JPH05323946A - Image terminal device - Google Patents

Abstract

PURPOSE:To use a general 9-bit memory and effectively utilize the memory and to facilitate the processing of data even when 8-bit image data are handled by the image terminal device which compresses a color image (natural image) like a photograph by a discrete cosine conversion encoding system, transmits and stores the natural image, and decodes its encoded data and displays the image. CONSTITUTION:One bit showing an underflow or overflow is added to the decoded 8-bit image data and the data are written in a natural image VRAM 7; and 9 bits are read out of the natural image VRAM 7 at a time and a decision is made according to the one bit showing the underflow or overflow and the most significant digit bit of the 8-bit image data to output the 8-bit image data as they are, or output the eight bits all as '1' or '0'.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention compresses a color image (hereinafter referred to as a natural image) such as a photograph by a discrete cosine transform (hereinafter abbreviated as DCT) coding method to efficiently transmit a natural image. The present invention relates to an image terminal device that accumulates and decodes and displays the encoded data, and further detects an underflow / overflow of the decoded image data that accompanies the DCT encoding, and displays the data within a displayable data width. The present invention relates to an image terminal device that performs image data processing for preventing deterioration of a display image due to underflow / overflow by clipping the image terminal device.

In general, a rectangular wave can be expressed as a DC waveform by using a concept that it can be expressed by a Fourier series as a composite waveform of a sine wave having the same frequency as its frequency and its harmonics.
In the T-coding method, the image data arranged on the time axis is replaced with a conversion coefficient (amplitude of each harmonic) on the frequency axis to correlate adjacent pixels existing in the image data on the time axis. This is a method of compressing a large amount of information by eliminating relationships.

However, as described above, in order to express a rectangular wave by superposing and combining sine waves to obtain a perfect restored waveform, harmonic components up to an infinite term are necessary, and a practical system is constructed. Therefore, it is known that undershooting / overshooting occurs on the rising / falling (white / black) of the restored waveform when the censoring is performed by the finite term.

When the system is constructed without considering the undershoot / overshoot, a black dot is displayed or a black dot is displayed only on a part where the white is continuous (high brightness part) in the display image. An image pattern that does not exist in the original image is generated such that white spots are displayed only in an area where undershoot occurs in a continuous area, resulting in local but very conspicuous image quality deterioration.

Therefore, in order to prevent such image quality deterioration, even if an undershoot / overshoot occurs in the decoding result, it is within the dynamic range of the image data (from "0" in the case of 8-bit image data). It is necessary to perform clipping processing so that the value falls within "255").

[0006]

2. Description of the Related Art FIG. 3 shows a general configuration example of an image terminal device for a natural image in which a DCT coding system is introduced.
Reference numeral 1 is a TV camera for inputting video, and 2 is for converting an image signal (NTSC signal) from the TV camera into a luminance signal (Y signal) or a color signal (C) and further Cr (R- It is an 8-bit output video input control unit that performs A / D conversion into a digital signal such as a Y) signal or a Cb (BY) signal and performs color decoding to extract a color difference signal from a chroma signal.

Reference numeral 3 is a CPU, 4 is a memory for storing data, 5 is a communication control unit, and these are 8-bit CPs.
It is connected via the U-bus 11.

Reference numeral 6 denotes a DCT / CODEC, which is composed of an encoding unit 62 and a decoding unit 61, and has a CPU bus 1
It is connected between 1 and also an 8-bit image bus 10. As described above, the video input control unit 2 and the DCT / COD
The image bus 10 connected to the EC 6 has a 8-bit structure and is a natural image V composed of the luminance Y signal memory surface, the Cr (RY) signal memory surface, and the Cb (BY) signal memory surface.
Although not shown in the figure, it is connected to the RAM 7, and the memory data is read out from the natural image VRAM 7 under the control of the display control unit 8. The display control unit 8 conversely outputs the luminance Y signal and Cr (RY). After the signal and the Cb (BY) signal are converted into RGB and D / A converted into an analog signal,
It is designed to be displayed on the display unit (CRT) 9.

When an image is displayed on the terminal itself, the video signal of the TV camera is converted into a digital signal by the video input control unit 2, and the digital signal is temporarily stored in the natural image VRAM 7 via the image bus 10, and It is displayed on the display unit 9 by the display control unit 8.

Further, in the case of storing in the memory 4 of the own terminal, in the DCT / CODEC 6 which receives the 8-bit image data from the video input control unit 2 via the image bus 10, the encoding unit 62 performs compression processing. When the data is stored in the memory 4 via the CPU bus 11 and transmitted without being stored in the memory 4, the compressed data is sent out via the communication control unit 5 via the communication line.

In the conventional image terminal device as described above, the clipping process is performed when the decoding result in the decoding unit 61 in the DCT / CODEC 6 exceeds the bit width of the natural image VRAM 7 (for example, natural image). When the VRAM 7 has an 8-bit structure and the image data X is X <0 or X ≧ 256), when X <0, “0” is forcibly written to the natural image VRAM 7, and when X ≧ 256, the natural image VRAM.
The display control unit 8 controls to compulsorily write "255" in No. 7.

