JP2751143B2 - Image processing system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing system for handling natural images, animation images and the like.

[0002]

2. Description of the Related Art Moving an image in the horizontal direction of the screen is called horizontal scrolling, and moving the image in the vertical direction is called vertical scrolling. In an image processing apparatus such as a computer game machine, horizontal scrolling usually corresponds to an H blank period, and vertical scrolling corresponds to a V blank period.

FIG. 1 shows an H blank and a V blank, and no image is displayed during this period. Since the scanning line returns, the V blank is referred to as a horizontal retrace period and the V blank is referred to as a vertical retrace period. The processing of the display data is performed during this period, and the image can be displayed on the video screen by transferring the image data in accordance with the display period.

[0004] Image data is represented by a color in dot (pixel) units. Color data can be represented by three primary colors (RBG) or by luminance (Y) and color difference (UV). In any case, these colors can be expressed as (c ₀ , c ₁ , c ₂ ,..., C _n-1 ) in horizontal direction dot units. This series of color data is called a color vector, and the color data c0, c1, c2,... One vector corresponds to one raster display. FIG. 2 shows the relationship between each vector factor and the dot.

In this case, the vector factor and the dot are one-to-one.
Therefore, if image data is read from the virtual screen (VRAM) before 1H blank, and data is transferred in the next horizontal display period, an image for one raster can be displayed.

[0006] The display data is stored in a virtual screen (VRAM). The virtual screen is larger than the real screen (video screen), and a part thereof is displayed on the real screen. Which part of the virtual screen is displayed on the real screen depends on BXR and B shown in FIG.
Determined by YR. BXR is a value in dot unit,
BYR is a value in raster units. The register that specifies BXR is the BGX register, and the register that specifies BYR is the BGY register.

For example, in order to scroll the screen to the right by X dots in the horizontal direction, the current horizontal coordinate value is BG.
If X is set, the value of BGR-X may be set in the BXR register.

That is, as shown in FIG. 4, if data is transferred by using a coordinate value in the left direction of X dots from the currently displayed position as a display start point, the image looks as if it has been scrolled to the right by X dots. Since the BXR register becomes effective from the next raster, reading the image data from the virtual screen is one raster before the data transfer.

[0009]

As described above, since one vector factor corresponds to one dot, it is usually sufficient to send out without processing. However, when one vector factor does not correspond to one dot, it cannot be achieved only by the BXR register. This is because BGX specified in the BXR register is a dot unit.

[0010] This will be described with an example of an image processing system that also handles natural images. In a conventional computer game apparatus, an animation image is mainly used, and display colors of image data are only a few colors such as four colors, 16 colors, and 256 colors. For this purpose, as shown in FIG. 5, it is only necessary that the color data can point to the address of the color palette in units of one dot.

However, when dealing with natural images, tens of thousands or millions of display colors are required. It is not a practical measure to cope with this with the color palette method as in the past, since the color palette becomes enormous and the memory is pressed.

Therefore, it is conceivable that color data corresponding to dots can be directly displayed with color information. However, when color information is directly provided, the color data is large, which also leads to pressure on the memory.

[0013] For this reason, by taking advantage of the fact that adjacent images have similar colors in a natural image, a method of associating one color data (color vector factor in FIG. 2) with a plurality of dots or the like is used. The point is addressed.

FIG. 6 is a system block diagram of an example of an image processing system capable of handling a natural image. The controller unit, the image data decompression unit, the VDP unit, and the video encoder unit of this device are each configured on an IC chip.

KRAM, RRAM and VRAM are a controller unit, an image data decompression unit and a VD
This is an external memory of the P unit. The controller unit reads the data from the CD-ROM and once
The compressed data is transferred to the image data decompression unit.

The image data decompression unit decodes the compressed data sent from the controller unit and sends the data to the video encoder unit. The video encoder unit displays image data directly from the controller unit or from the image data decompression unit and the VDP unit on a video screen.

The function of the image data expansion unit will be described. FIG. 7 is a block diagram of the image data decompression unit. Data bus buffer (DATA BUS BUF
FER) is a memory for receiving image data sent from a device such as a control unit, and distributes and transfers the data to predetermined blocks.

