JP2004104208A - Information encoding method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information encoding method and apparatus for a binarized still picture by the systematic dither method that promotes highly efficient encoding. <P>SOLUTION: Let the number of threshold values in the main scanning direction of a dither matrix be A, the number of encoded processing unit bits be B, and the greatest common measure of A and B be C, then the number A wherein B=C holds in the case of A≥B is selected, and the number A wherein the A is a divisor of the B in the case of A<B is selected. Thus, properly selecting a relation between the number of the threshold values in the main scanning direction of the dither matrix and the number of encoded processing unit bits reduces the learning time resulting in enhancing the compression efficiency of the encoding. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method and an apparatus for encoding information of a still image binarized by an organized dither method.
[0002]
[Prior art]
In recent years, high-definition still image communication has been demanded. In order to satisfy this demand, the speed of communication lines has been actively increased. However, there is a limit to speeding up communication lines. Accordingly, information coding technology is being developed in parallel with the increase in the speed of communication lines. As an example, a method has been put to practical use in which multi-valued image data is converted into binary pixel data by using an organized dither method, and further encoded to reduce the number of transmission bits.
[0003]
When converting multi-valued image data into binary pixel data using the systematic dither method, the size of the dither matrix is determined based on the properties of the image. That is, in the case of converting an image such as a photographic image in which the difference in brightness is slightly changed, the dither matrix is increased as much as possible, and the number of gradations is increased even if the resolution is reduced. On the other hand, in the case of a character image, the resolution is increased by reducing the dither matrix to reduce the gradient.
[0004]
In addition, the dither processing is generally performed in units of one byte, and threshold values of dither matrices arranged in the main scanning direction are repeatedly arranged from left to right without gaps. Image data in 1-byte units are arranged on this threshold in order from the left without any gap. Slicing is performed using the threshold value, and is binarized as 1 when the value is larger than the threshold value and 0 when the value is smaller than the threshold value. In this case, the leading value of the data in 1-byte units is arranged above a predetermined threshold of the dither matrix, and then the data until the leading value of the data is arranged above the predetermined threshold is determined. One line.
[0005]
This will be described with reference to numerical examples. For example, the number of thresholds in the main scanning direction of the dither matrix is set to nine. In this case, if dither processing is performed on a byte-by-byte basis, the first byte is processed by the first to eighth thresholds. The second byte is processed by the ninth threshold and the first to seventh thresholds. The third byte is processed by the eighth through ninth thresholds and the first through sixth thresholds. Similarly, the ninth byte is processed with the second to ninth threshold values, and the tenth byte is started with the first threshold value. That is, in this case, one byte to nine bytes are defined as one line.
[0006]
Further, the method of encoding to reduce the number of transmission bits does not employ a method of transmitting the pixel data itself. Instead, the data value of the pixel data to be transmitted (hereinafter referred to as input data) is predicted by a predetermined method from the pixel data immediately above (hereinafter referred to as reference data) transmitted one line before. If the predictions match, the reference data has already been transmitted, so that the input data can be significantly compressed and transmitted. There are various data prediction methods. An example is described below. In this method, two types of prediction modes are used together. The outline of two types of prediction modes (prediction mode 1 and prediction mode 2) will be described below.
[0007]
Prediction mode 1
Normally, input data is encoded (compressed) in units of 8 bits. These 8 bits are divided into upper 4 bits and lower 4 bits. The upper 4 bits and the lower 4 bits are represented by two hexadecimal numbers, and 4 bits (4 pixels) are processed collectively. For example, when encoding the upper 4 bits of the input data, first, the data value X of the upper 4 bits of the reference data is read. Next, an address X is searched from a prediction table of 16 addresses × 4 bits provided in advance.
[0008]
In the address X of the prediction table, when the data value X changes to Y in the next line in the past progress, the data value Y is described. Therefore, when the data value of the upper 4 bits of the reference data is X, when the address X of the prediction table is searched, it can be predicted that the data value of the upper 4 bits of the input data should be Y.
[0009]
The predicted data value Y is compared with the data value Z actually described in the input data. If the data value Y is equal to the data value Z, it is determined to be a prediction match, encoded as a prediction match, and transmitted. When the data value Y and the data value Z are different, it is determined that the prediction does not match, and the code of the prediction mismatch and the data value Z are transmitted.
