JP5597272B2 - Arithmetic method, arithmetic device, image processing device, and program - Google Patents

Description

  The present invention relates to a calculation method, a calculation device, an image processing device, and a program. In particular, the present invention is preferably used when a plurality of image data and correction data in the main memory are divided and handled for image processing, and necessary data are sequentially read out to perform image processing and correction processing. It is

  Conventionally, when image formation is performed, local (neighboring) image processing such as spatial filter processing has been performed. This local image processing is image processing in which some calculation is performed using all the pixels included in the spatial filter region including the pixel to be processed. In the following description, a pixel to be processed is abbreviated as a target pixel.

  For example, spatial filter processing such as edge enhancement processing and blurring processing is performed on the digital image data 300 in FIG. As a conventional technique of such local image processing, there is a technique in which digital image data is divided for each area, and local image processing is performed for each divided area (see Patent Documents 1 to 4). In general, in this technique, as shown in FIGS. 11A to 11D, the entire digital image data 300 is divided into strips (strips), and various images are sequentially obtained for each region. Processing is performed. This elongated and divided area is called a band area, a storage area where the band area is expanded is called a band memory, and an action of dividing an image is called a band division. The band memory is not necessarily reserved as a storage area in the main memory, and may be allocated in any storage area on the system. However, for the sake of brevity, the band memory is used as the main memory. An example of securing in the memory will be described.

  The coordinate system (main scanning direction-sub-scanning direction) of the digital image data defines a new coordinate system (band area coordinate system) of the length direction and the height direction, and the band area is expressed by (length Bdl × height). Bdh). When the image is band-divided in the sub-scanning direction, the band region is expressed by (length Bdl × height Bdh) as shown in FIG. On the other hand, when the image is band-divided in the main scanning direction, the band region is expressed by (length Bdl × height Bdh) as shown in FIGS.

  That is, the length Bdl of the band region is always a value of the width in the main scanning direction or the height in the sub scanning direction of the digital image data, and the height of the band is an arbitrary value.

  The band processing will be described in more detail with reference to the example of FIG. In addition, since it is fundamentally the same also in FIG. 12, the overlapping description is avoided.

  First, the first band region 301 shown in FIG. 11A is developed in the band memory on the main memory to perform image processing. Next, the second band region 302 shown in FIG. 11B is overwritten and expanded on the band memory in which the first band region 301 is expanded, and image processing is performed. Further, the third band area 303 shown in FIG. 11C is overwritten and expanded on the band memory where the second band area 302 is expanded, and image processing is performed. Finally, the fourth band area 304 shown in FIG. 11 (d) is overwritten on the band memory in which the third band area 303 is expanded, and image processing is performed. As is clear from FIGS. 11A to 11D, the band regions 301 to 304 have the same length, but the heights need not be the same. The height of the band memory which is a storage area secured in the main memory is determined by the band area having the largest size in the height direction (in the case of FIG. 11, the first to third band areas 301 to 303). The 12 is the same as FIG. 11 except that there are only three band divisions.

  In such a technique, in order to perform local image processing without a gap between the band regions, each band region is devised so that a part thereof overlaps each other at a boundary with an adjacent region. Further, in Patent Document 4, an extra pixel is scanned one by one in the same direction as the height of each band area, and the capacity of a delay memory that holds a target pixel necessary for local image processing is set to the height of each band area. By defining the size of the delay memory, the delay memory can be saved.

US Pat. No. 6,587,158 JP 2000-31327 A Japanese Patent No. 3733826 JP 2006-139606 A

  However, looking over the field of image processing methods and apparatuses, there are many image processes that cannot be handled only by considering local (neighboring) image processing such as the spatial filter processing described in Patent Documents 1 to 4 described above. An example is shown below with reference to FIG.

  First, a correction process 1010 for correcting individual differences of reading elements of the sensor / device with respect to an image read by an image reading apparatus such as a scanner shown in the upper part of FIG. Information necessary for such correction processing 1010 is, for example, read minimum value data and read maximum value data for each reading element of the sensor device. Based on these pieces of information, the read image to be corrected is corrected for each pixel data (pixel value) by, for example, calculation using the following (Equation 1).

X = (P-MIN) / (MAX-MIN) × 1023 (Formula 1)
Here, X is a correction pixel value (X is a 10-bit digital signal), P is a reading pixel value, MIN is a reading minimum value, and MAX is a reading maximum value.

  That is, the data necessary for the correction process 1010 is three kinds of different information, ie, minimum value data, maximum value data, and pixel data of the read image that are different for each pixel. When the sensor device of the scanner is a line sensor in which the reading elements for each pixel are combined into one line, this sensor moves in the sub-scanning direction to read a two-dimensional image. In the case of such a reading method, in the main scanning direction, different minimum value data and maximum value data are arranged for each pixel, and in the sub scanning direction, in the case of pixels having the same position in the main scanning direction, It becomes the same minimum value data and maximum value data.

  Further, as a correction process similar to the above, there is a correction process for a two-dimensional display device used in a flat-screen television such as a liquid crystal television or a plasma television. In the case of a two-dimensional display device, since individual differences of display elements spread in two dimensions are different, correction processing is performed using two-dimensional correction data instead of one-dimensional correction data as in the correction processing 1010. become. In the correction processing as described above, it is necessary to pay attention because the correction data repetition method differs depending on the device shape (how elements are spread).

  In addition, there is a synthesis process 1020 shown in the middle part of FIG. 13, for example, for synthesizing a plurality of rendering images according to synthesis data (α value). When there are two rendering images to be combined, for example, a combining process 1020 is performed for each pixel using a combining expression such as the following (Expression 2).