The external memory RRAM (R-RAM) is an area for storing decoded data, and has a memory size of 16 rasters (64 Kbits) × 2. The reason for having two areas is to alternately read and write data in order to increase the processing speed.

The image data decompression unit can handle IDCT images and run-lens images. The IDCT image is a natural image / moving image obtained by IDCT decoding, and the run-lens image is a run-length compressed animation moving image.

In both cases, the image to be compressed is 256 dots × 240 rasters per field. The display colors of the IDCT are about 16.77 million colors, and the run-length mode of the run-lens image is a pallet system, and there are four types of 16 colors / 32 colors / 64 colors / 128 colors.

Terminals KR0 to KR7 are data buses for transferring data from the control unit to the image data decompression unit. On the other hand, -REGR is a data request signal terminal passed from the image data decompression unit to the control unit. The data request signal requests compressed image data from the control unit. That is, -REQR = 0 ... data request -REQR = 1 ... data stop.

The image data decompression unit restores the compressed image data of 16 rasters in 16 raster periods. Therefore, according to the transfer start timing of the present invention, transfer of image data for 16 rasters to be displayed is started from 16 rasters before to the image data decompression unit, and is completed by 1 raster before display start.

Since the image data decompression unit does not have the raster information, the timing is taken only by the control unit. In the image data decompression unit, 16 bytes after the control unit starts data transfer.
The image display of the image data decompression unit image is performed in synchronization with the first HSYNC.

FIG. 8 conceptually shows the flow and timing of image data. In the figure, while the third raster is being written from the video screen, the control unit has already sent the data of the fourth 16 rasters to the image data decompression unit. This transfer process is completed before the third 16 rasters are displayed on the video screen. If rasters are sequentially processed in this manner, data for one screen can be displayed on the video screen. This process is called normal reproduction.

The image data decompression unit has an image data FIFO from the control unit.
When the FIFO becomes full, disable (-RE
The request signal QR = 1) is passed to the control unit. When this signal is received by the control unit, the data transfer is suspended.

The control unit includes an image data decompression unit transfer control register, an image data decompression unit data start address register, and a multiple performance block number register. The image data decompression unit transfer control register permits the transfer of the image data decompression unit data. If not, the transfer is not performed.
If the transfer is not permitted during the transfer, the transfer is stopped at that point.

The start address register specifies the start address of the KRAM in which the data for the image data decompression unit is stored.
Transfer is started from the RAM address. During block transfer, the address is automatically incremented.

The transfer start raster register specifies the timing at which the control unit starts data transfer to the image data decompression unit, and specifies a raster number. The transfer block number register specifies the number of transfer blocks when data corresponding to 16 rasters to be transferred to the image data expansion unit is one block.

If the contents of each register are not changed, the same image is displayed in the next same frame. Each register becomes effective immediately after being set. However, when a block is being transferred to the image data decompression unit, it becomes effective after the transfer. Each register is shown in FIG.

The data is actually transferred because arbitration of the K-bus (K-BUS) is performed so that the image data expansion unit data can be accessed, and the request signal (-REQR) is transmitted from the image data expansion unit. = 0) is output.

The data transfer is disabled because:
The following are the cases. (1) When the processing of the image data decompression unit has not been started. (2) When the FIFO of the image data decompression unit is full. (3) HSYNC after the first data of 16 lines comes
When reading of all data of the number of bytes embedded in the data is completed during counting of 16 times.

To stop the transfer from the control unit, the transfer start bit is disabled.

As described above, in this system, 16
There are five color modes of 32 colors, 64 colors, 128 colors, and 16M colors, and the colors are displayed in the YUV system. Here, Y represents luminance, and U and V represent color differences.

For the 16 colors, 32 colors, 64 colors and 128 colors, a color palette is used as before, and for 16M colors, color data is directly specified. The reason is to save memory size.

