent processing unit.

  That is, an image processing apparatus according to the present invention includes a condition setting unit that sets conditions for image processing of image data, a specifying unit that specifies a range for performing image processing on image data, and predetermined image data. The image processing apparatus includes: a resolution conversion unit that converts to a resolution; a rearrangement unit that rearranges the order of blocks in the image data; and a recording unit that records the rearranged block at a predetermined position of the recording device. .

  As described above, by performing clipping processing, resolution conversion processing, and rearrangement processing in units of blocks, it can be said that not only the processing operation itself is simplified, but also the load on the subsequent processing unit can be reduced. Further, when image data is stored in the middle of each process, the capacity required for the storage can be reduced.

  In addition, since sorting has already been performed in units of blocks, it is easy to generate addresses to the storage location, etc., and since sorting is performed, it is possible to store multiple data at once as continuous data. , Transfer efficiency is improved.

[Second Embodiment]
FIG. 11 is a block diagram showing a configuration of an image processing apparatus according to the second embodiment of the present invention. In the figure, 110 is a JPEG decoding circuit, 111 is an MCU clipping circuit, 112 is an MCU reduction circuit, 113 is an intra-block rotation circuit, 114 is a block buffer, 115 is a dot sequential circuit, 116 is an RGB conversion circuit, and 117 is a page. A buffer 118 is a pixel clipping circuit, and 119 is a resizing circuit. Only the parts different from the first embodiment will be described below.

  A JPEG compressed image is input from the input unit 1 to the JPEG decoding circuit 110, decoded, and output in units of MCU. The MCU clipping circuit 111 determines validity / invalidity for each MCU. The valid MCU is reduced to a desired size by the MCU reduction circuit 112 and stored in a predetermined position after the rotation of the block buffer 114 by the address generation of the intra-block rotation circuit 113. Is done. When the data for one MCU is stored in the block buffer, the dot sequential circuit 115 simultaneously reads Y, Cr, and Cb data from the block buffer 114 and converts them into RGB data by the RGB circuit 116. In addition, when Cr and Cb data are subsampled, they are interpolated and output during dot sequentialization. The converted RGB data is stored in the page buffer 117 at an address corresponding to the rotation mode in MCU units.

  When data for one page is stored in the page buffer 117, RGB data is read from the top of the page of the page buffer 117 according to a synchronization signal of a print engine (not shown). The pixel clipping circuit 118 clips the pixels that have not been clipped by the MCU clipping circuit 111, converts them to a desired size by the resizing circuit 119, and then outputs them to the print engine from the output unit.

  Next, a specific operation will be described.

  FIG. 12 is a diagram for explaining a specific example of image processing by the image processing apparatus according to the second embodiment. As shown in FIG. 2, in this embodiment, 2272 × 1704 pixels, sampling mode is 4: 2: 2 from a JPEG image, and 1600 × 1200 pixels are trimmed, and one page is output by a printer having 680 × 480 pixels. And The trimming range is a rectangular range (including the boundary) surrounded by the start point (300, 300) and the end point (1899, 1499), where the upper left vertex is (0, 0).

  Since the sampling mode is 4: 2: 2, the MCU size is 16 × 8. Accordingly, the number of MCUs in the horizontal direction is 2272/16 = 142, the number of MCUs in the vertical direction is 1704/8 = 213 (both may be 8 bits), and the horizontal direction MCU counter of the MCU clipping circuit 111 is set to 142 base. . Since the trimming start point coordinates are (300, 300) and the end point coordinates are (1899, 1499), the MCU effective range is (XS, YS) = (18, 37), and the end point coordinates are (XE, YE) =. (119, 187). Since the trimming area does not necessarily coincide with the MCU boundary, the start point (XS, YS) is a value that is rounded down to the end point (XE) when the start point and end point coordinates of the image data are divided by the block size. , YE) rounds up the value after the decimal point. Specifically, XS is the value shifted 4 bits to the right, YS is the value shifted 3 bits to the right, and XE is the value shifted 4 bits to the right, and the result of ORing the lower 4 bits before the shift (0 or 1) is added. The value YE is a value obtained by adding the result of ORing the lower 3 bits before the shift to the value shifted 3 bits to the right.

  The MCU clipping circuit 111 compares the above (XS, YS) = (18, 37) and (XE, YE) = (119, 187) with the value of the MCU counter. Set Crip_Valid to “H”.

  The MCU reduction circuit 112 reduces the MCU size within the effective range to a predetermined size. FIG. 13 is a block diagram showing a detailed configuration of the MCU reduction circuit 112. As shown in the figure, the MCU reduction circuit 112 includes two reduction circuits, a horizontal size reduction circuit 131 and a vertical size reduction circuit 132, and an effective signal generation circuit 133.

