JP2635614B2 - Image decoding thinning method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for decoding coded image information into an original image, and more particularly, to a method for receiving an image code in a high-speed facsimile, an optical disk and the like. The present invention relates to an image decoding thinning method suitable for decoding an image code stored on a magnetic disk or the like, arbitrarily reducing the pixel size, and displaying the image code on a CRT or printing on a printer.

[Conventional technology]

There are various techniques for decoding encoded image information,
As commonly used, the MRs specified in the International Telegraph and Telephone Consultative Committee (CCITT) Recommendations T4 and T6
(Modified Read) sign.

The MR code is a two-dimensional compression code for binary image data. An example of an encoding device for the MR code is disclosed in
Since it is described in detail in Japanese Patent No. 6274, the essential points of the MR code and the decoding process will be described here to the extent necessary for understanding the decoding process.

FIGS. 2A and 2B summarize the rules of the MR code, and the meanings of the symbols in the figure are as follows. Note that the encoding line is a scanning line that is currently undergoing encoding processing, and the reference line is a processed scanning line immediately before the encoding line.

a ₀ ... Reference or starting point change pixel on the coding line. At the beginning of the coding line, a ₀ is placed on a virtual auto-change pixel immediately before the first pixel of the scan line (hereinafter “line”).

a ₁ ... The first changed pixel to the right of a ₀ on the coding line.

a ₂ ... The first changed pixel to the right of a ₁ on the coding line.

b ₁ ...... reference of the change pixels on the line, at right of a _0, the first changing pixel with a color opposite to a _0.

b ₂ … The first changed pixel to the right of b ₁ on the reference line.

In FIG. 2B, “a ₁ b ₁ = 0” and “a ₁ b ₁
= 1 "expressions such represents a distance of a ₁ b ₁ are each 0 image, one pixel, or the like.

Also, M (a ₀ a ₁ ) and M (a ₁ a ₂ ) are each from a ₀
a ₁ to and a ₁ number of pixels up to a ₂ shall apply in MH (ModifiedHuffman) code corresponding to (run length) is defined in CCITT recommendation 4 also the coding rule.

 MR codes are roughly classified into the following three modes.

Pass mode (P mode), shall apply in the case where b ₂ is present on the left side of a _1, the sign of this time, regardless of the distance uniquely determined. (Briefly, even if thickness change point on the reference line, is on the coding line sometimes change point does not come.) Horizontal mode (H mode), b ₂ Do the same position as the a ₁ or located on the right, moreover, it shall apply in the case the distance of a ₁ and b ₁ are four or more pixels, the sign of this time represents the respective distances a ₀ and a ₁ and a ₁ and a _2. (Simply speaking, contrary to pass mode,
This is a case where there is no change point on the reference line and there is a change point on the coding line. ) Vertical mode (V mode), the cases other than the two modes, i.e., b ₂ is filed at the same position or the right side thereof and a _1, moreover, in case the distance of a ₁ and b ₁ is less than 3 pixels Yes,
Code at this time represents the relative position of a ₁ and b _1. (Simply speaking, this is a case where a change point on the coding line comes to the same position as or near the change point on the reference line.) When one coding process is completed, a reference point (starting point) a
_0, if been made in the path mode, the position of b _2, the position of a ₂ if there was filed in the horizontal mode, the position of a ₁ if there was filed in the vertical mode and transferred respectively (also, which Accompanying
The positions of a ₁ , a ₂ , b ₁ , and b ₂ are also moved to the next point), and the next encoding process is performed. As described above, the position where a ₀ is shifted is different.
This is because the number of target change points differs depending on the mode.