For example, 4 bits are required to represent 16 colors, but if these 4 bits are used as color data directly, 16 colors will be fixed. So 5 out of 65536 colors
A color palette that can specify twelve color data is prepared, and the offset of the color palette is specified for each screen, and the color palette within one screen can be selected with the above four bits. By doing so, any 16 colors can be selected and displayed from 65536 colors, and flexibility is created.

However, as described in the prior art, 16
If a color palette for M colors is prepared, many Mbytes of memory are required, which is not practical. For this reason, color information is directly given to color data in 16M colors.

The problem here is how many bits of color data are required to represent one dot.
In the case of 16M colors, which is the largest, a size representing 16M colors is required, so 24 bits are required to represent one dot of color information. This is because 16M = 16 × 1K × 1K = 2 ⁴ × 10 ¹⁰ × 10 ¹⁰ = 2 ²⁴ (1K = 1024 = 2 ¹⁰ ).

Therefore, the color data is set to 8 for Y, U and V respectively.
Allocate bits to represent color. The virtual screen in the KRAM is prepared larger than the real screen. Its size is 10
Since it is 24 × 1024 dots, if a virtual screen is prepared for the 16M color mode, 24M
Bit memory is required.

However, since the 16M color mode is generally handled by a video or a natural image captured by an image scanner, the size of the actual screen may be sufficient. That is, in practice, it may be about 1/16 of the virtual screen. However, 1.5 Mbits are required.

In the 16M color mode, (Y ₀ Y ₁ U ₀ V ₀ , Y ₂ Y ₃ U ₁ V ₁ ,..., Y _n-2 Y _{n- 1} U _m V _m ) where m = (n−1) / 2. FIG. 10 shows the relationship between the vector factor and the dots in the 16M color mode.

This is because U and V of adjacent dots are common, and the difference is represented by luminance Y. As it is shown in FIG. _{10, Y 0 Y 1 U 0} V 0 is divided into Y ₀ U ₀ V ₀ and Y ₁ U ₀ V _0. There is no problem with image data like this,
This is because there is a fact that there is no great difference between the colors of adjacent dots in a natural image.

As described above, two dots can be represented by 32-bit color data, so that the average of one dot and color data are not 24 bits but 16 bits, and 1 Mbit of data for one screen is sufficient. .

As described above, in the 16M color mode, one vector factor corresponds to two dots. This vector factor is sent from the controller unit to the image data decompression unit in units of 16 rasters.

In the case of a one-to-one vector factor, the read vector factor may be sent as it is, so that the ratio control is easy. However, this is not the case for a one-to-two vector factor. This is because non-display dots are also sent in the case of horizontal scrolling of odd-numbered dots. Therefore, the timing for reading the vector factors is required.

An object of the present invention is to develop a method for realizing horizontal scrolling when dots and color vectors are not in one-to-one correspondence.

[0046]

The present invention solves the above problems.
In order to decide, one vector factor must correspond to two dots.
Read the image data formed by the vector data,
The image data is expanded into data corresponding to one dot.
Image processing system to transfer to the video encoder unit for display
Standards prepared internally in the stem
Data acquisition timing to the desired horizontal scroll dot.
With shifting DOO few minutes, in the case of horizontal scrolling of (1) an even dot, the data
The image data is read at the acquisition timing. (2) In the case of horizontal scrolling of odd-numbered dots,
Image data is acquired one dot before the acquisition timing.
Data from the data acquisition timing
The display period is a predetermined period after the timing of the dot.
Image processing system that scrolls the image horizontally
It is.

Standards prepared in advance in the device
By reading the vector factor at the data acquisition timing of
The vector factor is divided into a luminance component and a chrominance component, and the data can be transferred to a video encoder unit for display. That is, the timing of reading the vector factor without horizontal scrolling is the acquisition timing.

The horizontal scroll is realized by shifting the acquisition timing. However, the read timing varies depending on whether the dot is an even dot or an odd dot. Specifically, it is as follows.

In the case of an even number, the acquisition timing is set to the desired water
It is shifted by the number of flat scroll dots and this shifted
Reads image data at the same timing as the acquisition timing
From the acquisition timing,
A predetermined period after that is a display period.