  The valid signal generation circuit 133 increments the internal horizontal counter and vertical counter in accordance with the input valid signal Crip_Valid. By these two counters, the pixel position in the block is grasped. In this embodiment, since the block size is 8 × 8 pixels, the horizontal counter and the vertical counter are both 3 bits, and the vertical counter is incremented by the carry of the horizontal counter. The value x of the horizontal counter is supplied to the horizontal size reduction circuit 131, and the value y of the vertical counter is supplied to the vertical size reduction circuit 132, which is used for selecting an interpolation coefficient. Further, the horizontal counter value x and the vertical counter value y are respectively input to the valid signal table and converted into valid signals. The valid signal table has an 8-bit output for each reduced size, and selects 1 bit from the value of each counter. Here, for simplicity of explanation, a table configuration is used. However, as described later, a valid signal may be directly generated from a counter value. The generated vertical and horizontal valid signals are ANDed with the input valid signal Crip_Valid at the same timing, and output as an MCU reduction circuit valid signal Resize_Valid.

FIG. 14 is a block diagram showing a specific configuration of the horizontal size reduction circuit 131 or the vertical size reduction circuit 132. The difference between the horizontal size reduction circuit and the vertical size reduction circuit is the delay amount of the buffer 144, which is one pixel in the case of the horizontal size reduction circuit and one line (8 pixels) in the case of the vertical size reduction circuit. Here, it is obtained by linear interpolation from the positions of adjacent pixels and interpolation pixels. For example, when the reduced sampling point is between the pixel A and the pixel B and the ratio of the distance from the sampling point to the pixel B and the distance from the sampling point to the pixel A is α: (1−α), The value P is
P = αA + (1−α) B (1)
It becomes.

  FIG. 15 is a diagram specifically showing sampling points after MCU reduction. Circles with halftone dots represent sampling points after reduction. 15, (a) is 7 × 7 pixels, (b) is 6 × 6 pixels, (c) is 5 × 5 pixels, (d) is 4 × 4 pixels, (e) is 3 × 3 pixels, ( f) shows a case where the image is reduced to 2 × 2 pixels. When the image is reduced to 1 × 1 pixel, it matches the position (x, y) = (3, 3) before the reduction. As is apparent from the figure, the cycle is 8 pixel cycles in both the vertical and horizontal directions. That is, the reduction is completed within the block. In order to remove aliasing noise, the JPEG decoding circuit 110 operates the Q table to eliminate high-frequency components that cause aliasing noise, and then performs inverse DCT. 1 × 1 pixel, 2 × 2 pixel, 4 × 4 It is assumed that the data of the corresponding position comes out as it is without interpolation in the pixel. In addition, the numbers surrounded by circles in the figure indicate the effective timing.

  FIG. 7 is a diagram showing interpolation coefficients (values obtained by multiplying by 256 to an integer) in the case of each sampling point shown in FIG. Here, α = 256 represents a case where the position of the input pixel coincides with the reduced sampling point. In this case, the input pixel data is output as it is without interpolation. As described above, since the interpolation circuit of the horizontal size reduction circuit and the vertical size reduction circuit are the same, the interpolation coefficient is also the same. Accordingly, in the vertical size reduction circuit 132, x in FIG. 7 is replaced with y.

  In FIG. 7, a column with a background dot indicates invalid data, and the valid signal at this timing is “L” (invalid). Therefore, any value may be set for the interpolation coefficient. Here, in order to simplify the circuit, an interpolation coefficient for invalid data is also defined. For example, all the coefficients are x ≧ 4 and the appearance order is reversed (α and (1-α) are also switched). Further, the coefficients when the size W after reduction is 6 and 3 are the same. For example, the coefficient table when W = 7 is set to (255, 219, 183, 146, 110, 73, 37, 0), and α is read from the left and (1-α) is read from the right. In order to reduce the number of bits in the coefficient table, when the coefficient is 255, the input data is output as it is without interpolation (through mode).

  Further, when the reduced size W is a power of 2, all are set to the through mode, and therefore no coefficient is required. Furthermore, since (1−α) is a 2's complement of α, it may be generated from bit inversion of α + 1. In this case, for example, the coefficient table at the time of W = 7 needs only four coefficients (0, 37, 73, 110). α is generated from bit inversion +1 of (1−α), and when x ≧ 4, the value read in reverse order is set to α, and the bit inversion +1 of the read value is set to (1−α).

