JP2001148789A - Mmr compression method for binary image and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a high speed MMR compression device for a binary image. SOLUTION: The image compressor adopting the MMR encoding system consists of a change point extract means 71, a serial/parallel control means 72, a 1st buffer 73, a 2nd reference line buffer 74, an address retrieval means 75 for retrieving changed pixels a1 and a2, an address retrieval means 76 for retrieving changed pixels b1, b2, a means 77 that generates a Huffman code and updates a reference pixel a0, and a packing means 78 for the generated Huffman code.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a binary image MMR.
The present invention relates to an image compression process using an encoding method.

[0002]

2. Description of the Related Art Conventionally, image compression processing of a binary image by the MMR coding method is performed by sequential processing based on a standard flow diagram of the MMR coding method shown in FIG.
MR compression was performed.

[0003] A standard flow diagram of the MMR coding system shown in FIG. 1 shows a coding procedure in an MMR (Modified Modified READ) coding system applied to a G4 facsimile apparatus.
Are known.

[0004]

The prior art MMR
If the compression processing is to be realized by software, each step must be executed by sequential processing, and the execution of one step requires several clocks, so that the compression processing speed is slow. If an attempt is made to shorten the compression processing by realizing the MMR compression processing according to the prior art as it is in hardware, the time required for one-step processing can be reduced. Has occurred, and the overall parallel processing efficiency has not been improved.

[0005]

The present invention SUMMARY OF] is intended to eliminate the drawbacks of the MMR compression method according to the prior art binary image is trying to increase the coding processing speed, the reference pixel a ₀ address updates, The address search of the changed pixels a ₁ and a _2, the address search of the changed pixels b ₁ and b ₂ , the generation of the Huffman code, and the packing of the Huffman code are simultaneously performed in parallel, and the address search of the changed pixel and the packing of the Huffman code are described below. Then, the input data is received in parallel, and high-speed processing is performed in parallel (multiple bit width) to increase the efficiency of the parallel operation.

[0006]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 2 is a block diagram showing a method of searching for the address of a changed pixel at a high speed, FIG. 3 is a block diagram showing a packing method of Huffman codes, and FIG. 4 is an explanatory diagram showing a specific example of the packing method. FIG.
Outputting Huffman codes in conformity with MR coding method, and is a block diagram illustrating a method of outputting the address and color of the reference pixel a _0. FIG. 6 is a block diagram illustrating a change point extraction method, and FIG. 7 is a block diagram illustrating a configuration of an MMR image compression apparatus.

In FIG. 2, n-bit input data is
The mask pattern input to and held in the first register 1 and generated by the mask pattern generation unit 4 is stored in the second register 2
Is entered and retained. The output data from the AND circuit 3 that calculates the logical product of the output data from the first and second registers 1 and 2 is input to a mask pattern generation unit 4 to generate a new mask pattern, and to the priority encoder 5 Input and output address (bit position). The input data consisting of n bits is processed as it is in parallel (a plurality of bits in width), and the address of the changed pixel (the bit whose value is 1 by preprocessing) is searched at high speed. In the address search for the changed pixels a ₁ and a _{2 and} the address search for the changed pixels b ₁ and b ₂ , the input data is received in parallel, and high-speed processing is performed in parallel. The functions of the blocks [A ₁ ] to [A ₅ ] in FIG. 2 are as shown in Table 1.

[0008]

[Table 1]

FIG. 3 is a block diagram showing a method for packing Huffman codes. A buffer 11 for holding data (M bits) during packing, an adder 16 and a subtractor 1 for holding and updating the number of empty bits in the buffer 11
And a first barrel shifter (right shift) 13 for shifting the buffer contents to join the input data (Huffman code) and the contents of the buffer 11
A second barrel shifter (left shift) 14 having an adder 19 for shifting the result of joining to re-store the result in the buffer 11, and a multiplexer for providing a shift amount N when the first N bits of the buffer 11 are filled. 15, the input data (Huffman code) is processed in parallel (multiple bit width), and packing is performed at high speed. The functions of the blocks [B ₁ ] to [B ₅ ] in FIG. 3 are as shown in Table 2.

[0010]

[Table 2]

FIG. 4 is an explanatory diagram showing a specific example of the packing method of the Huffman code. Referring to the block diagram shown in FIG. 3 and the functions of the blocks [B ₁ ] to [B ₅ ] shown in Table 2,
In this example, the maximum bit width B of input data (Huffman code) is 13 bits, and the bus width N on the output side is 16 bits. That is, the state where the input data is set immediately after the already stored data using the barrel shifters 13 and 20 is as shown in FIG. If the data has been accumulated in the buffer by 16 bits or more, the state of outputting the first 16 bits of the buffer is shown in FIG.
(). Further, a state where the leading 16 bits of the already output empty buffer are shifted and packed by the barrel shifter 13 and the barrel shifter 20 in FIG.
47 ().