For example, to scroll two dots to the left of the screen (horizontal scroll amount +2), (Y ₂ Y ₃ U ₁ V ₁ ,..., Y _n-2 Y _n-1 U _m V _m , Z) , M = (n−1) / 2 are the vectors to be read.

Regarding which data is read from,
It can be specified by the CX register. That is, a value corresponding to the value of BXR (actually the counter value) shown in FIG. 3 is set in this register, and the first data to be read is determined.

The content of Z differs depending on the tea caddy mode or the non-tea caddy mode. In the tea caddy mode, it is Y ₀ Y ₁ U ₀ V ₀ , and in the non-tea caddy, it is transparent color data. Note that the tea caddy mode is a mode in which the scrolled out vector factor is wrapped around the other end of the screen and displayed.

In the case of an odd number, the data acquisition timing
Reads image data one dot earlier
Only a predetermined dot after the data acquisition timing.
With the display period for a predetermined period after the imine grayed, most
Unnecessary dots among dots corresponding to the first vector factor
In the display period (HDISP).

[0054] For example, to screen left three dots scroll (horizontal scroll amount +3) _{is, (Y 2 Y 3 U 1} V 1, Y 4 Y 5 U 2 V 2, ...... Y n-2 Y n-1 U _m V _m , Z) Here, m = (n−1) / 2 is a vector to be transferred.

This is the same as the case of scrolling two dots to the left of the screen. However, by shifting the timing of reading, the first Y ₂ U ₁ V ₁ is prevented from covering the display period, and as a result, three dots are scrolled. So that it scrolls.

[0056]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In the system of the present invention, the screen display control signals are VSYNC (V blank) and HSYNC (H blank).
In addition to the interrupt signal, there is DCK.

The HSYNC signal is transmitted from the video encoder unit to each device. Interrupt processing is activated based on the signal, and the controller unit sends image data compressed in units of 16 rasters to the image data decompression unit. When this is received by the image data decompression unit, it is restored to the original data format and temporarily stored in the RRAM.

DCK is an interrupt signal in units of one dot, which makes it possible to interrupt each time one dot is displayed, so that operations in units of dots can be performed. The present invention also utilizes this DCK. Below,
The clock cycle of DCK is simply referred to as a clock.

Here, when there is no horizontal scrolling, 1
Three examples of scrolling to the left by one dot and scrolling to the right by one dot will be described. The number of dots to scroll is determined by the first vector factor to be read (2 dots, 3 dots described above).
(Refer to the dot example.) Then, even dot scrolling or odd dot scrolling can be resolved at the timing given here.

FIGS. 11, 12, and 13 show examples of read timings when the horizontal scroll amount is 0 (no scroll), +1 (one dot scroll left), and -1 (one dot scroll right).

When the horizontal scroll amount is 0, as shown in FIG. 11, Y ₀ Y ₁ U ₀ V ₀ is read in accordance with the acquisition timing, and it is divided into Y ₀ , Y ₁ , U ₀ , V ₀ and displayed. Transfer to the video encoder unit according to the period. At the timing when the first dot (pixel) is displayed, the next Y ₂ Y ₃ U
₁ V ₁ is read into the buffer. By this repetition, normal image display can be performed.

In the case of the horizontal scroll amount + 1, as shown in FIG. 12, Y ₀ Y ₁ U one clock before the acquisition timing.
Read the ₀ V _0. As a result, the first Y ₀ U ₀ V ₀ is prepared in the buffer, but since it has not entered the display period, it becomes an INVALID signal and is not displayed. Therefore, the first display is from Y ₁ U ₀ V ₀ . That is, it is shifted to the left by one dot.

In this case, the last dot is INVALID
However, in the tea caddy mode, the first Y ₀ U ₀ V ₀ is transferred to the video encoder unit. INV in non-tea caddy mode
Send ALID signal. That is, in the non-cascading mode, the last dot is transparent.

In the case of the horizontal scroll amount −1, as shown in FIG. 13, Y _{0 is} delayed by one clock from the acquisition timing.
Read Y ₁ U ₀ V ₀ . As a result, the first dot is not displayed, and Y ₀ U ₀ V ₀ is displayed from the second dot. That is, it is shifted to the right by one dot. However, the above example is the case of the non-cascading mode.