  Note that when x = 7 (or y = 7), the invalid data is used to compensate for the delay caused by the buffer 144. Normally, a latency (delay) is generated for one pixel in the horizontal size reduction circuit and for eight pixels in the vertical size reduction circuit. Therefore, the interpolation output of the MCU reduction circuit 112 is output with a latency (delay) of 9 pixels, and the input valid signal and the horizontal size reduction valid signal are ANDed after matching the timing with the vertical size reduction valid signal. There must be. However, as described above, when x = 7 (or y = 7), the invalid operation data can be used to synchronize the internal operation timing with the input valid signal, and the timing adjustment is not necessary. However, when latching the outputs of the horizontal size reduction circuit and the vertical size reduction circuit, latency compensation for two pixels is necessary. Further, data of x = 0 (or y = 0) is not used as much as possible. This is because the influence of distortion due to the reduction of the high-frequency component of the DCT appears greatly in the peripheral part, so that the influence of the distortion is reduced by avoiding the use of pixels in the peripheral part.

The effective signal of the reduction circuit can be easily generated. That is,
When W = 1, x = 3 and "H"
When W = 2, the lower 2 bits of x = 01 and "H"
When W = 3, x = 1, 4, 6 and “H”,
When W = 4, LSB of x = 0 and "H"
When W = 5, "H" other than x = (2, 5, 7)
When W = 6, “H” except x = (3, 7),
When W = 7, “H” except x = (7),
It becomes.

  Further simplification can be achieved if x = 7 and the effective signal is always "L". For example, when W = 7, the determination is not necessary, and W = 6 may be an inversion of W = 1. Also, W = 5 is an inversion of W = 3 + 1, and 2 and 5 (1 and 6 when W = 3) have a one's complement relationship, so the MSB of x and the lower 2 bits are respectively It may be determined by performing EXOR.

  As described above, the block buffer 114 performs rotation and dot sequentialization within the block. In the case of JPEG, the MCU pixel size is maximum when the sampling mode is 4: 2: 0. At this time, the JPEG decoding circuit 110 has a luminance signal of 4 blocks (Y0, Y1, Y2, Y3) and a color difference signal of 1 block (Cr, Cb) for each 8 × 8 pixel block as an MCU. Input sequentially. The block buffer 114 has three block buffers (Y block buffer, Cr block buffer, and Cb block buffer), each of which is composed of individual RAMs (however, the addresses of the Cr block buffer and the Cb block buffer can be shared). Therefore, it may be configured on the same RAM by storing the data bus separately on the upper and lower sides of the data bus).

  The internal write horizontal counter, vertical counter, and block counter are incremented in accordance with the valid signal Resize_Valid input to the intra-block rotation circuit 113. The vertical and horizontal counters are used to grasp the pixel position in the block, and the block counter is used to grasp the block position. In this embodiment, since the block size is 8 × 8 pixels at the maximum, both the horizontal counter and the vertical counter are 3 bits. When the size is changed by the MCU reduction circuit 112, the horizontal counter and the vertical counter become counters for the size. That is, assuming that the size after reduction is W, a carry occurs when the counter is W-1, and returns to 0 (W base counter). The vertical counter is incremented by the carry of the horizontal counter. Since the maximum number of Y blocks in the MCU is 4, the block counter has 2 bits and is incremented by AND of the carry of the horizontal counter and the carry of the vertical counter.

  If the horizontal counter value for writing is Xw, the vertical counter value is Yw, and the block counter value is Bw, the write address to the Y block buffer is (Bw, Yw, Xw) (where "," are bits Indicates binding). Similarly, the write addresses of the Cr block buffer and the Cb block buffer are (Yw, Xw).

  FIG. 16 is a diagram showing how the luminance signal data in the specific example of FIG. 12 is stored in the Y block buffer. The numbers in the squares indicate the storage timing. In this embodiment, even when the MCU is reduced, the data is stored in units of 8 × 8 pixels before reduction without being rearranged. As a result, the address at the time of storage does not depend on the rotation mode, and address generation is facilitated (the control in the address generation unit is only carry control based on the Y, Cr, Cb block sequence and the size W after reduction). ).

  FIG. 17 shows how the color difference signal data in the specific example of FIG. 12 is stored in the Cr block buffer and the Cb block buffer (the Cr block buffer and the Cb block buffer may be configured on the same memory). FIG. The numbers in the squares indicate the storage timing. As shown in the figure, in this embodiment, when the MCU is reduced to ½ or less (W ≦ 4) by MCU reduction, the pixel size after reduction in the sub-sampled direction is doubled. That is, when the pixel size after reduction of the luminance signal is W, the size after reduction of the pixel data in the sub-sampled direction is 2W. With this configuration, the degradation of the resolution of the color difference signal due to the reduction is minimized.

  On the other hand, at the time of reading, an address is generated by a horizontal counter, a vertical counter, and a block counter for reading. The configuration of the counter differs depending on the sampling mode and the rotation mode. Here, in order to correspond to each sampling mode, the horizontal counter for the Y block buffer is composed of 3 bits, the vertical counter is composed of 4 bits, and the block counter is composed of 1 bit, and the C (Cr, Cb common) block buffer is composed of the horizontal counter. The counter is composed of 4 bits and the vertical counter is composed of 3 bits.