X = P2 * A + P1 * (1024-A) (Formula 2)
Here, X: the pixel value after synthesis (when the α value is a 10-bit digital signal), P1: the pixel value of the first rendering image, P2: the pixel value of the second rendering image, A: for each pixel of the synthesis data α value. Note that the composite data (α value) may be the same for all pixels of the rendered image, or may be two-dimensional composite data (α value) that differs for each pixel of the rendered image.

  In the case of the composite data (α value) as described above, in the composite processing 1020 as described above, when the number of rendering images to be composited is N, the number of pixel data and composite data (α value) necessary for the processing is the above. It goes without saying that it is different from the example.

  Further, as shown in the lower part of FIG. 13, for example, a plurality of field images continuous in the time axis direction are referred to, such as interlace / progressive conversion of moving images (hereinafter abbreviated as I / P conversion 1030). In some cases, adaptive spatial filtering may be performed. In general, in image processing such as I / P conversion 1030 that refers to a plurality of field images, high-quality image processing can be performed by performing adaptive processing using more field images.

  However, the plurality of field images are generally stored in an external memory such as an inexpensive DRAM connected to a semiconductor chip (ASIC or the like) on which an image processing circuit is mounted. For this reason, it is necessary to read out a plurality of field images from the external memory each time for I / P conversion and input them to the I / P conversion circuit. In many cases, the memory bandwidth allowed when reading a multi-field image from an external memory is naturally limited due to product cost constraints. Therefore, for example, it is not possible to read out a high-resolution field image such as full high-definition (1920 × 1080i) indefinitely and perform I / P conversion.

  On the other hand, it is assumed that there is an image processing apparatus having a memory band that can read up to three field images of the full high-definition resolution, for example, for I / P conversion. In this image processing apparatus, if a low resolution moving image such as NTSC (720 × 480i) is subjected to I / P conversion, a high-definition I is referred to by referring to more field images than the full high definition resolution moving image. It is obvious that / P conversion can be performed.

  In the I / P conversion 1030 as described above, it is important to switch the image processing method flexibly according to the resolution and the number of reference fields in order to perform as high-quality image processing as possible when the usable memory bandwidth is constant. It becomes.

  What is common to the image processing shown in FIG. 13 described above is that it is not sufficient to perform image processing considering only one piece of image data in which pixel data is two-dimensionally arranged. In some cases, correction data is required, composite data is required, and in other cases, the number of image data to be used is arbitrary. Further, the image data, correction data, and composite data necessary for the image processing differ in acquisition method depending on the sensor shape. For example, it is necessary to repeatedly read from a storage device such as an external memory, the number of rendering images to be read differs depending on the number specified by the user, and the number of field images to read differs depending on the input resolution of the broadcast wave.

  Incidentally, in recent years, the resolution of digital image data has increased in order to achieve high image quality for the purpose of device differentiation, and the amount of data processing in image processing has increased accordingly. In addition, there is a wide variety of information required to improve product image quality and enhance functions. However, in product development, it is desired to meet various image processing requirements within limited cost constraints. In other words, there is a demand for an apparatus and method that can flexibly realize the various image processes described above while keeping the memory bandwidth and circuit scale constant.

  However, the above-described prior art refers to the capacity saving of the delay memory for the spatial filter processing by reading one piece of image data from the external memory for each band region when performing the local image processing. However, for a plurality of correction data, a plurality of image data, and a plurality of continuous field images for one image data, an examination from the viewpoint of flexibly changing the band area according to the resolution and the quantity of each data is not possible. Not done. Here, the quantity of data is the number of reference fields, the number of rendered images, and the number of correction data.

  The present invention realizes various types of image processing within a certain memory bandwidth while realizing the capacity saving of the delay memory as in the prior art. Provide an arithmetic method, an arithmetic device, and an image processing device that can flexibly cope with the above.

The calculation method of the present invention for solving the above problem divides input image data into a plurality of band areas, supplies partial image data in the divided band areas to an image processing means via a buffer, An arithmetic method for an image processing apparatus that performs image processing with reference to reference data,
Obtaining a transfer unit of the partial image data;
The first data amount data in the partial image data and the second data amount data in the reference data referred to for processing the first data amount data fit within the capacity of the buffer. A second acquisition step of acquiring the first data amount as the transfer amount of the partial image data based on the capacity of the buffer and the transfer unit ;
And a third acquisition step of acquiring the height of the band area based on the transfer amount and the transfer unit.

  According to the present invention, image data to be processed can be divided into a plurality of band areas with a small circuit scale, and image processing can be sequentially performed for each divided band area. In addition, the height of the band region can be flexibly changed for various image processing that has not been realized in the prior art, and image processing corresponding to an arbitrary resolution can be realized with a small circuit scale.

