JPS6365190B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a grayscale image compression method, and more particularly to a grayscale image compression method that can efficiently compress image data expressed in four gradations.

In order to store or transmit image data with a small number of bits, it is handled by facsimile etc. 2
There are various compression methods for digitized image data.

However, the data compression of multivalued grayscale images is limited to the extent that can be seen in the following papers.

Onoue et al., "Image data compression in a computer," Information Processing, August 1977 issue. Kaneko, "Coding method for seal imprint patterns," Institute of Electronics and Communication Engineers Image Engineering Study Group Material IE79-13 (May 31, 1979). , which performs Huffman encoding on the difference in gradation for each pixel, and is an effective method for images that have gradation changes over the entire screen, such as natural images, or for images that have a large number of gradations. However, especially for character patterns and seal patterns,
If the image has a small number of gradations, is close to a binary pattern, has gradation (shading) only in the character strokes and seal parts, and has little change in shading in the background, the compression ratio will not be very good.

Further, the above-mentioned method is for displaying a seal pattern in 4 gradations and 2 bits. In this method, when representing a stamp pattern with four levels of gradation from level 0 to level 3, pixels at intermediate levels such as 1 and 2 are located at transitional positions between the pixels at level 0, the lowest gray level, and the pixels at level 3, the highest. Since the 2-bit impression pattern representing the gray level of each pixel is mainly expressed, the 2-bit stamp pattern representing the gray level of each pixel is divided into the upper bit plane (“1” for levels 3 and 2, and “0” for levels 1 and 0) and the lower bit plane (level 3. Level 1 is "1"; level 2, 0 is "0"), the upper bit plane is encoded first, and the lower bit plane is encoded based on the information of the inner and outer points of the upper bit plane. It is something. However, compared to the above method, this method is more effective for things like seal imprint patterns, but for example, it is effective for parts of patterns that have a level of 1 other than the outer points, such as faded parts of a seal. It needs to be processed separately, and its configuration control becomes complicated.

Therefore, the present invention improves these problems and
In particular, we have developed a gradation image compression method that can reduce the amount of image data by compressing simple gradation images expressed in 4-gradation 2-bit images, such as images with many white areas such as characters, figures, and seal impressions. The purpose is to provide. To this end, in the grayscale image compression method according to the present invention, an image is scanned and converted into a grayscale signal, and in the grayscale image compression method that compresses this grayscale signal, the image scanning signal is converted into a four-tone grayscale signal. a transition code generating means for generating a transition code according to the changing state of the gray signal; A 4th floor It is characterized by compressing the image data of the tone.

Before describing the present invention in detail based on one embodiment, an outline of the present invention will be explained with reference to FIGS. 1 to 6.

As shown in FIG. 1A, the original image P is scanned with a scanning line _p0 , and as shown in FIG. 1B, pixels _a0 , _a1 , _a2 , _a3 ,
Regarding a ₄ , a ₅ ..., 4 gradations of gradation 0, 0, 1,
When obtaining grayscale image data of 2, 3, 3...
This gradation image data of gradations 0 to 3 is displayed in 2 bits, and is displayed in upper bits U and lower bits L as shown in FIG.

Then, when there is a change in the gray level of a pixel, a transition code as shown in FIG. 2A is added. That is, when the gray level changes from 0 to 1 as shown in FIG. → When changing from 2 to 2, a transition code "00" is added as shown in FIG.