  For example, in the case of a Y block buffer counter, in the 4: 4: 4 mode, both the horizontal counter and the vertical counter operate as a W up / down counter (3 bits), and in the 4: 2: 2 mode, the horizontal counter is up W. The down counter and the vertical counter operate as a 2W base up / down counter (0 °, 180 °) or a W base up / down counter + 1 bit block counter (90 °, 270 °), and in 4: 2: 0 mode, The horizontal counter operates as a W base up / down counter, and the vertical counter operates as a 2W base up / down counter + 1 bit block counter. The counter configuration is switched by a carry (borrow) generation method. Note that the initial value at the time of down-counting is W-1 (W base) or 2W-1 (2W base).

  FIG. 18 is a diagram showing the relationship between the rotation mode and the reading position of the Y block buffer. In the figure, squares indicate blocks of 8 × 8 pixels, numbers in the squares indicate the input order of blocks, and small squares indicate block positions after reduction. An arrow indicates the direction in which data is read. A block indicated by a dot indicates a target block in the 4: 2: 2 mode.

The value of the horizontal counter for reading of the Y block buffer is Xr, the value of the vertical counter is Yr (or Yr '), the value of the block counter is Br, and the value of the horizontal counter for reading of the C (Cr, Cb) block buffer Is Xcr (or Xcr ′), and the value of the vertical counter is Ycr (or Ycr ′), the read address of each block buffer is expressed by the following equation from FIG. In the following description, “,” indicates a bit combination, and “!” Indicates a downcount.
(1) When the sampling mode is 4: 4: 4 (common to Y, Cr, Cb)
When 0 °: Yr, Xr
At 90 °! Xr, Yr
At 180 ° :! Yr! Xr
At 270 °: Xr! Yr
(2) Sampling mode = 4: 2: 2 [Y block address], (Yr 'is 2W base)
When 0 °: Yr ′ [0], Yr ′ [3..1], Xr
At 90 °: Br! Xr, Yr
At 180 ° :! Yr '[0] ,! Yr '[3..1] ,! Xr
At 270 ° :! Br, Xr ,! Yr
[Cr, Cb block address] (When W> 4, Xcr ′ and Ycr ′ are in 2W base)
At 0 °: Ycr, Xcr ′ [3..1]
At 90 °! Xcr, Ycr '[3..1]
At 180 ° :! Ycr ,! Xcr '[3..1]
At 270 °: Xcr! Ycr '[3..1]
[Cr, Cb block address] (When W ≦ 4, Xcr ′ and Ycr ′ are in 2W base)
When 0 °: Ycr, Xcr ′
At 90 °! Xcr, Ycr '
At 180 ° :! Ycr ,! Xcr '
At 270 °: Xcr! Ycr '
(3) Sampling mode = 4: 2: 0 [Y block address] (Yr 'is 2W base)
At 0 °: Br, Yr ′ [0], Yr ′ [3..1], Xr
At 90 °! Yr '[0], Br ,! Xr, Yr '[3..1]
At 180 ° :! Br ,! Yr '[0] ,! Yr '[3..1] ,! Xr
At 270 °: Yr ′ [0] ,! Br, Xr ,! Yr '[3..1]
[Cr, Cb block address] (When W> 4, Xcr ′ and Ycr ′ are in 2W base)
At 0 °: Ycr '[3..1], Xcr' [3..1]
At 90 °! Xcr '[3..1], Ycr' [3..1]
At 180 ° :! Ycr '[3..1] ,! Xcr '[3..1]
At 270 °: Xcr '[3..1]! Ycr '[3..1]
[Cr, Cb block address] (When W ≦ 4, Xcr ′ and Ycr ′ are in 2W base)
At 0 °: Ycr ′ [3..1], Xcr ′
At 90 °! Xcr '[3..1], Ycr'
At 180 ° :! Ycr '[3..1] ,! Xcr '
At 270 °: Xcr '[3..1]! Ycr '
Note that, by reversing the function of the predetermined counter (up-counting and down-counting) among the above counters, arbitrary vertical and horizontal mirror images can be obtained.

Further, it is possible to make the counters of the respective block buffers common by manipulating the valid signal. In this case, the horizontal counter, vertical counter and block counter for reading may be 3-bit, 4-bit and 1-bit binary counters, respectively, and the read address of the Y block buffer is
When 0 °: Br, Yr [0], Yr [3..1], Xr
At 90 °:! Yr [0], Br,! Xr, Yr [3..1]
When 180 °:! Br,! Yr [0],! Yr [3..1],! Xr
When 270 °: Yr [0],! Br, Xr,! Yr [3..1]
become that way. However, “,” indicates bit combination, and “!” Indicates bit inversion (the same applies hereinafter).