However, in recent years, when natural image communication is performed using a communication line, in order to improve response and serviceability, first, a rough image having a small amount of data is quickly transmitted, and then, Hierarchical coding technology, in which data of a finer part is added step by step, is being introduced.

In the DCT decoding method, as a hierarchical coding method like this, there is a spectral selection in which only low-order frequency components are transmitted first and high-frequency components are gradually added. The so-called method is generally used.

In such a method, undershoot / overshoot with respect to the original image is very large at first,
By gradually adding high frequency components, the values converge to the values of the original image. Therefore, the undershoot / overshoot in the middle layer becomes information that cannot be lost when the original image is finally restored. If the conventional technique is applied as it is and the above-mentioned clipping process is repeated for each layer, the final image is obtained. The problem that the error of becomes a very large value occurs.

Therefore, in the hierarchical DCT system based on the spectral selection method, the following method has been used. In the following description, the original image data has an 8-bit structure as described above, and the dynamic range from "0" to
The case where the value is "255" is shown.

(1) As shown in the conventional example of FIG. 4, a natural image V
The RAM 7 has a 9-bit configuration and the image data is “-12
Writing / reading is performed as data "8" to "385". Therefore, the video input control unit 2 adds "128" to the 8-bit image data to generate 9-bit data of "-128" to "383", and the DCT
The coding unit 62 of the CODEC 6 subtracts "128" from the 9-bit image data to compress it into 8-bit image data, and
The data is sent from the memory 4 or the communication control unit 5 to the communication line via the PU bus 11.

In order to decompress such compressed 8-bit image data to the original 9-bit data, the decoding section 61 converts the data of "-128" to "383" into "0" to "5".
11 ”9-bit data converted to natural picture VRAM7
Store at.

In order to display the natural image on the display unit 9, the display control block 8 subtracts "128" from the 9-bit image data of the natural image VRAM 7 to obtain "0" to "255".
Generates bit data, and the image data at this time is "0"
If it is below, underflow processing is performed, and if it is "255" or more, overflow processing is performed.

(2) As shown in the conventional example of FIG. 5, the decoding unit 61 includes an overflow bit (1 bit) and an underflow bit (1 bit) for each pixel in the 8-bit image data written in the natural image VRAM 7. Provided as a result 1
The image data has a 0-bit width. And at the time of display,
The display control unit 8 reads the 8-bit image data and the corresponding underflow / overflow bit from the natural image VRAM 7 and supplies them to the selector (SEL) 12, and when each bit is "0", it is 8 bits as it is. Image data is given from the selector 12 to the display unit 9. When it is "1", and when the overflow bit is "1": image data X =
255 When underflow bit is "1": image data X =
Since the forced data conversion of 0 is performed, the selector 12 is selected and given to the display unit 9.

[0020]

The above-mentioned two conventional examples (1) and (2) have the following problems, respectively. (1) In the display control unit 8, it is necessary to determine whether the underflow or the overflow has occurred, and the data to be viewed becomes large.

Since the image data is generated as "0" to "255" (8 bits), the 9-bit natural image VRAM 7 is used.
The video input control unit 2 needs to add "128" to the 8-bit image data to generate (-128) to "383" data (getting out a get) when writing to the It gets complicated.

(2) The natural picture VRAM 7 has a 10-bit structure. Currently, a commonly used memory has a width of 8 bits or 9 bits, and in order to make a 10-bit structure, it is necessary to use two 8-bit ones to make a 16-bit structure. It is very bad and causes an increase in hardware scale.

Therefore, an object of the present invention is to realize an image terminal device satisfying both of the following conditions. (1) Even when handling 8-bit image data, it is possible to use a general 9-bit memory by performing overflow / underflow processing only by adding 1-bit data, and effective use of the memory ( Improving usage efficiency). (2) Basically, the data in the device can be handled in the range of "0" to "255", thereby eliminating the need to tamper with the data, such as adding or subtracting data for each process. ..

[0024]

Means and Actions for Solving the Problems First, the focus of the image terminal device according to the present invention will be described.

As general properties of an 8-bit natural image, (1) underflow and overflow do not occur simultaneously for one pixel when DCT encoding / decoding is performed. (2) Except for extremely rare situations, the width of underflow and overflow is when data of "0" to "255" is handled. Overflow does not exceed 255 + 128 = 383, and underflow is 0-128. ==-128 or less.

Therefore, as shown in FIGS. 1A to 1C, the most significant bit (MSB) of the 8-bit image data is always "0" when an overflow occurs, and the 8-bit image when an underflow occurs. The most significant bit (MSB) of the data is always "1".

Focusing on such a property, in the present invention, the decoding unit 61 of the DCT / CODEC 6 shown in FIG. 3D outputs 1 bit of data indicating whether an underflow or overflow has occurred and outputs 8 bits. It is added as the 9th bit of the image data and stored in the natural image VRAM, and the display control unit 8 determines 1 bit thereof and 2 bits of the most significant bit (MSB) of the 8-bit image data to obtain 8 bits.
Output the bit image data as it is, output all 8 bits as all "1", or all "0".
Is output as overflow / underflow processing.

