JPH0219072A - Transmitter - Google Patents

Abstract

PURPOSE:To send even a picture with low spatial correlation with compression coding by assigning a difference of a data of two picture element from a data of the specific picture element adjacent to a specific picture element to a digital data in a prescribed bit number according to a prescribed function, sending the result and changing the function into a function of an inverted value when the difference is large. CONSTITUTION:A central processing unit(CPU) 12 reads a digital data in a display memory 25 for each basic block to apply compression coding respectively and sent to a digital telephone line via a telephone line interface 16. Then the digital data of a specific picture element is sent in a prescribed bit number and a difference of a digital data of two picture elements adjacent from the specific picture element spatially in horizontal and vertical directions is assigned to a digital data with a smaller bit number than the prescribed bit number and sent. Moreover, the assignment is varied in response to the quantity of the difference and flag information representing the assignment change is added and sent. Thus, even a picture with a low spatial correlation is decoded with fidelity.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a transmission device suitable for use in, for example, a digital videotex communication terminal.

[Summary of the invention]

The present invention relates to a transmission device suitable for use in, for example, a digital videotex communication terminal, and compresses and encodes digital data consisting of a predetermined number of bits each of four pixels arranged in two rows in the horizontal direction and two columns in the vertical direction. In a transmission device that transmits digital data of a specific pixel using a predetermined number of bits, the specific ii! The digital data of two pixels spatially adjacent to the i-element in the horizontal and vertical directions is transmitted by assigning the difference value between the i-element and the specific pixel to digital data having a smaller number of bits than the predetermined number of bits; By changing this assignment according to the magnitude of the difference value and adding flag information indicating the change in this assignment for transmission, images with low spatial correlation can be compressed and encoded and transmitted. This allows a more faithful image to be decoded.

[Conventional technology]

Videotex (interactive character and graphic information system) systems configured on analog telephone networks are limited to receiving information at approximately 4,8K, bρ3, and upstream information from user terminals to approximately 75 bps. Therefore, upstream information from user terminals is used only for providing information.

On the other hand, for example, the digital videotex shown in Fig. 7, which is configured on the digital telephone network that is being constructed in recent years, can use transmission pit rates up to 64 kbps for both the upper and lower ends, so it is possible to transmit still images etc. from the user terminal. Various types of information can be provided in a short period of time. In this Figure 7, +1) is a digital exchange, (2a),
...(2z) is a digital telephone, (3a)
, ... (3z) are Videotex communication terminals, (
4a), . . . (4z) are video tape recorders (VTRs) as still image information sources, and (151 and (6) are information centers, respectively.

Videotex communication terminals (3a), (3z), etc. are stationary 1! It has a function of compressing and encoding i-image data and transmitting it, and in addition to manually transmitting still image information to the information center 15) or (6), image information can be exchanged between communication terminals.

One such compression encoding method for still image data is
BlockT coding method (＾adaptive BlockT)
runcation Coding -A B'r
There is a method called C method). In this block encoding method, the cathode ray tube (7) of the communication terminal shown in Figure 88
For example, 768 dots (pixels) in the horizontal direction displayed on
× Screen (8) consisting of 480 dots in the vertical direction is divided into basic blocks B1j (1≦l≦M, 1≦
j≦N). To transmit this screen data, first 8 basic blocks are compressed and encoded and then transmitted, and then basic blocks B12, . ...BINIB21.
. . . are each compressed and encoded and transmitted. In order to compress and encode each basic block, the basic block (
1/4 block (4×
4 blocks) (Figure 9B) and its basic blocks as 1
After introducing the concept of 1/16 block (2×2 block) (FIG. 9C) obtained by dividing into 6 blocks, the encoding modes shown in Table 1 are used in combination.

Table 1 Encoding mode Also, the processing block unit in the basic block is [m n
J, where m indicates an m x m block, n indicates the position within the basic block shown in Figure 1B, and the processing block rmnJ is encoded in each encoding mode in Table 1 When performing this, a "mode flag" shown in FIG. 10A is added before the data information of the basic block. This "mode flag" information allows the receiving side to decode the N-image data.