Also, the read address of the C (Cr, Cb) block buffer is
When W> 4 in 4: 2: 2 mode,
When 0 °: Yr [3..1], Yr [0], Xr [2..1]
At 90 °:! Yr [0],! Xr [2..1], Yr [3..1]
When 180 °:! Yr [3..1],! Yr [0],! Xr [2..1]
When 270 °: Yr [0], Xr [2..1],! Yr [3..1]
When W> 4 in 4: 2: 0 mode,
At 0 °: Br, Yr [3..2], Yr [0], Xr [2..1]
At 90 °:! Yr [0],! Xr [2..1], Br, Yr [3..2]
When 180 °:! Br,! Yr [3..2],! Yr [0],! Xr [2..1]
At 270 °: Yr [0], Xr [2..1],! Br,! Yr [3..2]
When W ≦ 4 in 4: 2: 2 mode,
When 0 °: Yr [3..1], Yr [0], Xr [1..0]
At 90 °:! Xr, Br, Yr [2..1]
When 180 °:! Yr [3..1],! Yr [0],! Xr [1..0]
At 270 °: Xr,! Br,! Yr [2..1]
When W ≦ 4 in 4: 2: 0 mode,
At 0 °: Br, Yr [2..1], Yr [0], Xr [1..0]
At 90 °:! Yr [0],! Xr [1..0], Br, Yr [2..1]
When 180 °:! Br,! Yr [2..1],! Yr [0],! Xr [1..0]
At 270 °: Yr [0], Xr [1..0],! Br,! Yr [2..1]
It becomes.

  Further, it is possible to output a mirror image in the horizontal direction only by reversing the address in the horizontal direction (X direction in FIG. 18), and a mirror image in the vertical direction only by reversing the address in the vertical direction (Y direction in FIG. 18). .

In this case, the valid signal is
When 0 °: Valid when Xr <W and Yr <W
When 90 °: Valid when Xr <W and Yr <W When 180 °: Valid when! Xr <W and! Yr <W When 270 °: Valid when Xr <W and! Yr <W Become.

Furthermore, in 4: 2: 2 mode, in order to reduce reading of invalid blocks, the Y block buffer read address is
When 0 °: 0, Yr [0], Yr [3..1], Xr
At 90 °: 0, Yr [3],! Xr, Yr [2..0]
When 180 °: 0,! Yr [0],! Yr [3..1],! Xr
At 270 °: 0,! Yr [3], Xr,! Yr [2..0]
When the read address of the C (Cr, Cb) block buffer is W> 4,
When 0 °: Yr [3..1], Yr [0], Xr [2..1]
At 90 °:! Xr, Yr [3..1]
When 180 °:! Yr [3..1],! Yr [0],! Xr [2..1]
At 270 °: Xr,! Yr [3..1]
When the read address of the C (Cr, Cb) block buffer is W ≦ 4,
When 0 °: Yr [3..1], Yr [0], Xr [1..0]
At 90 °:! Xr, Yr [3], Yr [1..0]
When 180 °:! Yr [3..1],! Yr [0],! Xr [1..0]
At 270 °: Xr,! Yr [3],! Yr [1..0]
You may make it set to. In this case, only the gray block in FIG. 18 is read out.

  In this embodiment, the internal rotation is performed at the time of reading the block buffer. However, as shown in FIGS. 19 and 20, the data is stored at the position rotated within the block at the time of data writing to the block buffer. May be. That is, FIG. 19 is a diagram showing a state in which the luminance signal data when rotating in the block at the time of writing is stored in the Y block buffer. FIG. 20 is a diagram illustrating a state in which the color difference signal data when rotated in the block at the time of writing is stored in the Cr block buffer and the Cb block buffer.

  Further, in this embodiment, since it is possible to identify in units of 8 × 8 pixels at the time of reading, it is possible to perform clipping in units of 8 × 8 pixels by manipulating the effective signal at the time of reading. In this case, the clipping is performed in three stages of MCU units, DCT block (8 × 8 pixels) units, and pixel units. Alternatively, clipping in MCU units may be omitted, and two-stage clipping in units of DCT blocks (8 × 8 pixels) and pixels may be used.

  Further, the dot-sequential YCrCb data is converted into RGB data by the RGB conversion circuit 116 and stored in the page buffer 117.

  Furthermore, storage in the page buffer 117 is performed in units of the MCU.

The head address of the MCU can be generated from the horizontal MCU counter and the vertical MCU counter in the same manner as the rearrangement in the block. However, in this case, a plurality of multiplications are required, and address generation becomes complicated. Therefore, the storage address of the page buffer is generated by a method of obtaining the address of the next pixel by adding the differential address to the address of the current pixel (hereinafter referred to as “differential addressing method”). That is, when the address of the immediately preceding pixel (original pixel) is A (n-1) and the difference address (address difference value) is D, the address A (n) of the current pixel is
A (n) = A (n-1) + D (2)
And That is, the next pixel data is stored at the buffer address indicated by A (n), and the pixel rearrangement process is performed.

