JPH0548658B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image data compression method for compressing image data by two-dimensional encoding and performing high-speed facsimile transmission.

[Conventional technology]

Currently, the Modified READ method (hereinafter referred to as the "MR method") is used as a compression method for image data in the digital network facsimile method.
It is called. ) has been adopted as one of the standard methods. The MR method targets two scanning lines, the encoding scanning line and the reference scanning line immediately before it, and detects the points where a white pixel on the encoding scanning line changes from a white pixel to a black pixel, and from a black pixel to a white pixel. (hereinafter referred to as the "change point") is determined by the relative distance (hereinafter referred to as "shift") to the change point of the same color tone on the reference scanning line, or to the immediately preceding change point of the opposite color tone on the encoded scan line. This is a method of encoding using relative distance (hereinafter referred to as "run length").

There are three changing point modes in the MR method: pass, vertical, and horizontal. For example, as shown in Figure 9a, from the black pixel at changing point b ₁ on the reference scanning line to the next changing point b The black pixel of P up to the white pixel of ₂ is
The continuous white pixels from the white pixel at the change point a ₀ on the encoded scan line to the white pixel a ₀ ′ immediately before the black pixel at the next change point a ₁ are continuous with the black pixels on the encoded scan line. Since it does not work, set this as pass mode, and
As shown in , the black pixel between the black pixel at change point b ₁ on the reference scan line and the white pixel at the next change point b ₂ , and the black pixel at change point a ₁ on the encoded scan line and the next change The case where the black pixel between the white pixel at point _a2 and the black pixel are continuous is considered to be the vertical mode. However, when there is a shift of 4 pixels or more in the horizontal direction [V( _{1 1} )≧4], the horizontal mode is set.

The changing points b ₁ , b ₂ on the reference scanning line and the changing points a ₀ , on the encoding scanning line correspond to each of these modes.
Depending on the relationship between a ₁ and a ₂ run lengths and deviations,
As shown in FIG. 10, codes are assigned.

For horizontal mode, the Modified Huffman method (hereinafter referred to as “MH
``Method''. ), a run length is expressed by a combination of a termination code that encodes each run length from 0 to 63, and a make-up code that encodes each run length that is an integer multiple of 64. A code is specified.

In the horizontal mode MR code in Figure 10,
MH indicates an MH code encoded by a combination of a terminating code and a make-up code.
In addition, in normal image data, 50% of the changing points can be expressed in the vertical mode V (0), and 30% can be expressed in the vertical mode V _R (1), V _L (1) For these, the code length is 1.
Alternatively, a short code of 3 is assigned to obtain a high compression rate.

[Problem that the invention seeks to solve]

Image data generally has not only horizontal correlations, but also
Vertical correlation is also often large. That is,
If you look at the boundaries of the images, you can see that they are connected not only horizontally but also vertically. In the above-mentioned MR method, since encoding is performed using only two scanning lines, the reference scanning line and the encoding scanning line, there is no method that fully utilizes the vertical correlation.

For example, as shown in Figure 11, the width is 500 x the height is 300.
When image data of a rectangular figure is encoded using the MR method, the black run length in the horizontal direction is BRL500, so the terminating code of BRL52 and
Together with the BRL448 make-up code, the total is 24 bits, resulting in a compression rate of 24/500. However, all vertical boundaries are represented by V(0) in vertical mode and are composed of 1 bit, but the boundaries on both sides of the rectangle are encoded in bit-by-bit format. It had the disadvantage of low compression efficiency.

The present invention aims to improve the above-mentioned conventional drawbacks and improve the compression rate by encoding also in the vertical direction.

[Means for solving problems]

The image data compression method of the present invention targets a reference scanning line and an encoded scanning line, and includes a detection means for detecting a start mode in which two change points exist only on the encoded scanning line; comprising a detection means for detecting a continuation mode in which a change point exists on the line and the encoded scanning line, and a detection means for detecting a termination mode in which a change point exists only on the reference scan line, Each boundary line that connects in the sub-scanning direction from the two change points of the start mode is traced in a continuous mode and encoded up to the end mode.

[Effect]

The length connected in the vertical direction is tracked by the continuation mode between the start mode and the end mode, the length continued in the vertical direction is encoded, and the data compression rate is improved by using the correlation in the vertical direction. It is something.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention.
1 is an image data memory having a capacity to store image data for one page; 2 is a start mode detection unit; 3;
7 and 8 are counters, 4 and 5 are registers, 6 is a continuation mode detector, 9 is an end mode detector, 10 is a multiplexer, 11 is a code conversion memory that outputs a code and its code length information, and 12 is a shift register , 13 is a code length counter. The image data is added to the image data memory 1, one page worth is temporarily stored, and from the image data memory 1, reference scanning lines and codes are sent to the start mode detection section 2, the continuation mode detection section 6, and the end mode detection section 9, respectively. The image data on two scan lines are added.