[0010]
The prediction table is configured as follows as an example. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The data value of the upper 4 bits of the reference data (corresponding to the data value X) is set to 16 addresses. The prediction data is stored in each of the addresses. That is, when there is the latest progress in which the data value X has changed to the data value Y in the next line (input data), the data value Y (predicted data) is described. However, since there is no past progress in the first first line, all 0s are described (this 0 is updated in the second and subsequent lines).
[0011]
When the data value Y (predicted data) matches the data value Z (the data value actually described in the input data), the data value Y is held as it is. If the data value Y is different from the data value Z, the data value Y is updated with the data value Z. As a result, the latest progress is always stored in the prediction table. The same processing is performed for the lower 4 bits of the reference data, and the predicted value of the lower 4 bits of pixel data is stored.
[0012]
Prediction mode 2
In this mode, when the upper 4 bits of the input data are encoded (compressed), they are directly compared with the data values of the reference data corresponding to the upper 4 bits. For example, when predicting the upper 4 bits of the input data, the data value X of the reference data is directly compared with the data value Z of the input data. If the data value X is equal to the data value Z, it is determined to be a prediction match, and the code of the prediction match is transmitted. If the data value X is different from the data value Z, it is determined that the prediction does not match, and the code of the prediction mismatch and the data value Z are transmitted.
[0013]
The difference between the prediction mode 1 and the prediction mode 2 is as follows. That is, in the prediction mode 1, it is assumed that each pixel changes for each line. Therefore, it is assumed that the pixel having the data value X one line before should change to the data value Y in the next line, and the data value Y is used as the prediction data. For this purpose, a prediction table is provided and constantly updated to the latest prediction data and held.
[0014]
On the other hand, in the prediction mode 2, it is assumed that each pixel has the data value X one line before, which assumes that the pixel does not change for each line. This data value X is used as prediction data. Usually, the above two prediction modes are used together based on a predetermined rule. Regardless of which prediction mode is used, as described above, the greater the number of pixels that match prediction, the more the number of bits to be transmitted is compressed.
[0015]
[Problems to be solved by the invention]
The prior art described above has the following unsolved problems.
(1) Since the size is determined only by the relationship between the resolution and the number of gradations when the dither matrix is determined as described above, the size of the dither matrix in the main scanning direction and the encoding processing unit (usually 1 byte) When the matching is poor, the compression ratio may decrease.
[0016]
(2) In the prediction mode 1, the prediction data should be updated to the Z only for the address Y indicated by the reference data one line before the input data. That is, if the input data is the first position on a certain line, the reference data must also be the first position on the previous line. However, it is also updated when the reference data at another position in the same line is Y.
[0017]
The longer the number of bits in one line, the greater the probability that such a state will occur. Here, it is updated with the data value of the input data at that position. This data value is not always Z. This prediction table no longer serves as the prediction data at the first position on the following line. It is an object of the present invention to solve the above problems and promote efficient encoding.
[0018]
[Means for Solving the Problems]
The present invention employs the following configuration to solve the above points.
<Configuration 1>
An information encoding method for binarizing multi-valued image data using a dither matrix, encoding the data in units of a predetermined number of bits, and transmitting the image, wherein the threshold number of the dither matrix in the main scanning direction is A, Is set to B, and the greatest common divisor of A and B is set to C. When A ≧ B, the above A that satisfies B = C is selected. When A <B, the above A that is a divisor of B is selected. An information encoding method characterized by selecting.
[0019]
<Configuration 2>
By accepting input data consisting of a bit string of the number of encoding processing units obtained by binarizing multi-valued image data using a dither matrix and reference data consisting of the bit string located one line before the input data, Reference is made to the prediction data from a predetermined prediction table that indicates, as prediction data, a data value that the data value indicated by the reference data should change after one line in the past, and compares the prediction data with the input data. An information encoding device including a prediction unit that determines whether the prediction matches and does not match and updates the prediction data with the input data when the prediction does not match, wherein the one line includes a plurality of the bit strings, The prediction table includes a plurality of sub-tables to which sub-index bits respectively corresponding to the plurality of bit strings are added, A sub-index bit calculating unit configured to calculate the sub-index bit from a position in the one line of the reference data; and a prediction unit configured to calculate the predicted data from the predetermined sub-table designated by the sub-index bit. An information encoding apparatus, comprising: a data calculating unit, wherein, when the values do not match, the prediction data is updated with the input data only in a sub-table designated by the calculated sub-index bits.
[0020]
<Configuration 3>
3. The information encoding device according to configuration 2, wherein the sub-index bits are composed of a plurality of bits.