The encoding algorithm of the conventional block encoding method for compression encoding an 8×8 basic block by combining the above-mentioned encoding modes is shown in FIG. 11. respectively indicate the standard deviation of the data values of all pixels of the processing block, threshold values E1 and E2 indicate the allowable values of the standard deviation, respectively, and ao and al are the upper left corner and upper right corner, respectively, of dividing the processing block into four equal parts. The data values of the corner pixels, dl, d2 and d3, are respectively at-a+1. That is, a kind of variation in data values, threshold values DI, D2 and D
3 indicates the allowable value of variation in data values, therefore,
When both the standard deviation 81 and the variation dl are smaller than the allowable values, the process proceeds from step (100) to step (104) and is processed in A8 mode, and all 64 WA values of the basic block are set to the same value (for example, the average value). ) is transmitted as Further, when the standard deviation εl or the variation d1 is larger than the respective allowable value, the process moves to step (105), and the process moves to step (108) depending on whether the standard deviation a2 of the 4×4 block is smaller or larger than the threshold value E2. )
Proceed below or step (112) or below, step (1)
12) The following shows the sign of the 2×2 block. First, the variation d3 is obtained in step (113), and then when the variation d3 is smaller than the threshold value D3, the process proceeds from step (114) to step (115). , when the variation d3 is greater than or equal to the threshold value D3, step (1
Proceed from step (14) to step (116).

Step (116) shows encoding in the D2 mode, and in this D2 mode, as shown in FIG. 12, the difference amount Δ1. Δ2. Find Δ3. If the signs of Ha and a o - a 3 are also used as data values that can be expressed by 8-bit digital data of each pixel, the difference amounts are Δ1ace and Δ1ace, respectively.
ao, Δ2-a2-ao.

Δ3-a3-a,). Then, the data value aQ is expressed as 8-bit digital data, that is, a value of θ to 255 in decimal notation, and the difference amounts Δl, Δ2 . Each of Δ3 was expressed as 4-bit digital data, that is, a value of θ˜15 in decimal notation. In this case, the difference amount Δ1(-ax aO)
The theoretical range of values is -255≦a1 ao≦
255, and let f be a function that makes these difference amounts correspond to a range of O to 15, then 0≦f (al-ao)≦15. Examples of such conventional functions are shown in Table 2.

Table 2 Conventional D2 mode conversion table Cf, ao (8 bits): 0 to 255 (10
According to this Table 2, for example, al-ao m25
5, the transmitted data is f (at ao)
-15, and when this transmission data is decoded, the difference a1-
aQ will end up being 100. Therefore, the difference tac-ao, which could be transmitted relatively accurately in the past, is (atao
)', then -100≦(ax ao)'≦100
...(1). However, - the smallest adjacent to boat fishing like D2 mode! When processing i elements, each pixel a
Q.

There is a strong correlation between the data values of all m21 a), and the amount of difference falls within the range of equation (1). As shown in FIG. 13B, this D2 mode transmission signal includes 8-bit 26, 4-bit f (a, -ao), f (a2-
ao) and f (am-ao),
All digital data consists of additional bits. On the other hand, without compressing and encoding the data, as shown in FIG.
If 1 is transmitted as is, the digital data will consist of feather bits in total. Therefore, the data compression ratio in this L)2 mode is 20/23=I2/3.