FIG. 2 (c) shows this encoding procedure in a flowchart. First, following the start of the operation, place the origin changing point a ₀ immediately before the first pixel (2-2) changing point a _1, b
_1, b ₂ detects (2-3), b ₂ is to determine whether the left a ₁ (2-4). If b ₂ is to the left of a ₁ , after performing pass mode encoding (2-5), a ₀ is moved directly below b ₂ (2-6),
Proceed to the next encoding. If b ₂ is not in the left of a ₁ is, | a ₁ b ₁ |
Determine whether ≦ 3 (2-7), if 3 or less vertical mode code from the row of connexion (2-8), were transferred a ₀ to a ₁ proceeds to the next coding. If | a ₁ b ₁ |> 3, a ₂ is detected (2-10), and horizontal mode encoding is performed using this (2-11). Then, a ₀ is moved to a ₂ . . The above operation is repeated until the end of the operation line.

FIG. 3 shows a basic flowchart of an MR code decoding procedure.

In the figure, the reference line change point search mode is
When detecting a pixel change point on the reference line while outputting dot data after detecting the V code or P code, whether the pixel change point to be searched is a white → black change point
This is a mode indicating whether it is a black-to-white transition point. An EOL (End of Line) code is a code added immediately after a code corresponding to one line of image data and indicating the end of a line.

In the figure, initially set an initial value a ₀ is (3-2), the reference line changing point is initialized to one of white → black or black → white (3-3), then the encoding line The code shifts to code analysis for analyzing any of the codes shown in FIG.
-4). As a result, when the V code is detected, (3
-5) Search for a change point on the reference line and continue to output the dot of the coding line until the change point comes near (3-6). For example, until just below the change point of the reference line if V ₀ mode, until just after the change point of the reference line if VR mode, immediately before the change point of the reference line if VL mode,
Outputs the dot. Next, the change point search mode of the reference line is inverted (black → white when previously white → black) and the same search is performed (3-7), and the dot of the coding line is inverted and output (3-7). 3-8). In the case of P code detection, (3
-9), search for a change point on the reference line and continue to output the dot of the coding line up to this change point (3-
Ten). Next, the change point search mode of the reference line is inverted (3-11) to search for a change point on the reference line, and the next change is made without inverting the dot of the coding line (it does not change in the P mode). The coded line dot is output up to the point (3-12), and the change point search mode of the reference line is again inverted (3-13). As described above, in the P mode, the dot is continuously output until two change points of the reference line are detected. When the H mode is detected, (3-1)
4) Regardless of the reference line, the length of white or black of the coding line is analyzed and dots corresponding to the length are output. That is, in the H mode, after setting the MH mode for analyzing the MH code representing each length of black and white, (3-1-1)
5) First, the length of one of black and white (for example, white) is analyzed (3-16), and dots corresponding to the length are output (3-1).
7). Next, the black and white of the output dot is inverted (for example, black).
(3-18), outputs a dot corresponding to the length (3-20),
Finally, it is inverted again (3-21) to end. Thus, V,
The procedure of making a decision by inputting any one of the codes P and H, and performing dot output when the decision is completed is repeated until the end of one line.

In a facsimile apparatus using a telephone line or a low-speed dedicated line, the flowchart shown in FIG. 3 is further subdivided, and its steps are sequentially programmed to realize the steps.

In order to reduce such a long processing time required for sequential processing by software, an apparatus combining a program-controlled processing circuit and a dedicated processing circuit for determining a pixel change point position has been disclosed in Japanese Patent Application Laid-Open No. H10-139,197. This is proposed in FIG. 4 of Japanese Patent Publication No. 58-194465.

In this device, the program control processing circuit cuts out the data of the reference line by a predetermined length and writes it into the shift register in the dedicated processing circuit. When the program control processing circuit detects the presence of a pixel change point on the reference line, the reference line data in the shift register is shifted out one bit at a time, and the presence or absence of a pixel change is inspected at its output end. The shift amount up to the detection is sent to the program control processor as position data of the pixel change point.

The program control processor uses the pixel change point position data and the output of the decoding circuit to generate dot data for the output line.

As described above, the conventional apparatus has a problem in that the processing speed is slow because the detection and decoding of the change point are all performed by software.