[0021]
<Configuration 4>
A reference consisting of input data consisting of a bit string of the number of coding processing units obtained by binarizing multi-valued image data using a dither matrix, and a bit string of the number of coding processing units located one line before the input data Data, and based on the reference data, refer to the prediction data from a predetermined prediction table that indicates, as prediction data, a data value that the data value indicated by the reference data should change one line later in the past. An information encoding apparatus that includes a prediction unit that compares the prediction data with the input data to determine whether the prediction matches or does not match, encodes the data, and when the prediction data does not match, updates the prediction data with the input data. The one line is composed of a plurality of bit strings of the number of encoding units, and the prediction table is longer than the bit string of the number of encoding units. A plurality of sub-tables, each of which comprises a plurality of sub-tables, each of which comprises a plurality of bit strings of a plurality of encoding processing units in which a read area is specified by changing a position of a leading bit in the prediction table. The reading area determining means for specifying the reading area from the reference data; and the predicted data reading means for reading the predicted data from the sub-table specified by the reading area determining means. A prediction table updating unit that updates the prediction data with the input data only in a bit range specified by an area.
[0022]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described using specific examples.
<Specific example 1>
As described in the section of the prior art, in the encoding method employing the present invention, reference data one line before input data is indispensable in encoding processing. The encoding method until the reference data for one line is acquired in the information encoding device (hereinafter, referred to as a learning time) cannot exert an effect. In order to minimize the learning time, it is necessary to appropriately select the relationship between the number of bits (usually 8 bits) of the encoding processing unit and the number of thresholds in the main scanning direction of the dither matrix. . Example 1 relates to this method.
[0023]
FIG. 1 is an explanatory diagram (part 1) of the learning time.
(1) is a case where the threshold value of the dither matrix is 12 in the main scanning direction and the number of pixel data encoding units is 8 bits, and (2) is a case where the threshold value of the dither matrix is 16 in the main scanning direction and the pixel is 16 pixels. This is a case where the number of data encoding units is 8 bits. Before describing the learning time, the configuration of the information encoding device, which is the minimum required for the description, will be described with reference to the drawings.
[0024]
FIG. 2 is a block diagram of the configuration of the first embodiment.
As shown in FIG. 2, the information encoding apparatus of the first embodiment includes an input processing unit 1, a buffer management unit 2, a prediction unit 3, a prediction table (upper 4 bits) 4, a prediction table (lower 4 bits) 5, a code generation unit 6, And a line buffer 7.
[0025]
The input processing unit 1 is a unit that receives input data and sends it to the line buffer 7 via the prediction unit 3 and the buffer management unit 2 described later.
The buffer management unit 2 is a unit that reads or writes data one line before from the line buffer 7 in response to a request from the prediction unit 3.
[0026]
When receiving the input data from the input processing unit 1, the prediction unit 3 receives the data one line before (hereinafter referred to as reference data) from the line buffer 7 via the buffer management unit 2. This is a part for calculating prediction data using the reference data, a prediction table (higher 4 bits) 4 described later, and a prediction table (lower 4 bits) 5 described later.
[0027]
Further, this is a part for sending a prediction result (prediction match or prediction mismatch) to the code generation unit 6 described later. Further, in the case of the prediction mode 1, when the prediction does not match, the data value of the prediction table is updated with the data value described in the input data.
[0028]
The prediction table (upper 4 bits) 4 is a table of 16 addresses and 4 bits, and is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The data value of the upper 4 bits of the reference data is set to 16 addresses. The prediction data is stored in each of the addresses. For example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described in the address X. However, since there is no past progress in the first first line, all 0s are described (this 0 is updated in the second and subsequent lines).
[0029]
The prediction table (lower 4 bits) 5 is also a table of 4 bits and 16 addresses like the prediction table (upper 4 bits) 4, and is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The data value of the lower 4 bits of the reference data is set to 16 addresses. The prediction data is stored in each of the addresses. For example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described in the address X. However, since there is no past progress in the first first line, all 0s are described (this 0 is updated in the second and subsequent lines).
[0030]
The code creation unit 6 is a unit that encodes the prediction result received from the prediction unit 3. The line buffer 7 is a part that stores data one line before under the control of the buffer management unit 2.
This concludes the description of the configuration of the information encoding device, and returns to FIG. 1 to describe the learning time.