As described above, according to the present invention, the overflow / underflow process can be performed with 9 bits without tampering with the data, and the natural image VRAM can also have a 9-bit configuration.

[0029]

FIG. 2 shows an embodiment of an image terminal device according to the present invention. In this embodiment, as compared with the embodiment shown in FIG. 5, the decoding unit 61 in the DCT / CODEC 6 is shown. However, the 8-bit image data is output to the same 8-bit image bus 10 and the 1-bit underflow / overflow bit is directly transmitted / received to / from the natural image VRAM 7.
9-bit connection with AM. Similarly, natural painting VRA
The M7 and the encoding unit 62 are also connected via the 8-bit image bus 10, and 1-bit underflow / overflow bit is directly transmitted / received to connect 9-bit.

Of the 9-bit output from the natural image VRAM 7, the 1-bit underflow / overflow bit and the most significant bit (MSB) of the 8-bit image data are used as decision bits, and the natural image VRAM 7 is used. Selector (SEL) 12 for selectively outputting any of 8-bit image data from 8 bits, 8-bit data of all "1" s, and 8-bit data of all "0s" Is provided, and the 8-bit output of the selector 12 is displayed on the display unit (CRT) via the display control unit 8.
9 is displayed. Although not shown, the display controller 8
Also controls the natural image VRAM 7, and in this embodiment, the selector 12 and the display controller 8 are shown separately, but they may be integrated.

The operation of the above embodiment will be described below.

First, the decoding unit 61 judges the decoding result, and if the value is "0" to "255", the value is directly used as 8-bit data as shown in FIG. 1B,
The 9th bit underflow / overflow bit is set to "0".

However, when the data is other than "0" to "255", the data is output as follows. (1) Underflow (see FIG. 1A): For example, when it becomes “−1”, bits 0 to 7 of the image data become “11111111”, and these bits 0 to 7 are output as pixels. Then, in addition, "1" is output to the underflow / overflow bit.

(2) In case of overflow: For example, "2
When it becomes "56", bits 0 to 8 of the image data are "100000000".
, Bits 0 to 7 are output as image data, and the underflow / overflow bits are also set to "1".
Is output.

The natural picture VRAM 7 is composed of 9 bits,
The most significant bit (MSB: bit 8) is the underflow / overflow bit. This bit 8 is pulled down by hardware, and the portion that does not require the 9th bit (for example, the video input unit) has an 8-bit bus configuration, and the display control unit 8 keeps the video data in the 8-bit configuration. It is possible to access the natural image VRAM 7. (Use only the lower 8 bits)

The display control unit 8 displays 9 of the natural image VRAM 7.
Bits are read out at the same time, and the judgment and display image data output shown in the following table are made from the underflow / overflow bit (bit 8) and the most significant bit (bit 7) of the image data.

[0037]

[Table 1]

[0038]

As described above, according to the image terminal device of the present invention, a natural image VRAM is added by adding 1 bit indicating that an underflow or an overflow has occurred to the decoded 8-bit image data. , And 9 bits are simultaneously read from the natural image VRAM, and the 8-bit image data is output as it is by judging from 1 bit indicating underflow or overflow and the most significant bit of the 8-bit image data, or All "1"
Since it is configured to output either as or as all "0", the following effects can be expected in the pixel terminal device for a natural image in which the DCT coding method, particularly the hierarchical system is introduced.

Underflow / overflow processing is possible with image data + 1 bit, and when handling 8-bit data, a 9-bit width memory can be used:
It is possible to significantly improve the memory usage efficiency and reduce the number of memories as compared with the conventional example.

Since the actual image data width inside the device can be unified to 8 bits, the structure of the image bus and the bus control can be simplified.

Underflow / overflow processing can be performed only by the display control unit, and the apparatus can be miniaturized / priced down when the hierarchical processing is applied to the image terminal apparatus for a natural image using the DCT coding method. There is a large contribution.

[Brief description of drawings]

FIG. 1 is a diagram for explaining in principle an image terminal device according to the present invention.

FIG. 2 is a block diagram showing a configuration of an embodiment of an image terminal device according to the present invention.

FIG. 3 is a block diagram showing a general configuration of an image terminal device for natural images.

FIG. 4 is a block diagram showing a conventional image terminal device (No. 1).

FIG. 5 is a block diagram showing a conventional image terminal device (No. 2).

[Explanation of symbols]

 2 video input control unit 6 DCT encoding unit 61 decoding unit 7 natural image VRAM 8 display control unit 10 image bus In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (1)

[Claims] 1. A display control unit (8) displays the 8-bit image data encoded by using the discrete cosine transform by a decoding unit (61) and a natural image VRAM (7) having a 9-bit structure. In the image terminal device to be displayed by the means (9), the decoding unit (61) adds 1 bit indicating that an underflow or an overflow has occurred to the decoded 8-bit image data, and the natural image is added. The display control unit (8) writes the data in the VRAM (7) and the natural image VRAM (7) to
Whether the 8-bit image data is output as it is by judging the 1-bit indicating the underflow or the overflow and the most significant bit of the 8-bit image data, or the 8-bit image data is output as all "1". , All “0” is output.