JPH0470823B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION An object of the present invention is to preprocess sampled image information in a two-dimensional binary image signal such as a facsimile signal to increase the effect of data compression processing performed later. do.

Conventionally, band compression technology such as facsimile uses an image sensor to sample the original image to be transmitted at a predetermined resolution, as shown in Figure 1, and redundancy is applied to the extent that the sampled original image signal can be faithfully reproduced on the receiving side. Attention has been focused on eliminating signals that result in . However, in reality, there are flaws in the original painting,
Due to the effects of stains, variations in sensor sensitivity, and when the sensor input is close to the black and white judgment threshold,
The sampling signal already contains noise components. For example, when an original image as shown in FIG. 2a is sampled, a concavo-convex signal as shown in FIG. 2b is output. This situation frequently occurs with documents such as paperbacks, graphs, or typefaces that contain many linear components. Moreover, the isolated black signal as shown in FIG. 2c is considered to be noise due to some cause. If we try to compress such original image information as it is, even though it does not essentially contain large information as shown in Figure 2 b and c, the presence of such information will significantly reduce the compression efficiency. Cause. For example, even in the case of run-length encoding, if patterns such as those shown in FIG. 2b and c exist, the run lengths will be cut off and the compression effect will be controlled.

The present invention is intended to eliminate defects caused by isolated black pixels, and as shown in Figure 3, sampling noise contained in the original image signal is removed before data compression processing is performed, and the subsequent The effect of data compression processing can be greatly improved.

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings. In FIG. 4, the center pixel a ₁₁ of 3×3 sample original image signals a has a pattern of surrounding pixels a ₀₀ , a ₀₁ , a ₀₂ , a ₁₀ , a ₁₂ , a ₂₀ , a ₂₁ , a ₂₂ If it is possible to replace it with white or black, replace it with a ₁₁ as shown in Figure 4b. Specific examples of replaceable patterns are shown in FIG. 5 a, c, and e, and changed patterns are shown in b, d, and f, respectively.

However, regarding a, b, c, and d, the same applies to the rotating bodies.

Next, a specific circuit configuration of the processing method is shown in FIG. The original image signal S ₁ outputted together with the clock signal S ₂ is sequentially stored in shift registers (SR) 10, 13, 16 each having a capacity for one line, and at the same time is stored in flip-flops 11, 12, 1.
It is also shifted to 4, 15, 17, and 18. At the start of one page transmission, all shift registers and flip-flops 10 to 18 are reset by the PRST signal, and at the start of one line, all flip-flops 11, 12, 14, 15, 17, and 18 are reset.
By resetting the shift register and flip-flops 10 to 18, the fourth
Sample values corresponding to a ₀₀ to a ₂₂ in the figure are output. This situation is shown in FIGS. 7 and 8. That is, assuming that each pixel in each line is given a symbol as shown in FIG. 7, the i-1st row, the i-th row, the
The state in which the rows are stored in the shift registers 10, 13, and 16 is shown in FIG. 8a. At this time, the start signal LRST of the next line causes the flip-flop (F/
F) 11, 12, 14, 15, 17, and 18 have been reset. Next, the state at the time when the first pixel of the i+2th row is input is shown in FIG. 8b. In order i
When the contents of the +2nd line are input and the j-th pixel is input, the timing is as shown in FIG. 8c, and when the (j+1)th pixel is input, the timing is as shown in FIG. Therefore, eight pixel patterns of sample values _a00 to _a22 , excluding _a11 , output from each shift register and flip-flops 10 to 18 are input to a read-only memory (ROM) 19 having a capacity of 256 x 2 bits. as output
a Get the replaceability signal (EN) of ₁₁ and the value (DATA) if it is replaceable. The contents of the ROM 19 are shown in FIG. Set EN = 1 only when a ₁₁ can be replaced from the pattern of a _{00 , a 01} , a 02 , a ₁₀ , a ₁₂ , a ₂₀ , a 21 , a ₂₂ , and set white to the signal for a ₁₁ ' (DATA). Depending on the blackness,
0 and 1 are stored. Therefore, by opening the gate of the AND circuit 21 when EN=0 and closing the gate of the AND circuit 22 at the same time, S ₁ '=a ₁₁ , and when EN=1, the AND circuits 21 and 22 open and close conversely, so S ₁ ′=a ₁₁ ′. 24 to 26 are the bit synchronization clock _S2 for the original image signal _S1 , the page synchronization signal _S3 , and the image signal modified from the line synchronization signal.
This is a 2 _- line + 1-bit delay circuit (DELAY) for generating the respective synchronizing signals S ₂ ′, S ₃ ′, and S ₄ ′ for S 1 ′.

As described above, in this embodiment, when the original image is scanned to obtain a binary signal, the black and white signal may be reversed due to noise, or there may be dust or other foreign matter in the area that should be white or black. This is to prevent the binary signal sequence from becoming uneven when it should be flat, and from losing its effectiveness when the binary signal example is later encoded into a run-length code and compressed. be. In other words, according to this embodiment, while noise caused by dust and scratches is removed, the white signal surrounded by the black signal of a complex kanji character such as "column" is not replaced, and the image is compressed as a clear image. Can be processed. In addition, in this embodiment, when burrs appear on the edges of characters, etc. in the original image itself, compression processing is performed by removing the burrs and reproducing an image that is different from the original image but is similar to the original image. This has the effect of improving efficiency.

As explained above, the present invention provides {(2n+1)2-1} pixel patterns (n=1,
2, 3, ...) are all white information, replace the binary information of the certain pixel with white information, and if the binary information of the surrounding pixel patterns are all black information, It stores the information of a certain pixel, and when scanning the original image to obtain a binary image, it improves the image quality by removing isolated black points caused by noise etc.
During later compression processing such as run-length encoding,
It has the excellent effect of preventing the compression effect from being impaired.

[Brief explanation of the drawing]

Fig. 1 is a system diagram explaining the conventional facsimile signal band compression transmission technology, Fig. 2 a to c are explanatory diagrams of the original image signal, and Fig. 3 is band compression transmission using the concept of preprocessing of the present invention. A system diagram explaining the technology; FIGS. 4a and 4b are diagrams explaining the principle of a signal processing method according to an embodiment of the present invention; FIGS. 5a to 5f are diagrams showing a specific example of pixel conversion; FIG. 7 and 8 are diagrams illustrating the operation of FIG. 6, and FIG. 9 is a diagram illustrating the operation of one component of FIG. 6. . 10, 13, 16...shift register, 11,
12, 14, 15, 17, 18...flip-flop, 19...read-only memory, 20...inhibition gate, 21, 22...and gate, 23,
27... OR gate, 28, 29... Noah gate, 24, 25, 26... delay circuit.

Claims (1)

[Claims] 1. As preprocessing for compression encoding of the binary image obtained from the original to be read,
+1) ² -1} pixel patterns (n=1, 2,
3, ...), when all the binary information of the original to be read is white information, the binary information of the center pixel is replaced with the white information, and the binary information of the surrounding pixel pattern is all black information. If so, the image signal processing method is characterized in that binary information of the central pixel is stored.