[0031]
Step S (1) 1
In (1) of FIG. 1, when the prediction unit 3 (FIG. 2) receives the bit string [1] via the input processing unit 1 (FIG. 2), the prediction unit 3 (FIG. 2) sets the buffer management unit 2 (FIG. 2). ) Requires acceptance of reference data. Since no data has been stored in the line buffer 7 (FIG. 2), the prediction results do not match. As a result, the prediction table (upper 4 bits) 4 (FIG. 2) is updated with the data value written in the upper 4 bits of the bit string [1]. Similarly, the prediction table (lower 4 bits) 5 (FIG. 2) is updated with the data value written in the lower 4 bits of the bit string [1]. Here, the bit string [1] is pixel data of an encoding processing unit (usually 8 bits) sliced by the first to eighth threshold values of the dither matrix.
[0032]
Step S (1) 2
In (1) of FIG. 1, when the prediction unit 3 (FIG. 2) receives the bit string [2] via the input processing unit 1 (FIG. 2), the prediction unit 3 (FIG. 2) ) Requires acceptance of reference data. However, since the data has not yet been stored in the line buffer 7 (FIG. 2), the prediction results do not match. As a result, the prediction table (upper 4 bits) 4 (FIG. 2) is updated with the data value written in the upper 4 bits of the bit string [2]. Similarly, the prediction table (lower 4 bits) 5 (FIG. 2) is updated with the data value written in the lower 4 bits of the bit string [2]. Here, the bit string [2] is pixel data of an encoding processing unit sliced by the ninth threshold to the fourth through fourth thresholds of the dither matrix.
[0033]
Step S (1) 3
In (1) of FIG. 1, when the prediction unit 3 (FIG. 2) receives the bit string [3] via the input processing unit 1 (FIG. 2), the prediction unit 3 (FIG. 2) ) Requires acceptance of reference data. However, since the data has not yet been stored in the line buffer 7 (FIG. 2), the prediction results do not match. As a result, the prediction table (upper 4 bits) 4 (FIG. 2) is updated with the data value written in the upper 4 bits of the bit string [3]. Similarly, the prediction table (lower 4 bits) 5 (FIG. 2) is updated with the data value written in the lower 4 bits of the bit string [3]. Here, the bit string [3] is pixel data of an encoding processing unit sliced at the fifth to twelfth thresholds of the dither matrix.
[0034]
Step S (1) 4
In (1) of FIG. 1, when the prediction unit 3 (FIG. 2) receives the bit string [4] via the input processing unit 1 (FIG. 2), the prediction unit 3 (FIG. 2) ) Requires acceptance of reference data. The data stored in step S (1) 1 in the line buffer 7 (FIG. 2) is pixel data of the encoding processing unit sliced by the first to eighth threshold values of the dither matrix. Here, the bit string [4] and the bit string [1] are pixel data sliced with the same threshold, and can be reference data. Therefore, after step S (1) 4, the prediction unit can operate normally.
[0035]
Thereafter, in step S (1) 5, the bit string [2] becomes the reference data, and in step S (1) 6, the bit string [3] becomes the reference data. That is, the bit string [3] becomes one line from the bit string [1]. Similarly, bit strings [4] to [6], bit strings [7] to bit strings [9],... Here, the processing time of the first line from bit string [1] to bit string [3] is the learning time. During this learning time, there is no predictive match of the input data. That is, the longer the learning time is, the lower the compression efficiency of the encoding is. The purpose of the specific example 1 is to reduce the learning time.
[0036]
The learning time is a time for processing the first line, and it is necessary to shorten one line as much as possible in order to reduce the learning time. To this end, it is necessary to appropriately select the relationship between the length of the bit string (the number of bits in the encoding processing unit) and the threshold number in the main scanning direction of the dither matrix.
[0037]
From FIG. 1A, the learning time is a time for processing 24 bits from the bit string [1] to the bit string [3]. These 24 bits are the least common multiple of the bit string (the number of bits of the encoding processing unit (here, 8)) and the threshold number (here, 12) of the dither matrix in the main scanning direction. That is, in order to shorten the learning time, it is necessary to select the relationship between the threshold number in the main scanning direction and the number of bits per unit of the encoding process so that the least common multiple is minimized. In order to satisfy this relationship, the following selection method is proposed.
[0038]
Let A be the threshold number in the main scanning direction of the dither matrix, B be the number of bits in the pixel data encoding processing unit, and C be the greatest common divisor of A and B. When A ≧ B, A is the case where B = C holds. When A <B, A that is a divisor of B is selected.