[Problem to be solved by the invention]

However, in the conventional D2 mode that compresses and encodes the 2×2 blocks mentioned above, it is assumed that the correlation between images is originally high in minute areas, and when images with low spatial correlation are compressed and encoded, the image becomes There was an inconvenience that distortion occurred. That is, for example, a thin black diagonal line (91 (11th place area value = 0) and a white background (10) shown in the 14th country)
, (11) (each value -255) is digitized and encoded into 8 bits using, for example, an image pickup tube and a digital conversion circuit. In this case, the human image has the same brightness for each pixel, so 8814
This results in a pattern as shown in Figure B. In FIG. 14B, for the 2×2 block Sl, aom255. ax-a2-127. al-0...
(2), and for the 2×2 block s2, ao −
0, at m32-127. The image (hereinafter referred to as encoded image) obtained by decoding the signal obtained by encoding this input image (al-255...13) in the conventional D2 mode is as shown in FIG. 14C. In FIG. 14C, block S
The value of each pixel of block S E t corresponding to 1 is a o = 255 + a 1- a 2- a 3
= 155, and the block S corresponding to block S2
The value of each pixel of E2 is ao-Olat-at-ax-100. Therefore, the entire encoded image has greater distortion than the input image.

In this case, the difference amount Δ1. Δ2. Distortion can be improved by increasing the number of bits for transmitting Δ3, but this has the disadvantage of increasing the amount of code.

The present invention has been made with consideration to these points, and is based on 2 lines x 2 lines.
The purpose of the present invention is to reduce distortion caused by encoding without significantly increasing the amount of code, even for images with low spatial correlation, in a transmission device that compression encodes and transmits digital data of four pixels in a column.

[Means to solve the problem]

A transmission device according to the present invention compresses and encodes and transmits digital data consisting of four pixels arranged in two rows in the horizontal direction and two columns in the vertical direction, each having a predetermined number of bits, for example, 8 bits. As shown in Figure 4,
The digital data of the specific picture IA ao is transmitted in its predetermined 8 bits, and the digital data of the two pixels a1 and at, which are spatially adjacent to this specific pixel aQ in the horizontal and vertical directions, is transmitted in its specific bits. The difference values Δ1 and Δ2 from the image iao are assigned to digital data with a smaller number of bits than the predetermined number of bits, for example, 4 bits, using a predetermined function, and transmitted. Accordingly, the function for the assignment is switched to, for example, f in Table 2 or g in Table 3, and information consisting of, for example, a 1-bit mode flag indicating the change in assignment is added and transmitted. It is.

[Effect]

According to the present invention, three strokes 1i ao, at,
When the absolute value of the difference value Δ1 (=ax ao) or Δ2 (−at−ao) of the data of at is, for example, 100 or less, the function f whose upper limit of the inverse transformation value in Table 2 is 100 is used.
Accordingly, the difference values Δ1 and Δ2 are converted into 4-bit data f (at -ao) and f (at -ao), respectively.
) and transmit it. Then, the data difference value Δ1
Or, when the absolute value of Δ2 exceeds 100, the difference value Δ
l and Δ2 are each 4-bit data g(ax ao)
and g (az-ao) for transmission. Then, 1-bit mode flag information is added to the transmission data, which becomes a low level @01 and a high level 61' when the allocation functions are f and g, respectively.

Therefore, the terminal receiving this data can identify the difference in the allocation function from the mode flag information, and if the function is g in Table 3, g
The difference values Δ1 and Δ2 can be obtained by inverse transformation of Δ1 and Δ2. In addition, the dynamic range of the inversely transformed value of the function g is wider than the function f J:, so the spatial correlation is low and the difference value is 1.
Even if an image exceeding 00 is compressed and encoded and transmitted,
By decoding, a faithful image can be reconstructed.

In this case, the dynamic range of the inversely transformed value of the function g is wide, but the resolution is also coarse in proportion to it. However, the human eye can discern the difference in brightness with fine resolution for images consisting of patterns with high spatial correlation and small differences in brightness and darkness, but for images consisting of patterns with low spatial correlation and large differences in brightness and darkness. There is no problem because it can only be identified with a relatively coarse resolution.

Further, regarding the increase in the amount of code due to mode flag information, for example, only about 1 bit is added to 20 bits of digital data, so there is almost no increase.

〔Example〕

Hereinafter, an embodiment of the transmission device of the present invention will be explained with reference to FIGS. 1 to 6. This embodiment is a stationary ll! It can compress and encode i@ data and transmit it, and can be used, for example, as a Videotex communication terminal.

