JP2006244094A - Data writing device and method, data reading device and method, program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data writing device and method, a data reading device and method, a program and a recording medium by which image data can be read efficiently. <P>SOLUTION: An address control section 5 writes image data in an image storage section 4 so that arbitrary N×N pixel data consisting of N pixels in a horizontal direction and N pixels in vertical direction may continue within the same row address. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a data writing apparatus and method for efficiently reading image data, a data reading apparatus and method, a program, and a recording medium.

  Conventionally, an image conversion apparatus 101 that converts an input image according to a conversion rule and obtains an output image as shown in FIG. 9 is known. As shown in FIG. 10, the image conversion apparatus 101 includes an image input unit 102 that inputs image data, an image buffer unit 103 that temporarily stores image data, and a conversion table unit 104 that stores an image data conversion table. An image storage unit 105 that stores image data, a pixel value interpolation calculation unit 106 that interpolates pixel values based on an image data conversion table, and a converted image output unit 107 that outputs converted image data. . Such an image conversion apparatus 101 temporarily stores an input image in the image storage unit 103 and uses the image data conversion table to determine which coordinates (X, Y) of the output image correspond to the input image ( For example, (X ′, Y ′)) is referred to.

  Further, the image conversion apparatus 101 can convert not only the coordinates of integer values but also the coordinates of decimal values. When converting to a decimal value coordinate, peripheral pixels are read from the reference input image from the memory, and linear interpolation is performed from the pixel values. For example, as shown in FIG. 11, the position P on the input image specified by the image data conversion table is obtained by reading out four pixels around the value indicated by the image data conversion table and performing linear interpolation from the pixel values. It is done. For example, P (155.25, 171.75) is the peripheral four pixels P1 (155,171), P2 (156,171), P3 (155,172) and P4 (156,172) as shown in FIG. Is read out and linear interpolation is performed. By performing pixel value interpolation in this way, an output image with high accuracy can be obtained.

JP 2003-237145 A JP 2004-72600 A

  Incidentally, SDRAM (Synchronous Dynamic Random Access Memory) is widely used as the type of memory for storing input images because of its large capacity. This SDRAM can read out pixels continuously every clock while incrementing the column address by activating the bank once within the same row address (hereinafter referred to as burst read).

  For example, when printing in the column direction (paper transport direction) with a printer, HV (Horizontal / Vertical) conversion is performed to sequentially cut out image data vertically, the image data is stored in a memory, and the vertical direction of the image is burst read ( For example, see Patent Document 1.)

  In the image rotation system, image data is stored by switching banks for each line, and high-speed image data is read (for example, see Patent Document 2).

  However, although the techniques described in Patent Documents 1 and 2 can read image data in the vertical direction or the horizontal direction at high speed, when the peripheral pixels are read like the image conversion apparatus 101 described above, the efficiency is high. Image data cannot often be read out.

  The present invention has been proposed in view of such a conventional situation, and provides a data writing apparatus and method, a data reading apparatus and method, a program, and a recording medium that can read image data efficiently. With the goal.

  In order to achieve the above-described object, a data writing device according to the present invention is a data writing device for writing image data into a data memory, a buffer for temporarily storing at least N lines of the image data, and an N line Writing means for writing the image data into the data memory every time, and the writing means stores data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction within the same row address. The image data is written in the data memory so as to be continuous.

  The data writing method according to the present invention includes a step of temporarily storing at least N lines of the image data in a buffer in the data writing method for writing image data into a data memory, and the image data for each N lines. In the data memory, and in the writing step, the data of any N × N pixels composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. The image data is written in the data memory.

  The program according to the present invention is a program for executing a process of writing image data into a data memory, a step of temporarily storing at least N lines of the image data in a buffer, and the image data for each N lines. In the data memory, and in the writing step, the data of any N × N pixels composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. The image data is written in the data memory.

Further, the data reading apparatus according to the present invention is a data read-in device in which image data is stored so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. A data reading device for reading out pixel data at a desired pixel position from a memory, wherein the acquisition unit reads out data of N ² pixels continuous according to the pixel position and acquires data of N × N pixels, and And generating means for generating pixel data at the pixel position based on N × N pixel data.

Further, the data reading method according to the present invention is a data in which image data is stored such that data of arbitrary N × N pixels composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous within the same row address. A data reading method for reading out pixel data of a desired pixel position from a memory, burst reading data for N ² pixels continuous according to the pixel position, and acquiring N × N pixel data; And a generation step of generating pixel data at the pixel position based on N × N pixel data.

Further, the program according to the present invention is provided from a data memory in which image data is stored so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. A program for performing a process of reading out pixel data at a desired pixel position, wherein an N × N pixel data is obtained by burst-reading data for N ² pixels continuous according to the pixel position, and the above N And a generation step of generating pixel data at the pixel position based on the data of × N pixels.

  In the recording medium according to the present invention, in the recording medium on which image data is recorded, arbitrary N × N pixel data composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. As described above, the image data is arranged.