[Problems to be solved by the invention]

By the way, in the electronic filling market where codes stored on optical disks, magnetic disks, etc. are decoded into images, there are various types of target image sizes, such as A and B plates. Is generally displayed on one fixed display device (CRT).

Also, when decoding and printing an image code stored on an optical disk or a magnetic disk, it is common to use a device having a limited image size when printing.

To meet such market needs, conventionally, there is no cotton density conversion function at the time of decoding or only a function having a limited function (for example, a conversion function only for the number of lines in the vertical direction). Even in the above publication, no consideration is given to such a conversion function at the time of decoding.

For this reason, after performing the decoding process, another image reduction device is used to convert the size of the image. In such a case, in addition to the decoding processing time, the image reduction processing time is sequentially increased.

SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide an image decoding method which solves the above-mentioned problems of the prior art and performs image size conversion at the time of decoding a decoded image more efficiently and at high speed. is there.

[Means for solving the problem]

In order to achieve the above object, the present invention provides a decoding line management unit that manages the number of decoded image lines, a decoding line reduction unit that decodes encoded image information, and simultaneously performs a thinning reduction according to a thinning ratio specified by the device. And a pixel change point detector that shifts n bits (n is an integer of 2 or more) until the corresponding pixel change point appears. The image decoding unit is provided with decoding thinning-out means for thinning out output dot data in accordance with the thinning-out magnification, and the decoding line management unit is provided with an effective line (a line which is not thinned out) and an invalid line (in a line unit). Means for determining the line to be thinned) is provided. In the valid line, the thinning-out by the decoding thinning-out means is performed, but in the invalid line, the thinning-out operation by the decoding thinning-out means is not performed, and the n-bit shift processing (n is an integer of 2 or more) of the decoded image is performed. To be configured.

As described above, in the present invention, the detection and decoding of the change point are all performed by hardware.

In the present invention, for example, the present invention is applied to MR codes and the like (MH, G3MR, G4MR codes, and the like) defined in the CCITT recommendation draft.

[Action]

Based on the above configuration, when a decoding thinning ratio is given,
It is determined whether the current line to be decoded is a valid line (a line that is not thinned out) or an invalid line (a line that is thinned out).

When the valid line is determined, the decoding thinning circuit
Start by giving the decoding thinning ratio. The decoding thinning-out circuit incorporates, for example, a decoding thinning-out ROM, and after code analysis, thins out the dot output clock according to the magnification and stops when processing for one line is completed.

On the other hand, when an invalid line is determined, the decoding is started without setting the decoding thinning-out magnification (for example, setting all “0”).
When the decimation factor is not set (in the case of an invalid line), the decoding decimation circuit and the pixel change point detection unit
Until the corresponding pixel change point is detected, the output dot data is discarded while performing the n-bit shift process. When the corresponding pixel change point is detected, the process returns to the 1-bit process, and when it is completed, the n-bit process is performed again. Then, the n-bit shift process and the one-bit shift process are repeated until the process for one line is completed.

Here, the n-bit shift processing means that in a place where the same dot pattern is continuous, rather than processing one bit by one clock, n bits (for example, 8 bits) are processed in a lump and processed once. This can improve processing efficiency, and refers to such processing.

In the present invention, the output dot data is decimated for the valid line, but the output dot data is not decimated for the invalid line, so that the processing speed can be improved. By performing the n-bit shift processing on the image, the processing efficiency can be further improved.

Such an n-bit shift process is not performed on an effective line.

Note that the effective line is thinned out as output data, but it is necessary to use it as reference line data for the next line. Therefore, the decoded data of the effective line is written to the line buffer without thinning out.

〔Example〕

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention.

The decoding line management unit 1 manages the number of lines of the decoding image, controls the image bus 1-1, and controls the image combining unit 2,
It is a microprocessor with a built-in microprogram that controls the pixel change point detection unit 3. The decoding line management unit 1 receives the number of image lines, the length of one line data, and the decoding thinning rate 1-2 via the image bus 1-1,
Line data length 1-7, thinning magnification 1-
Give 8. Next, an image decoding unit 2 and an image change inspection unit 3
Is supplied with an activation signal 1-9. Then, in response to an image bus access request signal 1-5 from the image decoding unit 2, an image bus control signal 1-2 is output.
The MR code 1-3 is supplied to the image decoding unit 2 or the decoded image 1
-3 to the image bus 1-1.