It is a block diagram which shows an example of the whole structure of an image processing apparatus. It is a figure explaining the example of storing to RAM of various data. It is a figure explaining the example of storing the pixel value of image data. It is a block diagram which shows an example of the circuit structure of an image process part. It is a figure which shows the operation example until it stores data required for image processing from RAM to a buffer, and outputs pixel data from an image processing input circuit. It is a flowchart which shows an example of an image processing process. It is a figure which shows an example of a structure of the image processing input circuit of an image processing apparatus. It is a figure which shows an example of operation | movement of the image processing input circuit of an image processing apparatus. It is a figure which shows an example of operation | movement of the image processing input circuit of an image processing apparatus. It is a flowchart which shows the example of an output of pixel data. It is a figure explaining an example of an image data format. It is a figure explaining an example of operation of band processing. It is a figure explaining an example of operation of band processing. It is a figure explaining the example of the conventional image processing.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

[Example 1]
In the first embodiment, image processing specialized for band processing is performed. As a result, the capacity of the delay memory in local (neighboring) image processing such as spatial filter processing described in Patent Documents 1 to 4 can be reduced, and the circuit scale can be reduced and high speed can be achieved by using a plurality of pixels described in Patent Document 4 as one processing unit. Realize. At the same time, correction processing and image processing of image data with reference to at least one of a plurality of image data and correction data are realized instead of processing one piece of image data. In order to achieve such an object, in the first embodiment, image processing is performed in a band area coordinate system different from the digital image data coordinate system (main scanning direction-sub-scanning direction).

<Overall configuration of image processing apparatus>
FIG. 1 is a block diagram illustrating an example of the overall configuration of the image processing apparatus according to the first embodiment.

  In FIG. 1, the image reading unit 120 includes a lens 124, a CCD sensor 126, an analog signal processing unit 127, and the like. In the image reading unit 120, image information of the original 110 formed on the CCD sensor 126 through the lens 124 is converted into analog electric signals of R (Red), G (Green), and B (Blue) by the CCD sensor 126. The The image information converted into the analog electric signal is input to the analog signal processing unit 127, and is subjected to analog / digital conversion (A / D conversion) after being corrected for each color of R, G, B. In this way, a digitized full color signal (referred to as digital image data) is generated.

  The generated digital image data is stored in the RAM 106 of the CPU circuit unit 100 via the shared bus 190 by a DMAC (Direct Memory Access Controller) 192 whose operation is set in advance by the CPU 102. Next, the CPU 102 reads out the digital image data set in the DMAC 194 and stored in the RAM 106 and inputs it to the image processing unit 150.

  The image processing unit 150 performs correction of individual differences of reading elements of a sensor device such as a scanner and color correction such as input gamma correction on the input digital image data, and normalizes the read image to be constant. Create standard image data. Then, the digital image data subjected to these processes is stored again in the RAM 106 by the DMAC 196 in which the writing operation has been set in advance. In some cases, the image processing unit 150 performs image processing for printing such as input color correction processing, spatial filter processing, color space conversion, density correction processing, and halftone processing on the input digital image data. To create printable image data. Then, it is stored in the RAM 106 by the DMAC 196 as described above.

  Finally, the CPU 102 reads the digital image data that has been set in the DMAC 198 and stored in the RAM 106, and outputs it to the image printing unit (printer) 170. The image printing unit (printer) 170 includes, for example, a print output unit (not shown) such as a raster plotter using an inkjet head, a thermal head, or the like, and an image is printed on paper by an input digital image signal. Record. Or in Example 2 mentioned later, it is displayed and output by video display parts 160, such as a display.

  The CPU circuit unit 100 includes a CPU 102 for arithmetic control, a ROM 104 that stores fixed data and programs, a RAM 106 that is used for temporarily storing data and loading programs, and an external storage device 108. The CPU circuit unit 100 controls the image reading unit 120, the image processing unit 150, the image printing unit (printer) 170, and the like, and comprehensively controls the sequence of the image processing apparatus of this embodiment. The external storage unit 108 is a storage medium such as a disk that stores parameters, programs, and correction data used by the image processing apparatus of the present embodiment. The data, programs, and the like in the RAM 106 are loaded from the external storage unit 108. It does not matter.

<Correction process>
First, a description will be given of a case in which correction processing is performed on an image read by the image reading unit 120 in FIG. In the case of such image processing, the minimum read value data and the maximum read value data for each reading element of the sensor device are required as described above. These various data are temporarily stored in the RAM 106. In general, the RAM 106 is often composed of an inexpensive DRAM. Therefore, in order to read and write such various data via the DMAC as described above, it is desirable to access data in units that allow the DRAM to read and write without degrading performance.

(Data configuration example of RAM 106)
A method of storing data in the RAM 106 will be described with reference to FIG.

  In FIG. 2, IMG_AREA_STR_ADDR to IMG_AREA_END_ADDR are areas for storing various data necessary for image processing. In the example of FIG. 2, six types of data from S0_IMG to S5_IMG are stored in this area. The minimum unit of the capacity of each stored data is 32 bits × 8 words of 32 bytes so that the data can be accessed without degrading the performance of the DRAM. Naturally, the storage capacity of data from S0_IMG to S5_IMG is an integral multiple of 32 bytes.

  For example, in the case of two-dimensional image data, an area corresponding to the maximum image area (window's area) 440 is indicated by 420 on the memory. An area of S0_IMG (image's area) 430 that fits in the area is stored in an area indicated by 410 on the memory. Reference numerals 422, 424, and 426 denote areas on the memory corresponding to the maximum image area in which S1_IMG to S3_IMG can be accommodated.

  When the line sensor device 450 of the scanner reads the original while moving in the sub-scanning direction (v) with respect to the original 490, correction data (MAX) in the main scanning direction (h) of the line sensor device 450 is read. 470 and correction data (MIN) 475 are distributed. The data is stored in S4_IMG 460 and S5_IMG 465 on the RAM 106 in the example of FIG.

  Further, the data structure will be described in detail with reference to FIG.

