JPS5941630B2 - Waveform shaping method in three-level transmission - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a waveform 5 shaping method for reducing the distortion peculiar to 3-value conversion caused by the transmission characteristics in a method for transmitting a binary signal such as a facsimile after converting it into 3-values. be.

Until now, almost no methods have been considered to improve the distortion peculiar to facsimile transmissions that involve three-value conversion. This was done only with

The problem was that the received image quality remained unnatural. For example, the transmission screens shown in Figures 1A and 1C are scanned, and the binary signals obtained are converted into three values and transmitted, and the reproduction screens of the received signals are as shown in Figures 1B and 15D, respectively. .

In FIG. 1A, there is an edge Ila of the signal of the image 11 which is orthogonal to the scanning lines 11 to 14 and is narrower than one pixel width, and an edge 12 of the signal of the black image 12 which is parallel to the scanning line and is longer than three pixels.
a, or a binary facsimile signal in which the edge Ilb of the black image 20 image 11 and the edge 13a of the long black image 13 coincide with each other on the scanning line, is converted into eight values and transmitted. Then, the reproduced image is shown in Figure 1B.
As shown in the figure, the thin black image 11 becomes thick like a single line,
Although the edge shifts, the edge of the long black image 25 remains the same, so the edge portions 12a', 13 of this black signal
Each a' is depressed. On the other hand, when the width of the thin black image 14 is about 2 pixels as shown in FIG. 1C, the thin black image 14' of the reproduced image becomes thin as shown in FIG. The edge portions 12a'' and 13a'' protrude. The relationship shown in FIG. 1 is shown quantitatively in FIG. 2.

In FIG. 2, the horizontal axis represents the pulse width of the black signal on the transmitting side, and the vertical axis represents the difference between the receiving side pulse width and the transmitting side pulse width of the black signal. The case where the received pulse width is wider than the transmitted pulse width is taken as positive, and the case where it is narrowed is taken as negative. Tp is 1
The pulse width of the pixel is C, and the transmission signal is band-limited by an ideal filter with a cutoff frequency of 1/4 Tp (Hz).

A curve 15 is the difference characteristic of the transmission and reception pulse widths when a ternary signal passes through the filter, and the difference becomes almost O when the transmission pulse width is 3Tp or more. Next, the sampling period of the transmission signal is Tp/2 (shown as an example in this explanation)
The difference characteristic of the transmission and reception pulse width when sampling and holding is shown by a curve 16, and the shaded area in the figure is the sampling jitter. As can be seen from this curve 16, the edge of the image with a long black signal (
(the difference corresponds to zero), in order to relatively cancel out the unevenness of the thin line edge (the difference value of the transmitting and receiving pulse width characteristics),
As shown by the dotted line 17 in the figure, a long black signal edge that touches a black signal edge of 1 sampling width has a black signal added by Tp/2, and a long black signal edge that touches a 4-sampling and 5-sampling black signal edge has a black signal of Tp/2. It is sufficient to suppress the black signal by Tp/2. In this invention, for a long black or white image on the next scanning line where the thin black or white image edges coincide, if the thin black or white image is less than a predetermined value, the edge of the long black or white image signal is black. Or, by adding a white signal and removing the black or white signal at the edge of the long black or white image signal when the thin black or white image has a certain width range larger than the predetermined value, the edge of the reproduced image is The present invention provides a waveform shaping method for three-level transmission that does not cause unevenness. FIG. 3 shows an embodiment of the present invention, and FIG. 4 shows waveforms at various parts thereof.

In FIG. 3, a binary signal such as a facsimile signal from an input signal terminal 21 is input to a sampling clock signal input terminal 22.
The sampling period from Tp/2 is stored in the shift register 23 for one scanning line. The output of the shift register 23 is supplied to the output terminal 24, and the black signal length is counted by the black signal counter 25.
Further, the output of the shift register 23 is supplied to a white signal counter 27 through an inverter 26, and the white signal length is counted. Based on the count value of the black signal counter 25, the detection circuits 28, 29, and 31 detect if the black signal length is an edge black signal of 1, 4, or 5 bits, respectively. If the length of the white signals on both sides of the black signal is, for example, 3 bits or more each, the circuit 32 determines that the black signal is a rising black signal. A circuit 33 detects a black signal edge time with a long pulse width in the input signal. This circuit 33 detects, for example, a pattern of a black signal of 7 bits or more before and after a white signal of 3 or more consecutive bits as a means of a long black signal. Load 10
This can be realized with a bit shift register. i.e. black
1F, if white is set to ``O'', the output is detected when the contents of the register 33 are 0001111111 or 1111111000. The edge of the long black signal in the current scan is detected by the detection circuit 33, and the edge of the long black signal in the current scan is detected by the detection circuit 33.
The 1, 4, and 5 bit edge black signal detection outputs are detected by circuits 34, 35, and 36, respectively. If the previous scanning signal is a 1-bit rising black signal, a 1-bit black signal is added to the black signal edge of the current scanning line by the output of the circuit 34.
If the previous scanning signal is a 4- or 5-bit rising signal, the output of the circuit 35 or 36 suppresses the 1-bit black signal from the black signal edge of the current scanning line. By the above operating procedure, the unevenness characteristic of the curve 17 in FIG. 2 is obtained. FIG. 4A shows the sampling clock signal waveform of the terminal 22, and FIG. 4B shows the signal waveform of the previous scanning line, with black set to a low level and white set to a high level.