Figure S1 shows the transmission device of this example, and in this Figure 1,
(12) is the central processing unit (CPU), (13) is R
OM, (14) is a RAM. Further, (15) indicates a connection terminal with a digital communication line, and the CPU (12) sends compressed and encoded data to the connection terminal (15) via the telephone line interface (16). moreover,(
17) is an image input device such as a digitizer, and (18) is various input/output circuits that perform input/output with a keyboard (19) and the like.

In addition, (20) indicates an input terminal, and this input terminal (20
) is a VTR (not shown) as a still image information source.
etc. to manually generate video signals. This video signal is supplied to a synchronization separation circuit (21), and the separated horizontal and vertical synchronization signals are supplied to a display timing pulse generation circuit (22), respectively. This display timing pulse generation circuit (22) mainly generates timing pulses for analog/digital conversion and timing pulses for writing into memory.

The above-mentioned video signal is also supplied to the I (GB separation circuit) at the same time, and this RGB separation circuit (23) generates a red signal R, an edge double signal, and a blue signal B from the video signal, and from the built-in multiplexer. These three primary color signals are time-divided and supplied to an analog/digital (A/D) converter (24). It is converted into digital data and supplied to the display memory (25).

The display memory (25) has a capacity to store screen digital data for one frame or more, and the display memory (25) has a capacity to store screen digital data for one frame or more.
5) is to convert the three primary color signals R, G, and B into analog signals in parallel using a digital/analog (D/A) converter (26), and convert them into analog signals via a cathode ray tube drive circuit (27). feed into tube (28).

In addition, the PU (12) sequentially outputs the digital data in the display memory (25) for each basic block Bij as shown in FIG. Send on line.

Stationary i! of the transmission device in this example! The compression encoding method for data in ill is basically a block encoding method (ABTC method), and its encoding algorithm is shown in Figure 2. In this Figure 2, the steps corresponding to Figure 11 have the same codes. , and detailed explanation thereof will be omitted.

The difference in this example from the example in FIG. 11 is the step (114
) is the processing after d3≧D3 (when the brightness change is relatively large), and in the subsequent steps, the horizontal direction 2 shown in FIG.
4 pixels arranged in rows x 2 columns in the vertical direction or 2 rows in the horizontal direction x 2 columns in the vertical direction (drag I') aO + al + a
2 + al is used. In this example, the signal of each of these pixels is either the red signal R, the edge double signal, or the blue signal B.
For example, it may be a luminance signal Y, etc., and the digital data of these four pixels are each 8 bits (0 to 2 in decimal).
55), containing a total of 32 bits of information,
In order to compress and encode this 32-bit data, in this example! In step (117) of the fsZ diagram,
Δ1. shown in FIG. 4 as a difference value. Δ2. Δ3 and Δ3'
Calculate. These difference values are 4 pixels aQ, al.

a2. The digital data of al are each given the same symbol aQ =
When expressed as aL * 82 + al, Δ1
= a t a o s Δ2 a2 − aQ Δ3 ”
23 a O%Δ3'-ai a2. The difference from the conventional method is that the difference value of pixel a3 is
The difference value Δ3' between the pixel a2 and the horizontally adjacent pixel a2 is also determined. In this case, the data of the vertically adjacent pixel a1 and ii! The difference value from the data of i-set a3 may be used as Δ3'. In this example, the absolute value of the difference value Δ3 is regarded as the representative value β of the difference values between the four pixels. That is, as β-1Δ31 = l al -aQ, when this representative value β is 10o or less, the second
The process moves to step (116) in the figure, and when the representative value β exceeds 100, the process moves to step (118).

4ttL, 4W! In addition to this example, the representative value β of the difference value between the four elements may be, for example, the difference value Δ1. Δ2. A value that is the inner value of the absolute value of Δ3 or the like may be used.