  Reference numeral 500 denotes image data of the area of the above-described S0_IMG (image's area) 430. The area 510 is the above-mentioned data of 32 bytes, and 8 pixel data indicated by 520 is packed therein. Further, as indicated by 530, each pixel data is packed with 10-bit R (Red), G (Green), and B (Blue) data. In the example of FIG. 3, the remaining 2 bits are not used as invalid data in order to simplify data access of the DRAM. The range of the thick frame 500 in FIG. 3 means an image having a size of 8M × N. In the case of the scanner sensor device correction data stored in S4_IMG460 and S5_IMG465 described above, the data is in the range of coordinates (0,0) to (8M, 0) because it is a line sensor of one line. .

  The image data and the correction data of the read image read by the image reading unit 120 are stored in the RAM 106 as described with reference to FIGS.

(Configuration example of image processing unit 150)
FIG. 4 is a block diagram illustrating a configuration example of the image processing unit 150 according to the first embodiment.

  The CPU 102 activates the DMAC 194, reads the read image data and correction data stored as described above from the RAM 106, and is an example of the main circuit configuration of the image processing apparatus of the present embodiment. Enter 150.

  The read image data and the correction data are input to the image processing unit 150 via the DMAC 194 and the bus 205. As will be described in detail later, the image processing input circuit 220 receives various types of data, and from the packed read image data, correction data (MAX), and correction data (MIN) described in FIG. Data (pixel value) is extracted in the order described later. The image processing circuit (1) 230, the image processing circuit (2) 240,..., And the image processing circuit (P) 280 are then sent to perform various correction processes (or image processing). Then, in the image processing output circuit 290, data repacked into 32-byte data shown in FIG. 3 is created, and the corrected image data (image processing) is written back to the RAM 106 via the bus 295 via the DMAC 196. . In any one of the image processing circuit (1) 220 to the image processing circuit (P) 290, the individual difference of the reading elements of the sensor device shown in the above (Expression 1) is corrected. Of course, image processing such as input color correction processing, spatial filter processing, color space conversion, density correction processing, and halftone processing is also performed.

<Data Transfer from RAM 106 to Image Processing Input Circuit 220>
FIG. 5 is a diagram for explaining details of a method of inputting the above-described image data and correction data from the RAM 106 to the image processing input circuit 220 when the image processing method of the first embodiment is used.

  Image data surrounded by a dotted line 605 is stored in the RAM 106. Reference numeral 600 in FIG. 5A denotes an overall image data image. A band region 610 is extracted from the image data, and the image processing method according to the first embodiment is performed. As shown in FIGS. 5B and 5C, the band area 610 to be processed has a band area height Bdh of 16 lines (16 pixels) in the band area coordinate system, and the length of the band area. Bdl is 8 × M pixels. First, the CPU 102 described in (d) of FIG. 5 sets, for the DMAC 194, the start address, the continuous read amount, the increment address, and the number of repetitions of the band area 610 of the RAM 106 via the shared bus 190. Here, the continuous read amount represents how many times 32 bytes of data are read continuously. In the example shown in the figure, the start address is S0_IMG_STR_ADDR in FIG. 2, the continuous read amount is 1 time (32 bytes), the increment address is the data amount of 1 line (32 bytes × M), and the number of repetitions is 16.

  In addition, as shown in FIG. 5D, a shared buffer 660 that allows data access from both modules exists between the DMAC 194 and the image processing input circuit 220. Therefore, the CPU 102 also instructs the DMAC 194 via the shared bus 190 also the write start address of the shared buffer that is the write destination of the acquired data. In the example of FIG. 5, the data is temporarily stored in the buffer 662, so that S0_BUF_STR_ADDR is an address to be set. In response to an instruction from the CPU 102, the DMAC 194 reads any one of the area (1) 640 to the area (M) 645 of the band area 630 shown in FIG. 5C of the image data in the RAM 106 (described as 650). Then, after the acquired data is stored in the area 662 of the shared buffer 660 via the bus 694, the transfer end is notified via the interrupt signal 692.

  Next, the CPU 102 sets parameters for image processing to be described later via the shared bus 190. Parameters for image processing include the number of processing data, data identifier (number), band area height, image data format, left edge deletion pixel number and right edge deletion pixel number, top edge margin pixel number and bottom edge margin pixel number And the left margin margin pixel number and the right margin margin pixel number. Furthermore, it includes a parallelism indicating N pixels (pixel group: N is an integer of 1 or more) continuous in the length direction of the band region, a scanning mode, a read start address S0_BUF_STR_ADDR of the shared buffer, and the like. Then, the CPU 102 activates the image processing input circuit 220. The image processing input circuit 220 reads and accesses the shared buffer 660 via a control signal 672 such as a chip select signal and an address signal, and acquires read data 674. Then, an operation described later is performed to select pixel data (pixel value) for each pixel, output the pixel data to the internal bus 225 of the image processing unit 150, and then notify the end of the input operation via the interrupt signal 678. .

  The shared buffer may be composed of two or more buffers 666 and 667 as indicated by 660 ′ in FIG. In the example of FIG. 5D described above, the DMAC 194 and the image processing input circuit 220 share one buffer, and thus operate in time division. However, with the configuration of the shared buffer 660 ′ in FIG. 5E, the DMAC 194 transfers data from the RAM 106 to the shared buffer 660 ′ while the image processing input circuit 220 is performing pixel data acquisition processing from the shared buffer 660 ′. The next image data can be transferred. Therefore, the processing of the DMAC 194 and the image processing input circuit 220 can be parallelized.

  In order to perform image processing on the band region of the image data, the same operation may be repeated M times from the region (1) 640 to the region (M) 645 shown in FIG. As described above, image data is acquired from the RAM 106 to the shared buffer. Then, regarding the subsequent two correction data, the number of repetitions set in the DMAC 194 is set to one (that is, the height of the band area is one line), and necessary data is sequentially acquired in the same operation as the image data.