  The difference address D is switched in the sequence shown below depending on the rotation and sampling mode. Here, the differential address at the MCU line end (the last pixel of the rightmost MCU) is D1, the differential address at the MCU end (MCU last pixel) other than the MCU line end is Dm, and the intra-block line end other than the MCU end (within the MCU) Let Db be the difference address in the rightmost pixel). Also, the horizontal counter value in the MCU is x, the vertical counter value in the MCU is y, the horizontal MCU counter value is Mx, the vertical MCU counter value is My, and the rotated MCU size is Wx × Wy, the number of MCUs in the horizontal direction is Wm, and the number of MCUs in the vertical direction is Hm. The switching timing of each differential address at this time is shown below.

D = Dl: When x = Wx−1, y = Wy−1, z = m−1 D = Dm: When x = Wx−1, y = Wy−1, z ≠ m−1 D = Db: When x = Wx-1, y ≠ Wy-1, D = 1: Other than the above, where z = Mx, m = Wm: 0 °, when rotated 180 °,
z = My, m = Hm: 90 ° and 270 °.

Also, each difference address and the initial value A (0) of the address is assumed to be As when the top address of the page buffer is As
Db = Wm × Wx−Wx + 1: common to each rotation mode [0 ° rotation]
A (0) = As
Dl = 1
Dm = −Wm × Wx × (Wy−1) +1
[90 ° rotation]
A (0) = As + (Wm−1) × Wx
Dl = −Wm × Wx × (Hm × Wy−1) −2Wx + 1
Dm = Wm × Wx−Wx + 1
[180 ° rotation]
A (0) = As + Wm × Wx × (Hm−1) × Wy + (Wm−1) × Wx
Dl = −Wm × Wx × (2Wy−1) + (Wm−2) × Wx + 1
Dm = −Wm × Wx × (Wy−1) −2Wx + 1
[Rotate 270 °]
A (0) = As + Wm × Wx × (Hm−1) × Wy
Dl = Wm × Wx × ((Hm−2) × Wy + 1) +1
Dm = −Wm × Wx × (2Wy−1) −Wx + 1
It becomes.

  For example, in the example of FIG. 12, an image of 101 × 151 MCU is obtained by the MCU clipping circuit, an MCU of 16 × 8 pixels is reduced to an MCU size of 6 × 3 pixels by the MCU reduction circuit, and 90 × by the intra-block rotation circuit. ° Rotated and output. Therefore, the MCU size after the MCU reduction rotation is 3 × 6. That is, Wx = 3, Wy = 6, Wm = 151, and Hm = 101. Thus, if the page buffer start address As = 0, then A (0) = 450, Dl = −274070, and Dm = Db = 451. The horizontal counter x in the MCU is a ternary counter, and the vertical counter y in the MCU counts up by the carry of this counter. The vertical counter y in the MCU is a hexadecimal counter, and the MCU counter My counts up with the carry of this counter. The MCU counter My is a 101-ary counter that counts up from 0 to 100 and returns to 0. Note that a horizontal MCU counter is not required.

  The differential address D is 1,451,1,451-1,..., 1,1, -274070,... Once in 451, 1818 times (101 × 6 × 3 times). Once switched to -274070. That is, switching is performed in synchronization with the carry of the counter.

  As described above, the parameters Wx, Wy, Wm, and Hm are determined by the rotation mode, the MCU size after MCU reduction and rotation, and the number of vertical and horizontal MCUs of the image, and the horizontal counter in the MCU and the vertical direction in the MCU. The configuration of the counter and MCU counter is determined. The differential addresses Dl, Dm, and Db are uniquely determined by the parameters and are constant as long as the parameters are not changed. Therefore, by storing the differential addresses Dl, Dm, Db and the initial value A (0) of the address in the register, it is possible to generate an address without complicated calculations. Further, as described above, since D = 1 is normally set (continues in the direction in which the address increases), it is possible to improve the memory usage efficiency by burst write.

  Since it is stored in a rotated state on the page buffer, reading from the page buffer 117 is performed sequentially (continuously) from the top address As of the page buffer regardless of the mode. The read image data is output from the output unit to the print engine after the pixels not clipped by the MCU clipping circuit 111 are clipped by the pixel clipping circuit 118 and converted to a desired size by the resizing circuit 119. The For example, in the example of FIG. 12, it is stored in the page buffer with a size of 453 × 606 pixels. This is clipped to 450x600 pixels (3/8 times 1200x1600), which is the user setting area, and further enlarged (16/15 times) to 480x640 pixels, which is the output image size. Output.