C in the same figure shows the input signal waveform of the current scanning line from the terminal 21. The signal at the beginning of the scan line is at the far right in FIG. D in the same figure is a detected waveform of a rising black signal in the signal waveform of the previous scanning line in B in the same figure.The length of the black signal 37 is 1 bit, and the white signal immediately before it is 3 bits or more. Immediately after that, when 3 bits of the white signal are counted, a pulse 43 is output as a 1-bit high-rise black signal. Similarly, 1-bit black signal 38
, 39, a rising 4-bit black signal 41, and a rising 5-bit black signal 42 are sequentially detected as pulses 44, 45, 46, and 47. As explained in FIG. 3, the long black signal 48, 4 connected with 7 bits or more is connected to the white signal connected with 3 bits or more.
9 and 51 are sequentially detected by the circuit 33 as pulses 52 to 57, respectively, as shown in FIG. 4E. The signal of the previous scanning line shown in FIG. 4B is now entering the shift register 23, and the signal at the right end of FIG. As time progresses, the shift register 23 is shifted, and the previous scanning line signal is output from the right end, and the current scanning line signal is input from the right end of FIG. It is also input to the bit shift register. When 9 clock signals are input to the terminal 22 and a 9-bit current scanning line signal is input to the shift register 23 as shown in FIG. 5A, the previous scanning line signal output from the shift register 23 is checked. As a result, a detection pulse 43 for the 1-bit high-rise black signal 37 is obtained from the circuit 28. In this way, input of the current scanning line signal to the shift register 23 progresses, and as shown in FIG. 5B, the first to seventh bits of the shift register 23 are input.
All bits are black signal "11 8th bit to 10th
When all the bits are in the state of white signal "0", the shift register in the detection circuit 33 is also in the same state 6111111.
10001, and the leading edge of the long black signal 48 in the current scan line signal is detected as a pulse 52. At this time, as shown in FIG. 4D, the 1-bit high-rise black signal 38 in the previous scanning line signal whose leading edge and leading edge coincide with each other is detected as a pulse 44 3 bits earlier. . The pulse 44 from the 1-bit edge black detection circuit 28 is delayed by 3 bits and the leading edge detection pulse 52 of the long black signal from the detection circuit 33.
, the output of the circuit 34 causes the shift register 23 to
Change the white signal "'O" of the 8th bit to the black signal "1" and add 1 bit of the black signal to the leading edge of the black signal 48 (
Figure 4F, 58). As shown in FIG. 5C, the input state of the current scanning line signal to the shift register 23 is such that the first to third bits are a white signal "O" and the fourth to tenth bits are a black signal "1". Then, the shift register in the detection circuit 33 becomes 600011111113, and the long black signal 4
The trailing edge of 8 is detected as pulse 53. At this time, the trailing edge of the long black signal 48 coincides with the trailing edge of the 1-bit standing black signal 3.
9 is detected as pulse 45, and these pulses 45, 5
3 match, the output of the circuit 34 is transferred to the shift register 23.
The third bit of the white signal "O" is changed to the black signal "1" and one black signal bit is added to the trailing edge of the black signal 48 (the fourth bit is
59 in Figure F). As shown in FIG. 5D, the shift register 2
3, the 1st to 7th bits become "1", the 8th to 10th bits become "O", and the leading edge of the long black signal 49 becomes the pulse 54.
Detected as .