Then, the one-line end signal 1-6 is sent from the image decoding unit 2.
Upon receiving the information, it is determined whether or not the last line has been reached, based on the specified number of image lines and the decoding thinning ratio, and whether the next line to be decoded is an output valid line (a line for outputting a decoded image) or not. It is determined whether the line is an output invalid line (a line that does not output a decoded image). If the line is an output valid line, the thinning-out factor 1-8 is output again and the activation signal 1-9 is output. On the other hand, in the case of the output invalid line, only the activation signal 1-9 is output without outputting the thinning-out magnification 1-8.

When the output line is invalid, the decoding line management unit 1 supplies the image decoding unit 2 with the MR code 1-3.
The decoded image 1-3 is not supplied to the image bus 1-1.

The image decoding unit 2 sends a line data length 1-7 and a thinning ratio 1-
In the case of an output invalid line, only the line data length 1-7 is received.

Then, upon receiving the activation signal 1-9 from the decoding line management unit 1, it outputs an image bus access request signal 1-5 to the decoding line management unit 1, receives the MR decoding 1-3 from the image bus 1-1, The MR code 1-3 is analyzed and V (vertical)
P (pass) / H (horizontal) determination is performed. After the determination, a change point search mode 1-A and an output clock 1-B are generated and output to the pixel change point detector 3. Also, if the judgment result is V
Alternatively, in the case of the P mode, the VP mode signal 1
-C is also output. Then, the pixel change point detection unit 3
The decoded data 1-3 and 1-4 continue to be output until the reference line change point detection signal 1-F is output.

Next, when the reference line change point detection signal 1-F is output from the pixel change point detection section 3, the output clock 1-B is stopped, the change point search mode 1-A is inverted, and the next MR code 1 is output. Move on to analysis of -3.

When the analysis result of the MR code 1-3 is in the H mode, the output clock 1-B is generated, and the decoded data 1-3 and 1-4 are continuously output for the length of the corresponding run length.

When the decoded data corresponding to the corresponding run length is output, the output clock 1-B is stopped, and the next MR code 1-B is stopped.
Move to the analysis of 3.

The pixel change point detecting section 3 has a built-in line buffer. Then, a change point search mode 1-
A and VP mode signal 1-C and output clock 1
-B, while reading the data of the built-in line buffer as reference line data,
Is written to the line buffer so as to be the reference line data of the next line. When a change in pixel color is detected in the reference line data, the reference line change point detection signal 1-F
Is output to the image decoding unit 2.

When the decimation factor 1-8 is set by the decoding line management unit 1 (that is, in the case of an effective line), the image decoding unit 2
Inside, the decoded images 1-3 are thinned out according to the magnification and output. However, since the decoded data 1-4 becomes reference line data of the next line, it is output to the pixel change point detection unit 3 without thinning. Then, when the decoded data 1-4 for a prescribed length is output by the line data length 1-7, the one-line end signal 1-6 is output to the decoded line management unit, and the operation is stopped.

If the decimation factor 1-8 has not been set by the decoding line management unit 1 (ie, in the case of an invalid line), in the case of the H mode, the image combining unit 2 outputs the decoded data 1-4 corresponding to the corresponding run length. Until the output, or in the case of the V and P modes, the 8-bit shift enable signal 1-D is sent to the pixel change point detector 3 until the line change point detection signal 1-G is output from the pixel change point detector 3. While outputting the data, the 8-bit shift output of the decoded image 1-3 and the decoded data 1-4 is continued.