  When N is an integer equal to or greater than 2, N processing target pixels are simultaneously acquired and provided to the N image processing circuits simultaneously.

<Procedure for determining the height of the band area>
Next, how to determine the height of the band region of the first embodiment will be described with reference to FIG. FIG. 6 is a flowchart illustrating a processing procedure example of a program executed by the CPU 102.

  The CPU 102 starts processing (S705), and sets the number of various data necessary for the correction processing as the processing data number S (S710). In this embodiment, the number of processing data is “3”. To the first processing data, 0 ″ is added as a data identifier (number) to identify the data as the read image data to be corrected. Next, for the second processing data, In order to identify the data as correction data (MAX) of the sensor device, “1” is added as a data identifier (number), and the correction data of the sensor device is added to the third processing data. In order to identify the data as (MIN), “2” is added as the data identifier (number), and how many pixels are included in the 32-byte data by the setting of the image data format described in FIG. In this embodiment, the data format of the image data to be processed and the correction data (MAX and MIN) are shown in FIG. It is shown to be the same format as that of the.

  Next, the following parameters are set for processing to be described later by the number of processing data S (S720). For example, the number of left edge deletion pixels and right edge deletion pixels required for the operation of the input image processing circuit, the number of upper edge margin pixels, the number of lower edge margin pixels, the number of left edge margin pixels, the number of right edge margin pixels, the length of the band area It includes a parallelism and a scanning mode indicating N pixels (pixel groups) that are continuous in direction. That is, the variable lp is initialized (S721), and various parameter settings (S723) and variable lp + 1 (S724) are repeated until the variable lp becomes S (S722).

  When the parameter setting is completed (Yes in S722), the height of the band area and the read start address of the shared buffer are calculated (S730). First, the format of image data to be processed is set (S731). In the example of the data format described above, data of 32 bytes unit represents one line of image data. In the case of the correction process of this embodiment, correction data (MAX and MIN) are required to process this one line of image data. However, in this embodiment, it is a line sensor, and one line of correction data is repeated in the sub-scanning direction of the read document. Therefore, all the images in the A column (AA to AN) of the image data in FIG. The correction data is the same for the data. The correction data (MAX) for processing the pixel data (8 × N lines) in the A column region may be 32 bytes for one line × 8 pixels, and the same applies to the correction data (MIN).

  Accordingly, the following calculation will be described on the assumption that the capacity of the shared buffer is {256 bits (= 32 bytes) × BL words} and various data of the BL line can be held.

  A minimum transfer unit of image data to be processed and a minimum transfer unit of correction data (MAX and MIN) for correcting the image data are calculated (S732, S733). In this example, the minimum transfer unit of image data is 32 bytes (corresponding to one word (one line) of the shared buffer), and the minimum transfer unit of correction data (MAX and MIN) for correcting this image data is also 32 bytes. is there. In the case of the present embodiment, the same correction data is repeated for the image data A row (Yes in S740). In this case, (32 bytes × 2) of correction data is required to process the minimum transfer unit 32 bytes of image data (S741). And even if the number of processing lines of image data increases, the number of lines of correction data does not increase. Therefore, a value obtained by subtracting “2” from the number of buffer lines BL is obtained as the transfer amount of the image data (S742), and the minimum transfer unit of the reference data is set to “0” (S750). Then, a value obtained by subtracting “2” from the number of buffer lines BL and dividing by 32 bytes, which is the minimum transfer unit of image data, is the height of the band area (S751).

Finally, the shared buffer start address S0_BUF_STR_ADDR, S1_BUF_STR_ADDR, and S2_BUF_STR_ADDR are set (S760). In the above example,
Image data: S0_BUF_STR_ADDR = 0
Correction data (MAX): S1_BUF_STR_ADDR = BL-2
Correction data (MIN): S2_BUF_STR_ADDR = BL-1
It becomes.

  In the above example, the correction process using the correction data is illustrated. However, other image processes using the reference image data can be realized in the same manner (see S733).

<Configuration Example and Processing Example of Image Processing Input Circuit 220>
(Configuration Example of Image Processing Input Circuit 220)
The internal configuration of the image processing input circuit 220 of this embodiment is shown in FIG. In FIG. 7, the same functional elements and signals as those in FIG. 6 are denoted by the same reference numerals.

  In FIG. 7, an operation parameter 810 is set from the CPU 102, and this value is stored in an internal register, and the sequencer 820 operates. The operation will be described later with reference to FIGS. 8A, 8B, and 9. The sequencer 820 calculates the address 672 of the shared buffer 660, and acquires 32-byte data 674 from the shared buffer 660. Then, the pixel data generation circuit 830 generates data for one pixel or data for N pixels continuous in the length direction of the band region, and outputs the generated data to the internal bus 225 of the image processing unit 150.

(Processing example of image processing input circuit 220)
In order to perform image processing on the band region of the image data, data processing of 32 bytes × number of lines (hereinafter referred to as a transfer block) from region (1) 640 to region (M) 645 in FIG. 8A may be repeated M times. . In addition, since the data processing is repeated M times for at least one of the image data and the plurality of image data to be referred to and the correction data, the correction processing of this embodiment is repeated 3M times in total. Further, in order to perform trimming processing (processing for discarding unnecessary end pixels) in the length direction of the band area for each data, the number of left end deleted pixels and the number of right end deleted pixels are set for each data.