As shown in Fig. 2a, the start mode is the case where there is a change point from a white pixel to a black pixel or from a black pixel to a white pixel only on the encoded scanning line, and the change point a ₀ and the change point a ₁ are defined as the start mode. between and this change point a ₁ and the next change point a ₂
The run length between is the target of encoding. Therefore, unlike the MR method, the horizontal mode is not set when the deviation a ₁ b ₁ is 4 or more, but there is no connection between the black pixels on the reference scanning line and the black pixels on the encoding scanning line. This will show that.

As shown in Fig. 2b, the continuous mode is defined as a state in which there is a change point on the reference scanning line and the encoding scanning line, and the boundary lines are connected vertically, and the object of encoding is the shift a ₁ b ₁ , and this deviation a ₁ b ₁ is always encoded even if it is 4 or more.

Further, as shown in FIG. 2c, the end mode is defined as the case where the change points b ₁ and b ₂ exist only on the reference scanning line, and this is the same as the pass mode of the MR method.

FIG. 3 shows the assignment of codes corresponding to each of the above-mentioned modes, and the encoding formats S, C, and E indicate a start mode code, a continuation mode code, and an end mode code. Also, (V ₀ ×n) means encoding the number n of consecutive modes in the vertical direction. V ₀ corresponds to V(0) in the vertical mode of the MR system. Also, C ₀ is MR shown in Figure 10.
Corresponding to the sign V(0), C _R (1) and C _R (2) are
Corresponds to V _R (1) and V _R (2) of MR code, and C _L
(1) and C _L (2) correspond to V _L (1) and V _L (2).

The encoding procedure detects the start mode of the change point between the reference scan line and the encoded scan line, and after the start mode code S, two run lengths, e.g., a ₀ a in a of FIG. ₁ , encode the run length between a ₁ a ₂ . Note that there are the following two methods for selecting the reference change point _a0 . (1) The second change point _a2 of the immediately preceding start mode is set as the reference change point. (2) In addition to the number of change points in the continuation mode that have passed from the second change point a ₂ of the previous start mode to the first change point a ₁ of the current start mode, the continuation of the continuation mode immediately before the first change point a ₁ The mode change point is defined as the reference change point. However, if there is no change point in the continuous mode in the middle, it is the same as (1).
Method (2) described above allows for a shorter run length than method (1), resulting in a higher compression ratio, but since the number of change points in the continuous mode is added, from the perspective of applying MR codes ( Method 1) is preferred.

Once the start mode is detected, the first change point
Trace the boundaries of a ₁ vertically and encode them sequentially in continuous mode. That is, (i) if the deviation a ₁ b ₁ = 0, convert it to a code representing a number that continues vertically, and (ii) if the deviation a ₁ b ₁ = 1,
2, convert to the mode codes of C(1) and C(2), and (iii) if the deviation a ₁ b ₁ ≧ 3, convert to the continuous mode code and the code representing the deviation a ₁ b ₁ . . Then, when the end mode is detected, the end mode code is finally changed to E, and the boundary line tracing of the change point _a1 is ended.

Next, the boundary line tracing of the second change point a ₂ in the start mode is encoded in the continuous mode in the same manner as in the case of the first change point a ₁ described above. However, in this case, the end mode code E is not added at the end. This is the first change point ₁
Since the boundary line tracing information of the second change point _a2 is included between the end mode code E indicating the end of boundary line tracing and the next start mode code, there is no need to add the end mode code E. , in order to know the end of the continuation mode.

Next, the start mode is detected and the above-described procedure is repeated to perform encoding.

If there is no other starting mode change point in the encoded scanning line, a line end signal (for example, the EOL code in the MR method or the above-mentioned end mode code E) is added to end the processing of that line. , move on to processing of the next line.

In order to perform the encoding according to the above-mentioned procedure, the start mode detection unit 2 in FIG. Identify whether the image data is in the state shown in Figure 2 a, and if it is in the start mode,
Counter 3 counts the run length between a ₀ a ₁ and a ₁ a ₂ , latches the absolute address of the first change point a ₁ from white pixel to black pixel in register 4, and changes from black pixel to white pixel. The absolute address of the second change point _a2 is latched into register 5.

The continuation mode detection unit 6 reads image data vertically from the image data memory 1 according to the absolute addresses of the change points a ₁ and a ₂ set in the registers 4 and 5, and reads the image data at the change points a ₁ and a ₂ in the vertical direction. The connected image contours are tracked, and the counter 7 counts the deviation, that is, the deviation between a ₁ b ₁ in b in Fig. 2, and the counter 8 counts the number of consecutive images in the vertical direction with zero deviation, that is, the deviation in Fig. 2. b
Count the number of times the state a ₂ b ₂ continues in .

In addition, the end mode detection section 9 identifies whether or not the image data is in the state shown in c in FIG. 2 in the continuation mode detection operation.