In addition, when the upper bits U and lower bits L of the grayscale image data are different, as shown in FIG. When the upper bit U and lower bit L are the same as shown in Figure 3 A and 3', a run length code shown as a run code as shown in Figure 5 is used to display the run length. use. In other words, when displaying the run length with a direct code, the run length and the number of bits of the direct code match, and the direct code is displayed as the run length is 1, 2, 3, etc. are "1", "01", "001", etc., and the number of bits up to the "1" position indicates the run length. On the other hand, when the run length is displayed using a run code, as shown in Figure 5, when the run length is 1 to 4, it is indicated by 3 bits _A1 to _A3 , and when the run length is 5 to 8, it is indicated by 3 bits A1 to A3. Sometimes indicated by 4 bits _A1 to _A4 ,
When the run length is 9 to 16, it is indicated by 6 bits _A1 to _A6 . As is clear from this figure 5, A ₃ ,
A ₄ , A ₆ , A ₈ , A ₁₀ , A ₁₂ , A ₁₄ , A ₁₆ , A ₁₈ , A ₂₀
When the bit is 1, it indicates the end of the run code, and the bit is always 0 if it is not the end. Then, the specific length of the range is designated by the values of "1" and "0" of the cross marks of A ₁ , A ₂ , A ₅ . . . in the run code shown in FIG. For example, as shown in Figure 6, the run length is 1.
When A ₁ and A ₂ are 0 and are indicated as "001", when the run length is 2, A ₁ and A ₂ are 10 and indicated as "101", and when the run length is 3, it is indicated as "101". In this case, A ₁ and A ₂ are 01 and are indicated as "011".

That is, in the present invention, two lines of upper bit U and lower bit L of image data input in four gradations are simultaneously processed as one frame, and the previous bit in the scanning line direction is i and the current bit is j. If there is a difference, the transition code shown in FIG. 2C is assigned, and if they are the same, a run-length code is assigned. When the upper bit U and the lower bit L are the same, the run length code shown in Fig. 5 is used, and when they are not the same, the run length code is used.
Using the direct code shown in the figure.

When encoding, the first pixel of each scanning line usually has a gray scale of "0", starting from the run code. Therefore, encoded data is encoded in the following order: run code, transition code, run code, transition code.

Also, among the run-length codes, the gradations are 1 and 2.
Because the run lengths are halftones, the run lengths are often relatively short, so a code is directly assigned to them.The run lengths for gradations 0 and 3 are used for the white areas of the background, strokes of letters, seals, etc. Since there are many long parts, the compression effect can be improved by assigning run codes to these parts. Similarly, in the transition code, it is thought that the parts where the gray level difference is ±1 are used frequently, so such parts are displayed as 1 bit "1" to increase the data compression rate. There is.

An embodiment of the present invention will be described below with reference to FIGS. 7 and 8.

FIG. 7 shows a data compression section, and FIG. 8 shows a compressed data decompression section.

In the figure, 1 is a conversion circuit, 2 is an upper plane buffer, 3 is a lower plane buffer, 4 is an address controller, 5 is a transition detection circuit, 6 is a run code encoding circuit, 7 is a transition code encoding circuit,
8 is a direct code encoding circuit, 9 is a compression buffer,
10 is a transition code decoding circuit, 11 is a run direct code discrimination circuit, 12 is a run code decoding circuit, 13 is a direct code decoding circuit, and 14 is an OR circuit.

The conversion circuit 1 converts the original image P shown in FIG.
The image scanning signal obtained by scanning with p ₀ is
It converts into 2-bit gradation image data, in which the upper bit U of the 2-bit gradation image data is transmitted to the upper plane buffer 2, and the lower bit L is transmitted to the lower plane buffer 3. do. Note that an address control signal for inputting and outputting data to and from the upper plane buffer 2 and lower plane buffer 3 is obtained from the address control 4.

The transition detection circuit 5 is provided with a register for comparison, and compares the current grayscale image data with the previous one to determine whether they are the same or not, and if they are different, the transition code symbol of each of these data is determined. The signal is transmitted to the conversion circuit 7. Furthermore, this transition detection circuit 5 detects the output bits of the upper plane buffer 2 and the lower plane buffer 2.
It is determined whether the output bits of the buffer 3 are the same or not, and if they are the same, the output is transmitted to the run code encoding circuit 6, and if different, the output is transmitted directly to the code encoding circuit 8.

The data transition code encoding circuit 7 is provided with a ROM that outputs a transition code corresponding to the previous bit i and the current bit j, as shown in FIG. 2C.