  The number of deleted pixels cannot be set beyond the pixel area of 32 bytes in the image data format. That is, since the data format of the present embodiment is 8 pixels / 32 bytes, the left end deletion pixel number and the right end deletion pixel number are values up to 7. When performing further trimming processing, it is not necessary to transfer the block to be transferred at the time of reading from the RAM 106. The setting of the left end deleted pixel number and the right end deleted pixel number is meaningful only in the first transfer block 640 and the last transfer block 645. Therefore, the setting of the left end deletion pixel number affects the first transfer block 640, the hatched area 856 (pixels 3 to 7) becomes effective, and the pixels 0 to 2 are deleted. Similarly, the setting of the right end deletion pixel number affects the last transfer block 645, the hatched area 858 (pixels 0 to 1) becomes valid, and the pixels 2 to 7 are deleted. Then, an image data area 860 after the trimming process is generated.

  Next, the pixel copy expansion of the edge pixels of the trimmed image 860 by the designated number of pixels (the number of upper edge margin pixels, the number of lower edge margin pixels, the number of left edge margin pixels, and the number of right edge margin pixels) is performed. To generate a margin area so that it becomes an image data area 870. The image processing circuit includes local image processing such as a spatial filter. In order to perform local image processing at the upper end, the lower end, the left end, and the right end of the image data, it is desirable to have a function for generating the margin area.

  In this embodiment, since the scanner correction is performed, a desired output image can be obtained even if there is no function for generating a margin area if the reading area of the scanner is set wide in advance. However, in the case of Example 2 described later, since the input is a standardized video signal such as a broadcast wave, the function of generating the margin area is necessary to obtain a desired output image. The sequencer 820 accesses the shared buffer 660 in consideration of the margin area by setting the upper margin margin pixel number, the lower margin margin pixel number, the left margin margin pixel number, and the right margin margin pixel number.

  Specifically, in the case of the transfer block (1) 640, the shared buffer is used while the upper margin margin pixel or the lower margin margin pixel is generated in consideration of the upper margin margin pixel number, the lower margin margin pixel number, and the left margin margin pixel number. Do not increment the address. Further, it is considered that there is a pixel by extending the left end of the read 32-byte pixel data like 872. As for transfer flocks (2) to (M-1), the address of the shared buffer is not incremented while the upper edge margin pixel or the lower margin margin pixel is generated as described above. There is no need to expand the 32-byte pixel area at both the left and right ends. In the last transfer block (M) 645, the address of the shared buffer is not incremented while the upper edge margin pixel or the lower edge margin pixel is generated as described above. Further, it is considered that there is a pixel by expanding the right end of the read 32-byte pixel data like 874.

  FIG. 8B illustrates how the image data 870 created by the above-described trimming processing and edge extension processing by the function of generating the margin area is output via the internal bus 225. In FIG. 8B, transfer block (1) 640 is taken as an example.

  The output line number L is the total of the upper edge margin pixel, the lower edge margin pixel, and the number of lines indicating the band height. The method of outputting pixel data varies depending on the degree of parallelism and the scanning mode indicating N pixels (pixel groups) continuous in the length direction of the band region set by the CPU 102. Pixels are output continuously in the band area length direction only by the setting of the degree of parallelism. Depending on the setting of the scanning mode, it is determined whether N pixels are output at 1 pixel / cycle or simultaneously (parallel) at N pixels / cycle. In the case of simultaneous (parallel) output at N pixels / cycle, the internal bus 225 of the image processing 150 needs to have a band that can be output at the same time, and the subsequent image processing circuit also needs to operate in N pixels in parallel. . In addition, there is a limitation that the number of pixels in the 32-byte unit of the image data format, the number of left end deleted pixels, the number of left end glue margin pixels, and the number PN of pixel groups determined by the above-described parallelism must be integers.

<Example of Procedure for Outputting Pixel Data to Internal Bus 225 of Image Processing Input Circuit>
The flowchart of FIG. 9 summarizes a procedure example of outputting pixel data to the internal bus 225 of the image processing input circuit 220. As can be seen from the loop (S961, S962, S970, S963) of step S960 in FIG. 9, pixel groups of image data or various reference data (a plurality of correction data and image data) are output in order. To the image processing circuit. The image processing circuits 230, 240,..., 280 in FIG. 4 sequentially receive N-pixel image data, correction data MAX, and correction data MIN, perform correction processing when the data is complete, and perform N-pixel correction processing. Output the data. The distinction between the various data is determined by the above-described data identifier (number). These processes are repeated for each line in the height direction of the band region as indicated by the loop (S951, S952, S960, S953) of step S950. However, in the case of correction data, the band height is one line, and the shared buffer address of the same line is always read repeatedly. After the pixel area of the number of output lines (L × N) is output, scanning is performed for a pixel group (N pixels) in the length direction of the band area as indicated by the loop (S941, S942, S950, S943) in step S940. Then, the process proceeds to the processing for pixels. When these processes are repeated and the process shifts by the number PN of the pixel groups and all the number of pixels (PN × N × L) is processed, one transfer block (1) is completed, and the next transfer block ( 2) to (M) will be repeated.

[Example 2]
Configurations and processes having the same functions as those of the first embodiment are denoted by the same reference numerals, and descriptions of those that are not structurally and functionally omitted are omitted.

  Next, a second embodiment in which the present invention is applied to the composition processing 1020 and the I / P conversion 1030 shown in FIG. 13 will be considered.

  The difference from the first embodiment is that various data are input via the video input unit 130 of FIG. 1 and the processed data is output to a display device such as a display by the video display unit 160. 1 are the same as those in the first embodiment. 4 is the same as that of the first embodiment.