While the 8-bit shift output is being performed, the decoded data 1-4 and the decoded image 1 are output until the corresponding run length is detected in the H mode and the line buffer change point detection signal 1-G is detected in the V and P modes. After the output of -3, the 8-bit shift enable signal 1-D is turned "OFF" and the decoded data 1-
4 and the decoded image 1-3 are switched to 1-bit shift output.

As described above, the feature of the present invention resides in that the image size conversion is performed while decoding the MR code into an image. At the time of image size conversion, the decoding process of the image line to be thinned out is to shift the output dot data by 8 bits.

FIG. 5 is a block diagram showing the inside of the image decoding unit 2.

The image decoding unit 2 includes an MR code analysis ROM 5-1 and an MH code decoding unit 5
-3, a clock control unit 5-4, a decoding thinning ROM 5-5, an output buffer 5-D, and the like.

First, the line data length 1-7 is determined by the clock control unit 5-
4, the decoding thinning-out magnification 1-8 is given to the decoding thinning-out ROM 5-5. Next, the start signal 1-9 is sent to the clock control unit 5
-4, the clock control unit 5-4 generates an input clock 5-6. The MR code analysis ROM 5-1 receives the MR code 1-reference in synchronization with the input clock 5-6, and
The MR codes 1-3 are sequentially analyzed according to the sequence address data 5-A.
・ H mode signal 5 when P mode signal 1-C is in H mode
After outputting -7, the corresponding run-length data 5-8
Is output.

The clock control unit 5-4 reads the data from the MR code analysis ROM 5-1.
Upon receiving the VP mode 1-C or the H mode 5-7,
Immediately stop the input clock 5-6 and output clock 1-
Generate B.

When the analysis result of MR code 1-3 is H mode, MH
The decoding unit 5-3 receives the H mode signal 5-7 and the MH run length data 5-8, performs a run length count operation in synchronization with the output clock 1-B, and counts the corresponding run length. Then, it outputs the count carry signal 5-B to the clock control section 5-4. Upon receiving the count carry signal 5-B, the clock control unit 5-4 stops the output clock 1-B and stops the input clock 5-6.
Is generated, and the process proceeds to the analysis of the next MR code 1-3.

When the analysis result of the MR code 1-3 is V or P mode, as described above, the clock control unit 5-4 receives the VP mode signal 1-C from the MR code analysis ROM 5-1. Stop input clock 5-6 and generate output clock B. When receiving the reference line change point detection signal 1-F from the pixel change point detection unit 3 (FIG. 1), the output clock 1-B is stopped, the input clock is generated, and the next MR code 1 is generated.
Move on to analysis of -3.

The toggle circuit 5-2 outputs V.V. from the MR code analysis ROM 5-1.
Upon receiving the P mode 1-C and the H mode 5-7, the change point search mode 1-A is inverted and supplied to the pixel change point detection unit 3 (FIG. 1).

When the decoding thinning rate 1-8 is set, the decoding thinning ROM 5-5 supplies the output clock thinning signal 5-E to the output buffer 5-D in synchronization with the output clock 1-B, and shifts the data by 8 bits. The prohibition signal 5-9 is output to the clock control unit 5-4.

The dockle circuit 5-C receives V.V. from the MR code analysis ROM 5-1.
In response to the PH mode 1-C-5-7, the decoded data 1 is inverted by the last input clock 5-6 after the mode is determined.
Output as -4.

The output buffer 5-D synchronizes with the output clock 1-B from the clock control unit 5-4 and the output clock culling signal 5-E from the decoding culling ROM 5-5, and outputs the decoded data 1-B.
4 is fetched while being decimated and output to the image bus 1-1 as a decoded image 1-3. In this case, the thinning rate of the decoded image 1-3 is given by the decoding thinning ROM 5-5. still,
The clock control section 5-4 has an input clock 5-
6 and a counter for counting the output clock 1-B, each of which has a width (32
Bit), the image bus access request signal 1
Outputs -5.

When the output clock 1-B has been counted for the line data length 1-7, a line end report 1-6 is output, and the input clock 5-6 and the output clock 1-B are stopped.