  In this embodiment, the size is reduced to the vicinity of the target size by the MCU reduction circuit. In the example of FIG. 12, since 1200 × 1600 pixels are desired to be 480 × 640 pixels, the reduction ratio is 480/1200 = 0.4. This value is closest to 3/8 (= 0.375) among the reduction ratios of 1/8 (= 0.125) units. Therefore, 3/8 × 16/15 = 0.4 is realized by reducing to 3/8 by the MCU reduction circuit and 16/15 times by the resizing circuit in the subsequent stage.

  In the case of outputting an image signal in accordance with the above-described print engine synchronization signal, if the reduction of 1/2 or less is executed at a time, the sampling points after the reduction are separated, so that the readout of pixels used for interpolation is discontinuous. turn into. Therefore, when the page buffer is configured on the DRAM, the memory use efficiency is lowered. On the other hand, in this embodiment, when the reduction ratio is 1/16 or more, the reduction ratio of the resizing circuit 119 becomes 1/2 or more, and it can be handled by continuous reading (when the resizing circuit 119 is equipped with a line memory). Just read the image data from the top). That is, the continuity of memory access is improved and the memory usage efficiency is improved. In addition, the configuration in which the Q table of the JPEG decoding circuit is operated to eliminate high-frequency components that are aliasing noise and then inverse DCT is performed, thereby eliminating the need for a prefilter for aliasing noise removal. In addition, when the reduction ratio is 1/2 or less, interpolation by the sub-sampling of the color difference signal is not performed, so that deterioration of the resolution of the color difference signal can be suppressed.

In the above description, rearrangement is performed in the page buffer, but not limited to this, a band buffer (a part of the page buffer) may be used. As described above, in this embodiment, the address difference value between the block end pixel and the next pixel of the last block of the block line stored in the buffer is D1, and the address difference value between the block end pixel and the next pixel of the block is Dm. When the address difference value between the rightmost pixel of the block and the next pixel is Db, the address difference value D is
D = Dl for the block end pixel of the last block on the block line,
D = Dm if D = Dl and not the block end pixel of the last block on the block line,
Neither D = Dl nor D = Dm, and D = Db in the case of the rightmost pixel of the block other than the last block on the block line,
In other cases, D = 1
It is.

  Although the embodiments of the present invention have been described above, the present invention is not limited to this, and the present invention can be applied to a system including a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.). You may apply to the apparatus (for example, a copying machine, a facsimile machine, etc.) which consists of apparatus.

  Another object of the present invention is to supply a computer-readable recording medium (or storage medium) that records a program code (computer program) of software that implements the functions of the above-described embodiments to a system or apparatus, and the system. Needless to say, this can also be achieved by reading and executing the program code stored in the recording medium by the computer (or CPU or MPU) of the apparatus. In this case, the program code itself read from the recording medium realizes the functions of the above-described embodiment, and the recording medium on which the program code is recorded constitutes the present invention. Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an operating system (OS) running on the computer based on the instruction of the program code. It goes without saying that a case where the function of the above-described embodiment is realized by performing part or all of the actual processing and the processing is included.

  Furthermore, after the program code read from the recording medium is written into a memory provided in a function expansion card inserted into the computer or a function expansion unit connected to the computer, the function is based on the instruction of the program code. It goes without saying that the CPU or the like provided in the expansion card or the function expansion unit performs part or all of the actual processing and the functions of the above-described embodiments are realized by the processing.

  When the present invention is applied to the recording medium, program code corresponding to the flowchart described above is stored in the recording medium.

Claims (22)