  In the second embodiment, unlike the first embodiment, a plurality of image data and composite data necessary for image processing of image data are two-dimensional data. Further, unlike the correction data (MAX and MIN) in the first embodiment, the same data value is not repeated in the sub-scanning direction. Therefore, the storage method is the same as that of the coordinate system 440 like S0_IMG which is a processing target like S1_IMG, S2_IMG and S3_IMG in FIG. For the sake of brevity, it is assumed in the second embodiment that the various image data and composite data are in the data format shown in FIG. Various data acquisition operations from the RAM 106 to the shared buffer 660 are the same as those in the first embodiment.

<How to determine the height of the band area>
The method of determining the height of the band region in the second embodiment is different from that in the first embodiment. The difference will be described below.

  The CPU 102 sets the number of various data necessary for image processing as the processing data number S. When combining two rendering images, the necessary data is three types of data: one image data to be processed, image data to be combined and two combined data, and S = 3. In addition, in the case of I / P conversion for three field images, image data for each past and future is required for the field image at the time to be processed, resulting in a total of three types of data, and S = 3 is there. In addition, in the case of I / P conversion for five field images, image data for each of the past and future fields is required for the field image of the processing target time, resulting in a total of five types of data, and S = 5 is there. It should be noted that the number of processing data S is the number of data identifiers (numbers), the number of left edge deleted pixels and the number of right edge deleted pixels, the number of upper edge margin pixels, the number of lower edge margin pixels, the number of left edge margin pixels, the number of right margin margin pixels, and the band The parallelism indicating the N pixels (pixel group) continuous in the length direction of the region and the scanning mode are set. However, since it is the same as that of Example 1, description is omitted.

  Next, the height of the band area and the read start address of the shared buffer are calculated. According to the above-described format, data of 32 bytes represents one line of image data. In the case of this embodiment, other image data and composite data are also 32 bytes (one line) in order to process this one line of image data. Therefore, the following calculation will be described on the assumption that the capacity of the shared buffer 660 is {256 bits (= 32 bytes) × BL} words and various data of the BL line can be held.

The minimum transfer unit of image data is 32 bytes (corresponding to one word (one line) of the shared buffer), and the minimum transfer unit of various data for processing this image data is also 32 bytes. In other words, in order to process the minimum transfer unit 32 bytes of image data, various data of (32 bytes × 2) are required when combining two rendering images. The same applies to the case of I / P conversion for three field images. When performing I / P conversion on five field images, various data of (32 bytes × 4) are required. Therefore, the height (number of lines) of the band region is as follows.
・ Composition processing of two rendering images BL line / (1 line + 2 line) = BL / 3
I / P conversion for 3 field images BL line / (1 line + 2 line) = BL / 3
-I / P conversion for 5 field images BL line / (1 line + 4 line) = BL / 5
A value obtained by dividing the number of buffer lines BL by the total value of the minimum transfer unit 32 bytes of image data and the minimum transfer unit of data necessary for processing the image data is the band area height.

  The head address of the shared buffer 660 is an address arranged at equal intervals for each band area height (number of lines). Refer to the calculation flow when the presence / absence of repetition of the reference data (S740) in FIG. 6 is “None”.

  According to the setting of the band height, the method of outputting the pixel data to the image processing circuit after writing the desired image data and various reference data in the shared buffer is the same as in the first embodiment.

[Other embodiments]
As shown in FIG. 12, the length direction of the band area is set in accordance with the sub-scanning direction of the image data, as in Patent Document 1. Further, even if the pixel area 520 included in the 32-byte unit of the image data format of FIG. 3 is set to 1 pixel in the main scanning direction and 8 pixels in the sub-scanning direction, the above-described first and second embodiments are established. Needless to say.

  Further, as shown in FIG. 10, a case is considered in which the pixel region 1282 included in the unit of 32 bytes is set to 4 pixels in the main scanning direction and 2 pixels in the sub-scanning direction. In this case, image data transfer to the shared buffer 660 by the CPU 102 and the DMAC 194 is wastefully transferred because the minimum transfer unit is two lines. However, if the sequencer 820 of the image processing input circuit 220 in FIG. 7 can perform access processing to the shared buffer 660 in units of one line, it goes without saying that the first to third embodiments are established.

  Further, the processing of each of the above-described embodiments may be realized by cooperation of a plurality of hardware and software. In this case, it can be realized by executing software (program) acquired via a network or various storage media by a processing device (CPU, processor) such as a computer.

  The present invention may also be realized by supplying a storage medium storing a program for causing a computer to realize the functions of the above-described embodiments to a system or apparatus.

Claims (21)