FIG. 5 is a block diagram showing a Hahma system according to the MMR coding system.
And outputs a reference pixel a ₀ Address and color
FIG. 6 is a block diagram illustrating a method of outputting. Change point a₁ And b
_Two A first comparator 51 for examining the positional relationship between
Point a₁ And b₁ Second comparator 5 for checking the positional relationship of
2 and change pixel a₁ And b₁ First to find the signed distance of
And the change pixel a₁ And a_Two Find the distance of
2 subtractor 54 and reference pixel a₀ And change pixel a₁ Distance
And a third subtractor 55 for obtaining the Huffman encoding process
And a register 56 for storing the state (the number of remaining encodings).
Has formed. Further, the comparison results of the comparators 51 and 52 are compared.
Result and input changed pixel a₁ , A _Two, B_Two Ad
Pixel a based on the₀ First act to find the address of
Based on the comparison result of the arithmetic unit 58 and the comparators 51 and 52,
Reference pixel a₀ And the second arithmetic unit 59 for calculating the color of
A Huffman code that complies with the MMR coding method
And outputs the reference pixel a₀ Ask for the address and color.

Blocks [C ₁ ] to [C] in FIG.
₉ ] are as shown in Table 3, and the block [C
Operation Table _7] - [C _9] are shown in Tables 4 to 6.

[0014]

[Table 3]

[0015]

[Table 4]

[0016]

[Table 5]

[0017]

[Table 6]

FIG. 6 is a block diagram of a change point extracting method, in which binary image data is serially received, a changed pixel is converted to 1 and other pixels are converted to 0, and the result is serially output. A point extraction unit 71 and a conversion unit 7 that receives the output of the change point extraction unit 71 and performs serial-to-parallel conversion
2 and outputs the image data after the change point extraction processing. The functions of the blocks [D ₁ ] and [D ₂ ] in FIG. 6 are as shown in Table 7.

[0019]

[Table 7]

FIG. 7 is a block diagram showing the structure of a binary image MMR image compression apparatus according to the present invention. The change point extraction means 71, the serial-to-parallel conversion means 72, and the second memory for buffering image data. 1 buffer 73, a _second buffer 74 for outputting reference line image data, address search means 75 for changed pixels a ₁ and a ₂ , changed pixel b
₁ and b _2, an address search unit 76, a unit 77 for generating the address and color of the Huffman code and the reference pixel a ₀ , and a Huffman code packing unit 78.

Binary image data (given serially)
Is input to the change point extracting means 71, and the changed pixel is converted into a column consisting of 0 and 1 with 1 being the other pixel and 0 being the other pixel.
The image data that has undergone the change point extraction processing by the change point extraction means 71 is input to the serial-to-parallel conversion means 72, and n
It is converted into bit parallel image data. The image data (change point extraction processed) is temporarily stored in the first buffer 73.
And address search means 7 for the changed pixels a ₁ and a ₂
5 is read as needed to provide data. At this time, the read image data is stored in the second buffer 74.
(Reference line buffer) and is also stored. In parallel with this, from the second buffer 74, the already stored image data of the previous line (change point extraction processing completed) is sent to the address search means 76 of the change pixels b ₁ and b _2.
Is read as needed to provide data to

Encoding output from first buffer 73
The line image data is the reference pixel a₀ With the address and color of
Change pixel a₁ , A_Two Into the address search means 75
Change pixel a on the coding line₁ , A_Two Address
Output. Also, from the second reference line buffer 74
The output reference line image data is the reference pixel a₀ Ad
Pixel b with color and color₁ , B_Two Address search means
76 to change pixel b on the reference line₁ , B_Two of
Output address. Further, the changing pixel a₁ , A_Two No
Changed pixel a output from dress search means 75 ₁ , A_Two 
Address and change pixel b₁ , B_Two Address searcher
Changed pixel b output from stage 76₁ , B_Two Address of
Is the Huffman code generation and reference pixel a₀ Enter update means 77
Huffman code according to the MMR coding method (variable
Long) and reference pixel a₀ Generate the address and color of the. In addition,
The reference pixel a₀ Address and color are changed pixel a₁ , A_Two 
Address search means 75 and changed pixel b₁ , B_Two Address of
This is fed back to the search means 76.

The generated Huffman code (variable length) is input to packing means 78 and packed. Incidentally, the address searching means changes the pixel a _1, a _2, change pixels b _1,
Packing means generating a Huffman code of the address searching means and the Huffman code b ₂ receives input data in parallel, is processed at high speed while the parallel (multiple bits wide). The functions of the blocks [E ₁ ] to [E ₈ ] in FIG. 7 are as shown in Table 8.

[0024]

[Table 8]

[0025]

As described in the foregoing, MMR compression method and apparatus for binary image according to the present invention, among the coding procedure in MMR encoding method, the address of the reference pixel a ₀ updated, changed pixel a ₁ , A ₂ and b ₁ , b ₂ , the Huffman code generation and the Huffman code packing are simultaneously performed in parallel, and the address search and the Huffman code search for the changed pixels a ₁ , a ₂ and b ₁ , b ₂ are performed. Regarding code packing, the input data is received in parallel, and high-speed processing is performed in the parallel state, whereby the MMR compression method for a binary image can be speeded up.