Now, in FIG. 7, when an image scanning signal is input, the conversion circuit 1 converts it into 2-bit grayscale image data, transmits its upper bit U to the upper plane buffer 2, and transmits the lower bit L to the lower Transfer to plane buffer 3. In this way, the upper plane buffer 2 and the lower plane
The grayscale image data input to the buffer 3 is sequentially detected by a transition detection circuit 5 and sent to a run code encoding circuit 6 or a direct code encoding circuit 8, and if necessary, to either one of these and the transition code encoding circuit 7. communicated. At this time, the run code encoding circuit 6 and the direct code encoding circuit 8 have a built-in counter to count the number of consecutive pieces of the same grayscale image data and send out an output signal corresponding to the number of consecutive pieces of the same grayscale image data.
In this case, the run code encoding circuit 6
A run code as shown in the figure is output, and the direct code encoding circuit 8 outputs a direct code as shown in FIG. After the run code, transition code, direct code, etc. thus obtained are set in the compression buffer 9, the compressed data thus obtained is output.

Note that when encoding image scanning signals in this way, the first bit of each image scanning signal has a gray scale of 0.
Therefore, in this compressed data, the run code is output first, and then the transition code, run code, or direct code are output in this order.

When the compressed data transmitted from the data compression section shown in FIG. 7 in this way is input to the compressed data decompression section shown in FIG. is transmitted, the transition code decoding circuit 10 checks A ₃ , A ₄ , A ₆ . . . A ₂₀ in FIG.
Detects the position of and identifies the transition code that follows it. In response to this, the run direct code discrimination circuit 11 operates, and if the transition code identification result is for a run code, the run code decoding circuit 1
If it is a direct code, the direct code decoding circuit 13 is controlled to decode it; in any case, the OR circuit 14
The decoded grayscale image data will be output.

As described above, according to the present invention, it is possible to provide extremely efficient data compression for a grayscale image having four grayscale levels. In other words, with four gradation levels, generally levels 0 and 3 (the white part of the paper and the black part, such as printed characters) remain the same density for a long time, but the middle level 1 or 2 part remains the same for a long time. is a changing region, so it is unlikely that the same concentration will continue for a long time. Therefore, in the present invention, a run-length code is used for a gray level signal of 0 or 3 levels, which often have the same density, that is, the upper and lower bits are the same, and intermediate level parts are compressed by using a direct code. Efficiency can be improved. Therefore, it is highly compressed when processing images that are close to binary patterns, such as character patterns or seal impression patterns, where gradations exist only in the character strokes or seal parts, and where there are many background parts with little change in shading. We can provide high rates.

[Brief explanation of drawings]

Fig. 1 is an explanatory diagram of the original image scanning state and 4-gradation gray scale display state, Fig. 2 is an explanatory diagram of transition code, Fig. 3 is an explanatory diagram of when run code or direct code is used, and Fig. 4 is direct 5 and 6 are explanatory diagrams of the run code, FIG. 7 is a configuration diagram of an embodiment of the present invention, particularly showing its data compression section, and FIG. 8 is a compressed data decompression section. . In the figure, 1 is a conversion circuit, 2 is an upper plane buffer, 3 is a lower plane buffer, 4 is an address controller, 5 is a transition detection circuit, 6 is a run code encoding circuit, 7 is a transition code encoding circuit,
8 is a direct code encoding circuit, 9 is a compression buffer,
10 is a transition code decoding circuit, 11 is a run direct code discrimination circuit, 12 is a run code decoding circuit, 13 is a direct code decoding circuit, and 14 is an OR circuit.

Claims (1)

[Claims] 1. In a grayscale image compression method that scans an image, converts it into a grayscale signal, and compresses this grayscale signal, a conversion means that converts the image scanning signal into a four-tone grayscale signal, and a converting means that converts the image scanning signal into a four-tone grayscale signal, and a run-code encoding circuit that outputs a run-length code in accordance with the number of consecutive identical gray-scale signals in which the upper and lower bits are the same gray-scale signal; A direct code encoding circuit is provided which directly outputs a code run length code in accordance with the number of consecutive identical grayscale signals in which the bits and lower bits are grayscale signals of different bits.
A grayscale image compression method that is characterized by compressing four-tone image data.