An image processing apparatus that divides input image data into a plurality of band areas, supplies partial image data in the divided band areas to an image processing means via a buffer, and performs image processing with reference to reference data An arithmetic method for
Obtaining a transfer unit of the partial image data;
The first data amount data in the partial image data and the second data amount data in the reference data referred to for processing the first data amount data fit within the capacity of the buffer. A second acquisition step of acquiring the first data amount as the transfer amount of the partial image data based on the capacity of the buffer and the transfer unit ;
And a third acquisition step of acquiring a height of the band region based on the transfer amount and the transfer unit. In the second acquisition step, if the reference data is repeated when the partial image data is processed with reference to the reference data, the second data in the reference data considering the repetition. The calculation method according to claim 1, wherein the transfer amount is determined using an amount. The calculation method according to claim 1, further comprising a step of storing the partial image data and the reference data in the buffer according to the transfer amount and the second data amount . Each of N pixels (where N is an integer of 1 or more) continuous in the length direction of the band area from the partial image data corresponding to the height of the band area stored in the buffer and the reference data Reading out each pixel group from each stored data;
A providing step of adding an identifier for distinguishing the partial image data and the reference data and providing each pixel group to the image processing means in an order necessary for image processing. The calculation method according to claim 3.   The calculation method according to claim 4, wherein in the reading step, scanning is performed pixel by pixel in the height direction of the band region.   5. The read-out process according to claim 4, wherein in the reading step, trimming processing is performed on the image data and the reference data in a length direction of a band area in accordance with designation of a left end deletion pixel number and a right end deletion pixel number. Calculation method.   In the reading step, for the image data after the trimming process in the length direction of the band area and the reference data, the number of upper margin margin pixels, the number of lower margin margin pixels, the number of left margin margin pixels, the right margin margin pixel The calculation method according to claim 6, wherein the end portion is expanded according to the designation of the number.   5. The calculation method according to claim 4, wherein, in the providing step when N is an integer equal to or greater than 2, one pixel at a time is sequentially provided to the image processing unit from N pixels to be processed.   5. The calculation according to claim 4, wherein, in the providing step when N is an integer of 2 or more, N pixels to be processed are simultaneously acquired and provided to the N image processing means simultaneously. Method. 5. The image processing unit according to claim 4, wherein the image processing unit receives and distinguishes whether it is partial image data to be processed or reference data according to the identifier added to a pixel, and performs image processing. Calculation method. An image processing apparatus,
When the input image data is divided into a plurality of band areas, the data of the first data amount in the partial image data in the divided band area and the reference that is referred to for processing the data of the first data amount The first data amount is calculated based on the capacity of the buffer and the transfer unit of the partial image data so that the second data amount of data in the image data can be accommodated in the buffer capacity of the image processing apparatus . acquired by the transfer amount of the image data, holding the height of the band area obtained by the arithmetic unit, for obtaining the height of the band area on the basis of the transfer unit of the partial image data and the transfer amount Holding means;
A dividing means for dividing the input image data into a plurality of band regions;
An image processing apparatus comprising: image processing means for acquiring partial image data in the divided band region via a buffer and performing image processing with reference to reference data. For the partial image data corresponding to the height of the band area stored in the buffer and the reference data, the image processing means N pixels that are continuous in the length direction of the band area (where N is 1 or more) (Integer) pixel groups, an identifier for distinguishing the partial image data from the reference data is added, and the pixel groups are input to the image processing means in the order required for image processing. The image processing apparatus according to claim 11, further comprising an input unit configured to perform input.   The image processing apparatus according to claim 12, wherein the input unit scans one pixel at a time in the height direction of the band region.   The input means performs a trimming process in a length direction of a band region on the image data and the reference data according to designation of a left end deletion pixel number and a right end deletion pixel number. Image processing apparatus.   In the input means, with respect to the image data after the trimming process in the length direction of the band area and the reference data, the number of upper margin margin pixels, the number of lower margin margin pixels, the number of left margin margin pixels, and the right margin margin pixel The image processing apparatus according to claim 14, wherein the end portion is expanded in accordance with designation of the number.   The image processing apparatus according to claim 12, wherein when the N is an integer equal to or greater than 2, the input unit sequentially inputs pixels one by one from N pixels to be processed to the image processing unit. .   13. The input unit according to claim 12, wherein when the N is an integer of 2 or more, the N processing target pixels are simultaneously acquired and input to the N image processing units at the same time. Image processing device. The image processing means determines whether it is partial image data to be processed or reference data based on an identifier added to a pixel, and performs the image processing. The image processing apparatus described. An image processing apparatus that divides input image data into a plurality of band areas, supplies partial image data in the divided band areas to an image processing means via a buffer, and performs image processing with reference to reference data An arithmetic unit for obtaining the height of the band region set for
Obtaining means for obtaining a transfer unit of the partial image data;
The first data amount data in the partial image data and the second data amount data in the reference data referred to for processing the first data amount data fit within the capacity of the buffer. Second acquisition means for acquiring the first data amount as the transfer amount of the partial image data based on the capacity of the buffer and the transfer unit ;
An arithmetic device comprising: third acquisition means for acquiring the height of the band region based on the transfer amount and the transfer unit. When the input image data is divided into a plurality of band areas, the data of the first data amount in the partial image data in the divided band area and the reference that is referred to for processing the data of the first data amount The first data amount is calculated based on the buffer capacity and the transfer unit of the partial image data so that the data of the second data amount in the data for use can be accommodated in the buffer capacity of the image processing apparatus . acquired by the transfer amount of data, holding the height of the band area obtained by the arithmetic unit, for obtaining the height of the band area on the basis of the transfer unit of the partial image data and the transfer amount holding The image processing apparatus having means,
A dividing means for dividing input image data into a plurality of band regions;
A computer program for obtaining partial image data in the divided band area via a buffer and functioning as image processing means for performing image processing with reference to reference data. An image processing apparatus that divides input image data into a plurality of band areas, supplies partial image data in the divided band areas to an image processing means via a buffer, and performs image processing with reference to reference data An arithmetic unit that acquires the height of the band region set for
Obtaining means for obtaining a transfer unit of the partial image data;
The first data amount data in the partial image data and the second data amount data in the reference data referred to for processing the first data amount data fit within the capacity of the buffer. Second acquisition means for acquiring the first data amount as the transfer amount of the partial image data based on the capacity of the buffer and the transfer unit ;
A computer program for functioning as third acquisition means for acquiring the height of the band area based on the transfer amount and the transfer unit.

Images