Next, the 8-bit shift processing will be described. Since the 8-bit shift process is a process of an output dot data sequence,
It is always performed in synchronization with the output clock 1-B.

First, since the decoding decimation factor 1-8 from the decoding line management unit 1 (FIG. 1) is not set, the decoding decimation ROM 5-
5 does not operate, and does not output the 8-bit shift inhibit signal 5-9 and the output clock thinning signal 5-E. As a result, the 8-bit shift enable signal 1-D is output from the clock control unit 5-4, and the run length counter in the MH code decoding unit 5-3 and the output clock counter in the clock control unit 5-4 are output. , One clock of output clock 1-B is counted up by "8", and output buffer 5
The decoded data 1-4 in -D is also shifted up by "8". In the H mode output, the line buffer change point detection signal 1-G is output from the pixel change point detection unit 3 (first time) until the count carry 5-B is output, and in the VP mode, the pixel change point detection unit 3 (first time). Up to 8 bit shift processing.

 FIG. 4 shows the internal configuration of the pixel change point detection unit 3.

After receiving the activation signal 1-9, the line buffer 4-1 reads the reference line data 4-6 in synchronization with the output clock 1-B, and uses the decoded data 1-4 as the reference line of the next line. Write.

The reference line change point detection section 4-3 inputs the reference line data 4-6 in synchronization with the output clock 1-B, while the VP mode 1-C and the change point search mode 1-.
The corresponding pixel change points are sequentially verified by A. And
When the change point is detected, the reference line change point detection signal 1-F
Is output to the image decoding unit 2 (FIG. 1).

In the 8-bit shift operation, when the 8-bit shift enable signal 1-D is input to the counter 4-2, the counter 4-2 outputs "8" for one clock of the output clock 1-B.
Counted up by one. As a result, reading and writing of the line buffer 4-1 are processed in units of 8 bits.

The EXOR gate 4-5 constantly monitors the 8-bit data 4-4 read from the line buffer 4-1 for a pixel dye change point. If a change point is found, the line buffer change point detection signal 1- 6 to the image decoding unit 2.

〔effect〕

As described in detail above, in the present invention, in the decoding thinning method of the image, the thinning of the decoded image is performed simultaneously with the decoding of the image, the thinning of the image is reduced in the effective line, and the thinning of the image is performed in the invalid line. Is not performed, and instead n-bit processing (n is an integer of 2 or more) of the decoded image is performed, so that the size of the decoded image can be converted simultaneously with the decoding of the encoded image, and the invalid line ( In the thinning line, the output dot data is processed by n bits (8 bits), so that an excellent effect such as faster image size conversion can be achieved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the entire configuration of an embodiment of the present invention, FIGS. 2 (a) and 2 (b) are summaries of MR code regulations, and FIG. 2 (c) is a flowchart of MR code encoding. Fig. 3 shows MR
FIG. 4 is a block diagram showing the internal configuration of a pixel change point detecting section, and FIG. 5 is a block diagram showing the internal configuration of an image decoding section. 1. Decoding line management unit, 2 .... image decoding unit, 3 .... pixel change point detection unit, 4-1 ... line buffer, 4-2 ...
... Counter, 4-3 ... Reference line change point detector, 5-
1 ... MR code analysis ROM, 5-2,5-C ... Toggle circuit, 5
-3: MH code decoder, 5-4: Clock controller, 5
-5 ... Decimation thinning ROM, 5-D ... Output buffer

Claims (1)

(57) [Claims] 1. A method for decoding coded (MH, G3MR, G4MR code or the like) image information, comprising: means for determining a valid line or an invalid line on a line-by-line basis according to a specified thinning-out magnification; According to the thinning rate, the decoding of the decoded image dots is performed simultaneously with the decoding of the active line,
An invalid line is provided with image decoding thinning-out means for not thinning out decoded image dots, and processing means for performing n-bit processing (n is an integer of 2 or more) until a corresponding pixel change point appears on an invalid line. Image decoding thinning method.