  According to the present invention, N × N pixels are written in the data memory so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. The data of N pixels can be burst read efficiently.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing an image conversion apparatus according to the present embodiment. The image conversion apparatus 1 controls an image input unit 2 for inputting image data, an image buffer unit 3 for temporarily storing image data, an image storage unit 4 for storing image data, and writing / reading of image data. An address control unit 5 that performs conversion, a conversion table storage unit 6 that stores a conversion table for converting an image, a conversion table decoding unit 7 that decodes the conversion table, and a pixel value interpolation calculation unit 8 that performs pixel value interpolation. It is configured.

  The input image input from the image input unit 2 is temporarily stored in the image buffer unit 3. The FIFO 31 and the FIFO 32 of the image buffer unit 3 each store one line of the input image. The FIFO 31 stores a line next to the line accumulated in the FIFO 32. That is, the FIFO 31 and the FIFO 32 store two consecutive upper and lower lines of the input image data. As a result, the image can be stored in combination with the next lower line as will be described later.

  The selector 33 switches from continuous two lines of image data stored in the FIFO 31 and the FIFO 32 to one line of image data. The D-Flipflops 34 to 36 determine pixels to be overlapped by switching banks from one line of image data, and divide one line of image data into each bank. The image data divided into each bank is output as write data via the selector 37.

  The image storage unit 4 includes a memory such as an SDRAM (Synchronous Dynamic Random Access Memory), and stores image data. This SDRAM can read out pixels continuously every clock while incrementing the column address by activating the bank once within the same row address (hereinafter referred to as burst read). The SDRAM can execute a command in another bank while executing a command in a certain bank.

  The address control unit 5 controls the selector 33 and the selector 37 to generate write data and write image data to the image storage unit 4 and read image data from the image storage unit 4.

  The conversion table storage unit 6 has a conversion table for converting pixel values in order to convert an image. This conversion table includes a table that converts to a decimal pixel value.

  The conversion table decoding unit 7 decodes the integer part and the decimal part of the conversion table. The pixel value interpolation calculation unit 8 linearly interpolates the pixel value of the decimal part with the peripheral pixels of the integer value. That is, the peripheral pixels of the integer value are read from the image storage unit 4 from the integer part decoded by the conversion table decoding unit 7, and the peripheral pixels of the integer value are input to the pixel value interpolation calculation unit 8 as read data, and linear interpolation is performed. Do.

  By converting the image in this way, a highly accurate output image can be obtained.

  Next, a method for storing image data in the image storage unit 4 will be described. FIG. 2 is a diagram illustrating a case where an input image of 640 × 480 pixels is stored in a 4-bank memory. In this example, peripheral 4 pixels, that is, 2 × 2 pixels are continuously read out.

  As shown in FIG. 2, the input image is stored in the order of banks 0, 1, 2, 3 if the Y coordinate is an even number, and in the order of banks 2, 3, 0, 1 if the Y coordinate is an odd number. That is, the input image data is divided into a plurality of blocks, and each block is stored in a bank different from the surrounding blocks. Thereby, for example, when correcting distortion in the camera, it is possible to perform high-speed reading from different banks.

  Further, by the continuous two lines of image data stored in the FIFO 31 and the FIFO 32, each bank has (X, Y), (X, Y + 1), (X + 1, Y), as shown in FIGS. Images are stored while being combined with the next lower line in the order of (X + 1, Y + 1),. As described above, the peripheral four pixels are stored continuously in the same row address, so that the peripheral pixels can be burst-read with one activation of the bank.

  When one horizontal line of the image cannot be stored in one bank, the last pixel stored in the bank is also stored in the head of the next bank by D-Flipflop 34-36. For example, in the example of FIGS. 3 and 4, the pixels at the column addresses COL319 and COL320 of the bank 0 are stored in the column addresses COL0 and COL1 of the bank 1. Thereby, any four peripheral pixels can be read out from one bank.

  Further, it is preferable that the end of the input image stores the same pixel value or stores 0 as in the row address 479 of the bank 0, for example.

  In this way, the input image is stored in the order of (X, Y), (X, Y + 1), (X + 1, Y), (X + 1, Y + 1),. By burst access to the integer value pixels with the decimal point of the value truncated, any of the surrounding four pixels can be burst read.

  Next, burst read of four peripheral pixels will be described with reference to FIG. Reading data from the SDRAM is performed by inputting a row address simultaneously with the ACTIVE command to activate the bank, and inputting a column address simultaneously with the READ / WRITE command, and the data is output after the CAS latency time.

  Conventionally, as in the timing of FIG. 7A, an ACTIVATE command must be issued every time a pixel is read, and the bank once activated must be precharged. Therefore, since a series of precharge operations are required to read out four peripheral pixels, a read time as shown in FIG. 7A is required.