At this time, the detection pulse 4 of the 4-bit edge black signal 41 whose leading edge coincides with the leading edge of the black signal 49 in the previous scanning line signal
6 is obtained in circuit 29. Due to the coincidence of these pulses 46 and 54, the output of the circuit 35 changes the 7th bit black signal "1" of the shift register 23 to the white signal "0n" and suppresses the 1-bit black signal from the leading edge of the black signal 49. (Figure 4 F
61). As shown in FIG. 5D, when the first to third bits of the shift register 23 become a white signal "0" and the fourth to 10th bits become a black signal "1", the trailing edge of the long black signal 51 becomes a pulse 57. At this time, the detection pulse 4 of the 5-bit rising black signal 42 whose trailing edge coincides with the trailing edge of this black signal 51
7 is obtained from circuit 31.

Due to the coincidence of these pulses 47 and 57, the output of the circuit 36 turns the fourth bit of the black signal "1" of the shift register 23 into a white signal "O", suppressing the 1-bit black signal from the trailing edge of the long black signal 51. (62 in Figure 4 F). Note that the detection of the arc black signal is performed using the signal output from the shift register 23,
In other words, it is performed from a signal obtained by adding or removing 1 bit of the black signal from a long black signal, but for the rising black signal, 1 bit and 4 bits are added.
This is a 5-bit stand-up signal, and the long black signal to be corrected is 7 bits or more, so even if this long black signal is corrected, the length is 6 bits or more, so the shift register 23 is The edge black signal in the previous scanning line signal can be correctly detected from the output.

As described above, a long black signal edge 13a is determined from the pulses 44 and 52, which coincides with the 1-bit edge 11b of the scanning line 14 in FIG. 1, and the 1-bit edge 58 in FIG. Similarly, pulses 45,
53, the edge 12a of the scanning line 12 in FIG. 1 is detected, and 1 bit 59 is added. From the relationship between the pulses 46 and 54, the edge 13a of the scanning line 16 in FIG. 1 is detected, and the 1 bit 61 is removed. , similarly from pulses 47 and 57 to the first
Edge 12a of scanning line 15 in the figure is detected and 1 bit 6
2 is removed. The output of FIG. 4F whose waveform has been shaped in this way is transmitted from the output terminal 24 of FIG. 3. Although the case of the black signal has been explained above, in the case of the white signal as well, the output of FIG. 4F whose waveform has been shaped by the same method is transmitted from the output terminal 24 of FIG.

Although the case of the black signal has been explained above, it is clear that waveform shaping can be performed using a similar method in the case of the white signal as well. As explained above, according to the waveform shaping method of the present invention, in facsimile transmission involving three-value conversion, etc., a transmission screen in which the positions of black signals with different pulse lengths on adjacent scanning lines on the edge scanning line coincide with each other. It is possible to reduce the distortion that occurs when the image is transmitted, and it is possible to improve the image quality of the received screen.

[Brief explanation of the drawing]

Figure 1 shows an example of screen distortion that occurs uniquely in facsimile transmission involving three-value conversion, where A and B are input screens, C,
D is a playback screen, FIG. 2 is a graph quantitatively showing the distortion in FIG. 1, FIG. 3 is a block diagram showing an embodiment of the waveform shaping method according to the present invention, and FIG. 4 is an operation waveform diagram of each part thereof. , 5th
The figure shows a shift register 23 for explaining the operation of FIG.
FIG. 3 is a diagram showing a current scanning line signal input manually. 21: Input terminal, 22: Sampling clock input terminal, 23: Memory circuit for 1 scanning line, 24: Output terminal, 25
: Black signal counter, 28, 29, 31: 1, 4, 5 bit arc rising black signal detection circuit, 32: Circuit for counting and controlling the white signals before and after arc rising black signal detection, 33: Long black signal Detection registers 34, 35, 36: Circuits that add or suppress one bit of black signal.

Claims (1)

[Claims] 1 In a method that converts a binary signal waveform into three values and transmits it, the position of a black or white signal edge with a long pulse width of approximately three pixels width or more on the scanning line and the scanning line immediately before it are scanned. Detecting a waveform in which the position of an edge of an edge black or white signal whose pulse width is shorter than the long pulse width that is detected coincides with the position on the scanning line, and determining the pulse width of the edge of the edge black or white signal; When the width is approximately 1 pixel or less, a black or white signal having approximately 1/2 pixel width is added to the edge of the black or white signal having the long pulse width, and the determined width is approximately 1 pixel width or more and 3 pixels width. A waveform shaping method in ternary transmission, characterized in that a black or white signal of approximately 1/2 pixel width is removed from an edge of the black or white signal having the long pulse width when the width is in the following range.