The multiplexer 10 switches the detection signals of the counters 3, 7, 8 and the end mode detector 9 according to the code procedure and adds them to the code conversion memory 11.
The code conversion memory 11 is composed of a read-only memory, converts the code into the code shown in FIG. 3, sets the code data in the shift register 12, sets the code length in the counter 13, and calculates the number of bits according to the code length. is sent out from the shift register 12 as compressed data.

FIG. 4 shows an example of compression-encoded data, in which a run length RL of _{0 1} and a run length RL of _{1 2} are added next to the disclosure mode code S to become start mode data. The change point a ₁ ,
b ₁ is continuous in the vertical direction, and the continuation mode code C ₀ and n expressed by the MH code, and the codes C _R (1), C _L (2), etc. corresponding to the shift of the change point, The first change point a ₁ becomes C mode data, the end mode code E is added, and then the second change point a _{2 becomes C mode data of a 2 by} boundary line tracing. At the end of the line, the end mode code E or MR is used as described above.
The EOL code in the code is added.

In compressing the image data of the figure shown in Figure 11 above, in the conventional MR method, if the length of the left side of the rectangle is 100, then H (W100, B500) + 2 (V ₀ × 299) + P = 3 + 5 + 8 + 12 + 12 + 2 x 299 + 4 = 642 (bits). On the other hand, when applying the compression method of the embodiment of the present invention and using MR code, S (W100, B500) + 2 (C ₀ + B 299) + E = 3 + 5 + 8 + 12 + 12 + 2 (1 + 12 + 12) + 4 = 94 (bits), compared to the conventional example. This enables significant compression. This is done by expressing and encoding the number of zero deviations in the vertical direction in the continuous mode as a run length.

In addition, it is possible to sufficiently compress image data not only for figures but also for characters; for example,
When the image data for the characters "pi1" in pixels No. 1 to 42 and lines No. 1 to 29 shown in Fig. 5 are compressed and encoded, the image data shown in Fig. 6 is obtained, and the total number of bits is 252. . Further, in FIG. 5, .about. indicates the boundary line tracing path of the change point, and these .about. are shown correspondingly in the mode column of FIG. Also the 7th
The figure shows the corresponding case of compression encoding using the conventional MR method, and the total number of bits is 261.
It became. In this case, the compression rate according to the embodiment of the present invention is not very high, but if code allocation is optimized, the compression rate can be further improved.

In addition, when the boundary line is complex, for example, in the case shown in Figure 8, it is defined that the images are connected from the upper left to the lower right, and the changing points as shown by S and C are processed as the start mode and the continuation mode. I can do it. In order to simplify the process, (1) pixels at the start and end of all lines are made white pixels. (2) The last line of one page is all white pixels. With such a definition, there is no need to perform virtual change point processing as in the MR method, and it becomes easy to detect the end of the boundary line based on the final line.

〔Effect of the invention〕

As described above, the present invention provides a start mode detection unit 2, etc. for detecting a start mode in which two change points exist only on the encoded scan line, with reference scan lines and encoded scan lines as targets. means such as a continuation mode detection unit 6 for detecting a continuation mode in which a change point exists on the reference scan line and the encoded scan line; and an end mode in which a change point exists only on the reference scan line. By detecting the start mode, each boundary line connected in the sub-scanning direction from two change points of the start mode is tracked in the continuous mode. , until the end mode is detected, and has the advantage of being able to compress data by fully utilizing vertical correlation.

[Brief explanation of drawings]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a mode explanatory diagram of an embodiment of the present invention, Fig. 3 is an explanatory diagram of each mode and code assignment, and Fig. 4 is an explanation of an example of a compression code. Figure 5 is an explanatory diagram of an example of characters, Figure 6 is an explanatory diagram of the image data in Figure 5 compressed by the embodiment of the present invention, and Figure 7 is the data of the image in Figure 5. Figure 8 is an illustration of the continuation and cutting definitions when the boundary line is complex. Figure 9 is an illustration of the path, vertical and horizontal modes of the MR method. FIG. 10 is an explanatory diagram of code assignment in the MR method, and FIG. 11 shows an example of a rectangular figure. 1 is an image data memory with a capacity for one page, 2 is a start mode detection section, 3, 7, and 8 are counters, 4, 5
1 is a register, 6 is a continuation mode detection section, 9 is an end mode detection section, 10 is a multiplexer, 11 is a code conversion memory, 12 is a shift register, and 13 is a counter.

Claims (1)

[Claims] 1. In an image data compression method that compresses image data by two-dimensional encoding, detection of a start mode in which two change points exist only on the encoded scan line, with a reference scan line and an encoded scan line as targets. means for detecting a continuation mode in which a change point exists on the reference scan line and the encoded scan line; and means for detecting an end mode in which a change point exists only on the reference scan line; An image data compression method characterized in that each boundary line connected in the sub-scanning direction from two change points is traced in a continuous mode and encoded up to an end mode.