  However, when the peripheral four pixels are stored continuously in the same row address as in this embodiment, the peripheral is continuously activated by activating the bank once as shown in the timing of FIG. 7B. Four pixels can be burst read. According to this timing, after burst read, the peripheral four pixels can be read out only in one clock data non-output period for precharging the read bank, so that the read time is longer than the timing of FIG. It can be shortened to about one third. In the continuous readout of the surrounding 4 pixels in this 1-clock data non-output period, when the surrounding 4 pixels to be continuously read are different from the bank in which the preceding surrounding 4 pixels are stored, or before the surrounding 4 pixels are stored Even if the banks are the same, this can be done when they are stored at the same row address as the previous four surrounding pixels.

In the above-described embodiment, the peripheral four pixels are stored so as to be burst-read. However, the present invention is not limited to this. For example, the peripheral nine pixels may be burst-read. In this case, the image buffer unit is configured as shown in FIG. That is, the image buffer unit 300 includes FIFOs 301 to 303 for temporarily storing three consecutive lines, a selector 304 for switching from image data stored in the FIFOs 301 to 302 to one line of image data, and switching of image data banks. D-Flipflops 305 to 310 that determine pixels to be overlapped and divide one line of image data into each bank, and a selector 311 that outputs the image data divided into each bank as write data. That is, in order to store N × N pixels so that they can be read in bursts, N FIFOs and (N−1) × (number of banks−1) D-Flipflops may be provided. Thereby, the pixel data is in the XY coordinate system at the row address of the memory,
(X, Y), ..., (X, Y + (N-1)),
(X + 1, Y), ..., (X + 1, Y + (N-1)),
...
(X + (N-1), Y), ..., (X + (N-1), Y + (N-1))
Arranged in this order.

  As is apparent from the above description, according to the present invention, N × N pixels can be burst read by writing image data continuously to the row address for each N × N pixels.

It is a block diagram which shows the structure of the image conversion apparatus in this Embodiment. It is a figure explaining the case where the input image of 640x480 pixels is stored in memory of 4 banks. It is a figure explaining the bank 0 in the case of storing the input image of 640x480 pixels in the memory of 4 banks. It is a figure explaining the bank 1 in the case of storing the input image of 640x480 pixels in the memory of 4 banks. It is a figure explaining the bank 2 in the case of storing the input image of 640x480 pixels in the memory of 4 banks. It is a figure explaining the bank 3 in the case of storing the input image of 640x480 pixels in the memory of 4 banks. It is a figure explaining the timing of reading of a peripheral pixel. It is a block diagram which shows the structure of the image buffer part in other embodiment. It is a figure for demonstrating the conventional image converter. It is a figure for demonstrating the conventional image converter. It is a figure for demonstrating pixel value interpolation. It is a figure for demonstrating pixel value interpolation.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image converter, 2 Image input part, 3 Image buffer part, 4 Image storage part, 5 Address control part, 6 Conversion table storage part, 7 Conversion table decoding part, 8 Pixel value interpolation calculation part

Claims (8)

In a data writing device for writing image data to a data memory,
A buffer for temporarily storing at least N lines of the image data;
Writing means for writing the image data into the data memory every N lines,
The writing means writes the image data into the data memory so that arbitrary N × N pixel data composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. Data writing device. The data memory is composed of a plurality of banks,
2. The data according to claim 1, wherein the writing means divides the image data into a plurality of blocks and writes the data of each block so that the bank is different from the bank in which the data of the adjacent block is written. Writing device. In a data writing method for writing image data to a data memory,
Temporarily storing at least N lines of the image data in a buffer;
A writing step of writing the image data into the data memory every N lines,
In the writing step, the image data is written into the data memory so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. Data writing method. In a program that executes a process of writing image data to a data memory,
Temporarily storing at least N lines of the image data in a buffer;
A writing step of writing the image data into the data memory every N lines,
In the writing step, the image data is written into the data memory so that data of an arbitrary N × N pixel composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. Program to do. Pixel data at a desired pixel position is read out from a data memory in which image data is stored so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. A data reading device,
Acquisition means for burst-reading N ² pixels of data according to pixel positions and acquiring data of N × N pixels;
A data reading device comprising: generating means for generating pixel data at the pixel position based on the data of the N × N pixels. Pixel data at a desired pixel position is read out from a data memory in which image data is stored so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. A data reading method comprising:
An acquisition step of burst-reading data for N ² pixels continuous according to the pixel position and acquiring data of N × N pixels;
And a generation step of generating pixel data at the pixel position based on the N × N pixel data. Pixel data at a desired pixel position is read out from a data memory in which image data is stored so that data of arbitrary N × N pixels including N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address. A program for processing,
An acquisition step of burst-reading data for N ² pixels continuous according to the pixel position and acquiring data of N × N pixels;
And a generation step of generating pixel data at the pixel position based on the data of the N × N pixels. In a recording medium on which image data is recorded,
A recording medium, wherein image data is arranged so that data of arbitrary N × N pixels composed of N pixels in the horizontal direction and N pixels in the vertical direction are continuous in the same row address.

Images