In step (117), when the representative value β of the difference value between the four pixels is less than the predetermined value 100, it means that the spatial correlation of the input image is high and there is little change in brightness, and in this case step (116) The same process as step (116) in FIG. 11 (processing in D2 mode) can be performed in step (116) in FIG. There is. Further, in step (117), when the representative value β of the difference values between the four pixels exceeds the predetermined value 100, it means that the spatial correlation of the input image is low and the brightness change is large,
In this case, in step (11B), encoding is performed using the D2' mode, which is an improved version of the conventional L)2 mode. Therefore,
The encoding mode of this example has a configuration shown to third country nationals. Third
Diagram B shows the location of the corresponding block.

In the 02' mode of this example, the above-mentioned difference values Δl, Δ
2. Δ3' is converted into 4-bit data g (at-ao) and g (al-ao
), g (a3-al).

Table 3: Conversion table for 02' mode in the embodiment example ao (8 bits): 0 to 255 (decimal) The range is widened. That is, a difference value that can be transmitted relatively accurately, for example, al-ao (
ai ao)', then from Table 3, 127≦(ax-ao)'≦127...-(4
It becomes 1. Then, as shown in FIG.
First, 1-bit mode flag information of high level 61" indicating that the current processing mode is 02' mode is transmitted. Then, from time t-t2, iii element a
o 8-bit digital data is transmitted, and then 4-bit differences 1g (at-ao) and g (al-a
o) Sg (at-al) is transmitted sequentially up to time t"'t3. Therefore, according to this example, the digital data to be transmitted, excluding other mode flag information, may be 21 bits,
The data compression rate is 217 birds - 2/3, which is almost the same as before. In particular, in this example, as shown in FIG. 2, processing in the D2 mode or D2' mode is performed only when the luminance change is large, and the overall amount of encoded data is almost the same as the conventional one.

On the terminal side that received this digital data, the received difference amount g (ax ao) + g (al-ao)
g (a3-al) is inversely transformed according to Table 3 to obtain the difference values Δl, Δ2. Obtain Δ3'. Then, using the digital data of pixel aO as a reference, calculate the digital data of pixels a1 and al from L "2 ""0 + Δ2 to al-ao +, and further use the values of ii!ii and 1Ii-al found using the above formula. Te ax-3
The digital data of pixel a3 is obtained from 2+Δ3' and the decoding is completed.

The original input image of thin "diagonal lines" with low spatial correlation as shown in FIG. 14A by the transmission device of this example is shown in FIG. The digital data of each pixel of the 2×2 blocks S1 and S2 of the input image, whose encoded images are shown in FIG. 6B, is calculated using the conventional formula (2), ! It is the same as 3). On the other hand, Decryption North West @! The digital data of each pixel of the 2×2 block SEX (FIG. 6B) is ao -255+at-al-121La3-1 ・
・151, and the digital data of each pixel of 2×2 block SE2 is ao ”O+at -at-L27+a3-254
...16) Therefore, the encoded image of this example as a whole is approximately equal to the input 1jtlIl, and it can be seen that the distortion is significantly improved compared to the conventional encoded image of FIG. 14C. .

In this example, the 4 pixels aO+ al+a2+ in FIG.
When the digital data of al is compressed and encoded for transmission, the difference value Δ3' of pixel a3 is based on the data of pixel a2, which is horizontally adjacent to image sa3. in this case,
The pixel to be processed may be rectangular, but for convenience it is assumed that it is square, and if the size of one pixel is LoXLo, then the distance between pixels a, ) and at is #l L 2 and pixel a
The distance L1 between pixels 2 and a3 is Ll -L2 -LO, and the distance l1IIIL3 between pixels a6 and 83 is L3-
ft.