In the tea caddy mode, at the same timing as the horizontal scroll amount + 1, the last vector factor Y _n-2 Y _n-1 U _m V _m
Is read, Y _n-1 U _m V _m is displayed in the first dot. From the second dot is not it read in normal sequence from the _{Y 0 Y 1 U 0 V 0} . 8 to 10 have been mainly focused on the reading timing, FIGS.
FIG. 3 is a diagram mainly showing the timing of sending to the video encoder unit.

Data to be transferred to the video encoder unit is YYUV units, and decoding to YUV is performed by the video encoder unit. Even if transfer is performed in YYUV, YUV that does not take the display period is not displayed. In the present invention, scrolling is realized at this timing.

In FIGS. 14, 15 and 16, RT0 to RT
Reference numeral 7 denotes a data bus, and RTCODE1 to RTCODE0 represent types of image data on the data bus. FIG. 17 is a diagram showing an interface between the image data decompression unit and the video encoder unit.

FIG. 18 shows an example of the above-mentioned horizontal scroll. Here, m is (n-1) / 2, INV
ALID is not displayed (transparent color display), "reading timing" represents a delay relative to the acquisition timing of the vector factor to be read first, "+" is read earlier, and "-" is read later. FIG. 19 shows the arrangement of YYUV on the memory.

[0069]

According to the present invention, the horizontal scroll can be smoothly performed by controlling the timing of reading and sending the image information. As a result, smooth scrolling can be performed even if the vector factor information and the dots for coloring are not necessarily one-to-one with respect to the image forming the image.

Therefore, data is provided in a compressed form of color data having abundant color information such as a natural image.
This has the effect of saving the memory and increasing the transfer speed.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of an H blank and a V blank.

FIG. 2 is an explanatory diagram of a relationship between a color vector and a dot.

FIG. 3 is an explanatory diagram of a relationship between a virtual screen and a real screen.

FIG. 4 is an explanatory diagram of scrolling of a screen.

FIG. 5 is an explanatory diagram of a relationship between a dot and a color palette.

FIG. 6 is a system block diagram of an example of an image processing system capable of handling a natural image.

FIG. 7 is a block diagram of an image data decompression unit.

FIG. 8 is a conceptual explanatory diagram of the flow and timing of image data.

FIG. 9 is an explanatory diagram of a register of the image data decompression unit.

FIG. 10 is an explanatory diagram of a relationship between a vector factor and a dot in a 16M color mode.

FIG. 11 is an explanatory diagram of transfer timing when the horizontal scroll amount is 0.

FIG. 12 is an explanatory diagram of transfer timing when the horizontal scroll amount is +1 dot.

FIG. 13 is an explanatory diagram of transfer timing when the horizontal scroll amount is −1 dot.

FIG. 14 is an explanatory diagram of the timing of sending to the video encoder unit during horizontal scrolling.

FIG. 15 is an explanatory diagram of the timing of sending to the video encoder unit during horizontal scrolling.

FIG. 16 is an explanatory diagram of the timing of sending to the video encoder unit during horizontal scrolling.

FIG. 17 is an explanatory diagram of an image data decompression unit / video encoder unit interface.

FIG. 18 is a diagram showing a horizontal scroll amount and read timing.

FIG. 19 is a diagram showing an arrangement of YYUV on a memory.

Claims (1)

(57) [Claims] 1. A vector in which one vector factor corresponds to two dots.
Image data formed by the
Video that expands image data into data corresponding to one dot
Image processing system to transfer to the encoder
System, a standard data acquisition
Is shifted by the desired number of horizontal scroll dots.
In the case of (1) an even dot horizontal scrolling, the data
The image data is read at the acquisition timing. (2) In the case of horizontal scrolling of odd-numbered dots,
Image data is acquired one dot before the acquisition timing.
Data is read, and the timing is a predetermined dot after the data acquisition timing.
By setting the display period to a predetermined period after
An image processor comprising means for performing horizontal scrolling.
Management system.