Block image processing means for executing image processing in units of blocks for image data input in units of blocks;
Conversion means for converting the image data processed in block units into a raster shape;
Pixel image processing means for performing image processing in units of one pixel using the image data converted into a raster shape,
An image processing apparatus characterized in that image processing to be performed on input image data is performed in two stages of the block image processing means and the pixel image processing means.   The image processing apparatus according to claim 1, wherein the block image processing means extracts image data in a specified range from the input image data.   The image processing apparatus according to claim 1, wherein the block image processing unit converts a resolution of the image data. The conversion means is a pixel rearrangement process in each block for the image data converted into a raster shape for each block;
The image processing apparatus according to claim 1, wherein the conversion processing is performed in two stages including rearrangement processing of the blocks themselves.   The image processing apparatus according to claim 4, wherein the conversion unit rearranges the block and the pixel at a rotation angle of 90 °.   When extracting the image data of the specified range from the image data as a rectangular range determined by determining start point coordinates and end point coordinates, when extracting the rectangular image data of the specified range in block units The start point coordinates and end point coordinates of each block are coordinates based on values obtained by rounding up the numerical values after the decimal point when the start point coordinates and the end point coordinates of the image data are divided by the block size of the block. The image processing apparatus according to claim 2.   3. The image processing apparatus according to claim 2, wherein the extraction processing for extracting the image data in the specified range by the block image processing means is performed by controlling an effective signal related to the input image data.   When the block size is Bx × By, the horizontal size after the resolution conversion of the block unit by the block image processing means is an integer from 1 to Bx, and the vertical size after the resolution conversion is The image processing apparatus according to claim 3, wherein the image processing apparatus is any integer from 1 to By.   9. The image processing apparatus according to claim 8, wherein the resolution conversion processing in units of blocks by the block image processing means is performed independently in a horizontal direction and a vertical direction. When the horizontal reduction ratio in the resolution conversion process is Rx and the vertical reduction ratio is Ry, the output size of the image data in the block-unit resolution conversion process is:
n / Bx ≦ Rx <(n + 1) / Bx, where n is an integer from 1 to Bx,
m / By ≦ Ry <(m + 1) / By, where m is an integer from 1 to By,
The image processing apparatus according to claim 8, wherein n × m satisfies the following condition.   The horizontal and vertical resolution conversion processing by the block image processing means switches the interpolation coefficient according to the position on the block of the input image data, and sets data other than the interpolation point as invalid data. The image processing apparatus according to claim 9.   The block image processing means uses invalid data at the time of pixel calculation at the right end of the block in the horizontal resolution conversion processing, and invalid data at the time of pixel calculation at the bottom end of the block in the vertical resolution conversion processing. The image processing apparatus according to claim 11.   In the horizontal and vertical resolution conversion processing by the block image processing means, when the interpolation coefficient is a maximum value that can be set, a pixel multiplied by the coefficient that becomes the maximum value is used as an interpolation output. The image processing apparatus according to claim 11 or 12.   In the horizontal and vertical resolution conversion processing by the block image processing means, when the interpolation coefficient is 0, a pixel corresponding to a pixel to be multiplied by a coefficient of 0 and not multiplied by 0 is interpolated The image processing apparatus according to claim 11, wherein the image processing apparatus is an output. The rearrangement processing of the pixels in the block by the conversion means is performed such that the address of the original pixel is A (n−1), the address difference value selected in a predetermined sequence is D, and the address of the next pixel is A (n )
A (n) = A (n-1) + D
The image processing apparatus according to claim 4, wherein the next pixel data is stored at an address of a buffer indicated by The address difference value between the block end pixel and the next pixel of the last block of the block line stored in the buffer is D1, the address difference value between the block end pixel and the next pixel of the block is Dm, and the block right end pixel of the block When the address difference value with the next pixel is Db, the address difference value D is
D = Dl for the block end pixel of the last block on the block line,
D = Dm if D = Dl and not the block end pixel of the last block on the block line,
Neither D = Dl nor D = Dm, and D = Db in the case of the rightmost pixel of the block other than the last block on the block line,
In other cases, D = 1
The image processing apparatus according to claim 15, wherein: Block decoding means for decoding image data from encoded data encoded for each block;
Block image processing means for image-processing the decoded image data in units of processing blocks at the time of encoding;
Conversion means for converting the image data subjected to image processing in units of the processing blocks into a raster shape;
Pixel image processing means for executing image processing in units of one pixel using the image data converted into a raster shape,
An image processing apparatus characterized in that image processing to be performed on input image data is performed in two stages of the block image processing means and the pixel image processing means. A block image processing step for executing image processing in units of blocks for image data input in units of blocks;
A conversion step of converting the image data processed in block units into a raster shape;
A pixel image processing step of performing image processing in units of one pixel using the image data converted into a raster shape,
An image processing method characterized in that image processing to be applied to input image data is performed in two stages, the block image processing step and the pixel image processing step. A block decoding step of decoding image data from block-coded encoded data;
A block image processing step of image-processing the encoded image data in units of processing blocks at the time of block encoding;
A conversion step of converting the image data subjected to image processing in units of the processing blocks into a raster shape;
A pixel image processing step of performing image processing in units of one pixel using the image data converted into a raster shape,
An image processing method, wherein image processing to be performed on input image data is performed in two stages, the block image processing step and the pixel image processing step. At the computer
A block image processing procedure for executing image processing in units of blocks for image data input in units of blocks;
A conversion procedure for converting the image data processed in units of blocks into a raster shape;
A pixel image processing procedure for performing image processing in units of one pixel using the image data converted into a raster shape,
A computer program for executing image processing to be applied to input image data in two stages of the block image processing procedure and the pixel image processing procedure. At the computer
A block decoding procedure for decoding image data from block encoded data;
A block image processing procedure for performing image processing on the decoded image data in units of processing blocks at the time of block encoding;
A conversion procedure for converting the image data subjected to image processing in units of the processing blocks into a raster shape;
A pixel image processing procedure for performing image processing in units of one pixel using the image data converted into a raster shape,
A computer program for executing image processing to be applied to input image data in two stages of the block image processing procedure and the pixel image processing procedure.   A computer-readable recording medium storing the computer program according to claim 20 or 21.

Images