Therefore, Lx<Lt holds, and the spatial correlation between pixel a2 and pixel a3 is pixel aO and iyI! i element a2
generally higher than the correlation with Therefore, in this example, l! i element a 3 digital data can be transmitted. Furthermore, if the pattern of the original image is a pattern with low correlation in the diagonal direction as shown in FIG. 14A, for example, ax -ao--128, ay -32--12
It becomes 7. At this time, in this example, the difference value of pixel a2 is almost exactly 4 according to function g in Table 3 with m element aQ as a reference.
The difference value of the ii element a3 is converted to exactly 4 bits according to the function g in Table 3 with pixel a2 as a reference. Therefore, pixels aQ arranged diagonally
The difference value between pixel a3 and pixel a3 can be compressed and encoded almost accurately up to about a3 - aQ -255 and transmitted.

This means that in the compression encoding according to this example, the dynamic range in the diagonal direction is expanded to more than twice that of the conventional one, and in this example, patterns with low correlation in the diagonal direction are compressed more accurately. There is an advantage in that it can be encoded and transmitted.

In the above embodiment, the data of pixel a2 is used as the reference for the difference value of pixel a1, but the present invention is not limited to this, and the data of pixel a2 is used as the reference for the difference value of pixel a1+''2+a3. In this case, instead of the function g in Table 3, when the difference value a 2 - a o = -255, r 0OOOJ is obtained, and the difference value a2-ao=2
By using a function G such that rIIIJ when 55, the problem of the dynamic range in the diagonal direction can be solved. Furthermore, the difference value al −a, )a
The function g in Table 3 may be used for the conversion of t-ao, and the function G may be used for the conversion of the difference value a3-a,).In this way, the present invention is limited to the above-mentioned embodiments. Of course, changes may be made without further developing the gist of the present invention.

〔Effect of the invention〕

As described above, the transmission device of the present invention has two pixels adjacent to a specific pixel.
The data of one pixel is transmitted by assigning the difference value from the data of that particular pixel to digital data of a predetermined number of pits according to a predetermined function, and when the difference value is large, the dynamic range of the inversely transformed value is wider. Since it is converted into a function, there is an advantage that even images with low spatial correlation and large luminance changes can be compressed and encoded more accurately and transmitted.

Also, the only code that has increased compared to the conventional one is about 1 bit of flag information that indicates the type of function for the assignment.
The amount of code to be transmitted is approximately the same as the conventional method.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the transmission device of the present invention, FIG. 2 is a flowchart showing an algorithm of the block encoding method according to the embodiment, and FIG. 3 is a diagram showing types of encoding modes in the embodiment. , FIG. 4 and FIG. 5 are diagrams for explaining the D2' mode processing according to the embodiment, FIG. 6 is a diagram showing the encoding of "diagonal lines" according to the embodiment, and FIG. 7 is a diagram for explaining the processing of the D2' mode according to the embodiment. FIG. 8 is a diagram showing an example of the configuration of TESOX. Figure 9 is a diagram showing the configuration of the i1 plane, Figure 9 is a diagram showing the configuration of processing blocks, Figure 10 is a diagram showing the conventional encoding mode, and Figure 11 is a diagram showing the encoding algorithm of the conventional block transmission system. Flow chart diagram, 12th
13 and 14 are diagrams for explaining processing in the conventional 02 mode, respectively. (12) is a CPU, (16) is a telephone line interface, (25) is a display memory, Sl and Sl are each a 2x2 block of a human image, SEl and SE2 are each a 2x2 block of an encoded image, and aO+al+a2+a3 is Each is a pixel.

Claims (1)

[Claims] In a transmission device that compresses and encodes and transmits digital data consisting of four pixels arranged in two rows in the horizontal direction and two columns in the vertical direction, each of which has a predetermined number of bits, the digital data of a specific pixel is The digital data of two pixels spatially adjacent to the specific pixel in the horizontal and vertical directions is transmitted using a predetermined number of bits, and the difference value from the specific pixel is converted into digital data with a bit number smaller than the predetermined number of bits. A transmission device characterized in that the transmission device is configured to allocate data to data for transmission, change the allocation according to the magnitude of the difference value, and add flag information indicating a change in the allocation before transmission.