[Brief description of the drawings]

FIG. 1 is a standard flow diagram of an MMR coding scheme.

FIG. 2 is a block diagram showing a method for searching for an address of a changing pixel at a high speed.

FIG. 3 is a block diagram showing a packing method of Huffman codes.

FIG. 4 is an explanatory view showing a specific example of a packing method.

5 is a block diagram illustrating a method for outputting a Huffman code in conformity with MMR coding method, and outputs the address and the color of the reference pixel a _0.

FIG. 6 is a block diagram showing a change point extraction method.

FIG. 7 is a block diagram illustrating a configuration of an MMR image compression device.

[Explanation of symbols]

 1, 2 register 3 AND circuit 4 mask pattern generation unit 5 priority encoder 11 buffer 13 barrel shifter 15 multiplexer 16, 19 adder 17, 53-55 subtractor 51, 52 comparator 73, 74 buffer

 ────────────────────────────────────────────────── ─── Continuing from the front page (72) Inventor Mizio Tsukidate 1-9-9 Motohongo-cho, Hachioji-shi, Tokyo F-term in Chuo Denshi Co., Ltd. 5C078 BA23 BA27 5J064 AA03 BA09 BA16 BC04 BC08 BD01 9A001 BB02 BB04 EE02 FF03 FF05 KK56

Claims (6)

[Claims] In the compression processing method according to claim 1] MMR encoding method according to the binary image, the address update of the reference pixel a _0, the address search of change pixels a _1, a _2, change pixels b _1, b ₂ of the address search , And Huffman code generation and Huffman code packing are simultaneously executed, and the address search of the changed pixels a ₁ and a ₂ , the changed pixel b
_1, for the packing of b ₂ address search and Huffman coding, MMR compression method of a binary image, characterized in that the input data so that high speed processing remains parallel (multiple bits wide) received in parallel. 2. A first register for holding input data, a second register for holding a mask pattern, an AND circuit for calculating a logical product of the input data and the mask pattern held in the first register, A mask pattern generating unit that receives the output data from the AND circuit to generate a new mask pattern; and a priority encoder that receives the output data from the AND circuit and generates an address. 2. The method according to claim 1, wherein the processing is performed in parallel (a plurality of bit widths), and an address of a changed pixel (a bit having a value of 1 by preprocessing) therein is searched at high speed. MMR compression method for value image. 3. A buffer for holding data during packing, a register having an adder and a subtractor for holding and updating the number of empty bits in the buffer, and joining input data (Huffman code) with the contents of the buffer. A first barrel shifter (right shift) for shifting the contents of the buffer, a second barrel shifter (left shift) with an adder for shifting the joined result back into the buffer, and a head N of the buffer. And a multiplexer for providing a shift amount N at the time of packing bits, and processing input data (variable length) in parallel (multiple bit width),
2. The method of claim 1, wherein packing is performed at high speed. 4. A changing pixel a₁ , B₁ , B_Two The relationship
First and second comparators for examining;₁ And b₁ First subtraction to find the signed distance of
Vessel, changing pixel a₁ And a _Two The second subtractor that calculates the distance of
And reference pixel a₀ And a₁ Third subtractor for finding the distance of
And a register for holding a state of the encoding process; and the two comparators, the three subtractors, the output of the register, and
And the current a₀ Table that generates Huffman codes from colors
And a reference pixel a based on the calculation results of the comparator and the subtractor.
₀ And a first arithmetic unit for determining the address of the reference pixel a based on the calculation results of the comparator and the subtractor.
₀ And a second calculation unit for calculating the color
Output a Huffman code conforming to the R encoding method, and
Quasi-pixel a₀ Address and color.
2. The MMR compression method for a binary image according to claim 1, wherein: 5. A change point extracting unit that receives binary image data serially, converts a changed pixel to 1 and converts the other pixels to 0, and outputs the result in a serial manner. A conversion unit for performing conversion, extracting a change point of the image data with a small circuit scale, and outputting the change point extraction result in parallel (with a plurality of bits). Item 2. An MMR compression method for a binary image according to Item 1. 6. A compression processing apparatus for a binary image according to the MMR encoding method, comprising:
Change point extracting means for converting the other pixels to 0 and outputting the result (encoded line image data) serially; converting means for performing serial-to-parallel conversion on the result; a buffer for outputting the image data of the change point already extracted as the current of the reference line image data, the reference line image data and the address of the changed pixel b _1, b ₂ of the reference on the line by entering the address and color of the reference pixel a ₀ a first address search means for outputting a second address output address of the changed pixel a _1, a ₂ of the encoding on the line by entering the address and color coded line image data and the reference pixel a ₀ Huffman code (variable length) in accordance with the MMR coding method by inputting search means and addresses of changed pixels a ₁ and a ₂ and b ₁ and b ₂
Constituted by a packing means for packing means for updating the address and color of the reference pixel a _0, the generated Huffman codes (variable length) to generate a change pixel a _1, a ₂
An MMR compression apparatus for a binary image, wherein an address search means for changing pixels b ₁ and b ₂ , a Huffman code generation and a Huffman code packing means are processed in parallel.