[0039]
This will be described with reference to a numerical example. In FIG. 1A, C = 4 because A = 12 and B = 8. A = B, and B = C must be satisfied. However, B = 8 and C = 4, which do not satisfy the condition. That is, FIG. 1A is an inappropriate selection. On the other hand, in FIG. 1B, C = 8 because A = 16 and B = 8. A = B, and B = C must be satisfied. Since B = 8 and C = 8, the condition is satisfied. That is, FIG. 1A is an appropriate choice.
[0040]
The points to be noted here are as follows. 1 (2) uses a larger dither matrix than FIG. 1 (1), but the learning time is shorter due to the proper selection. That is, in FIG. 2B, one line is constituted by the bit string [1] and the bit string [2]. As a result, the compression efficiency of the encoding is improved. Next, the case of A <B will be described with actual numerical examples.
[0041]
FIG. 3 is an explanatory diagram (part 2) of the learning time.
(3) is a case where the threshold value of the dither matrix is 6 in the main scanning direction and the encoding processing unit of the pixel data is 8 bits, and (4) is a case where the threshold value of the dither matrix is 4 in the main scanning direction and the pixel data is (5) is a case where the threshold value of the dither matrix is two in the main scanning direction and the coding processing unit of the pixel data is 8 bits.
[0042]
In FIG. 3C, C = 2 because A = 6 and B = 8. A <B, and A = C must be satisfied. However, A = 6 and C = 2, which do not satisfy the condition. That is, FIG. 3C is an inappropriate selection. On the other hand, in FIG. 3D, C = 4 because A = 4 and B = 8. A <B, and A = C must be satisfied. Since A = 4 and C = 4, the condition is satisfied. That is, FIG. 3D is an appropriate choice. In FIG. 3 (5), C = 2 because A = 2 and B = 8. A <B, and A = C must be satisfied. Since A = 2 and C = 2, the condition is satisfied. That is, FIG. 3 (5) is an appropriate choice.
[0043]
In the above description, only the case where the number of encoding processing units is 8 bits is described. However, it is needless to say that even if the number of encoding processing units is other than 8 bits, an appropriate selection can be made if the above condition is satisfied.
[0044]
<Effect of Specific Example 1>
As described above, the learning time is shortened by appropriately selecting the relationship between the threshold value in the main scanning direction of the dither matrix and the number of bits per unit of the encoding process, and consequently the encoding compression efficiency is improved. be able to.
[0045]
<Configuration of Specific Example 2>
As described above in (2) of the problem to be solved by the invention, the update of the prediction data in the prediction mode 1 should be performed only for the reference data one line before. However, if the other reference data in the same line also has a data value Y, the prediction table is updated with this data value Y as well. With this, we cannot expect correct prediction results. It is an object of the second embodiment to solve such a problem and promote efficient encoding.
[0046]
FIG. 4 is a block diagram of the configuration of the second embodiment.
As shown in FIG. 4, the information encoding apparatus according to the second embodiment includes an input processing unit 1, a buffer management unit 22, a prediction unit 23, a prediction table (upper 4 bits) 24, a prediction table (lower 4 bits) 25, a code generation unit 6, And a line buffer 7. Hereinafter, only the differences from the first embodiment will be described.
[0047]
The buffer management unit 22 is a unit that reads or writes data one line earlier in the line buffer 7 in response to a request from the prediction unit 23. Further, it is a part for recognizing the number of the bit being processed currently in one line and informing the prediction unit 23 of the number. With this recognition, the prediction unit 23 can recognize the number of the threshold of the dither matrix at which the bit currently being processed is sliced.
[0048]
When receiving the input data from the input processing unit 1, the prediction unit 23 receives the reference data from the line buffer 7 via the buffer management unit 22. This is a part for calculating prediction data using the reference data, the prediction table (upper 4 bits) 24, and the prediction table (lower 4 bits) 25. Further, it is a part for sending a prediction result (prediction match or prediction mismatch) to the code creation unit 6. Further, in the case of the prediction mode 1, when the prediction does not match, the data value of the prediction table is updated with the data value described in the input data. A sub-index bit calculating unit that calculates a sub-index bit (described later) from the reference data when the reference data is received, and a predicted data calculating unit that calculates predicted data from a predetermined sub-table specified by the sub-index. It is also a part to be provided.
[0049]
The prediction table (upper 4 bits) 24 is a set of a plurality of sub tables in which a sub index 3 bits is added to a 4-bit 16 address as an example. The sub-table is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The data value of the upper 4 bits of the reference data is set to 16 addresses. The prediction data is stored in each of the addresses. For example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described. However, since there is no past progress in the first first line, all 0s are described (this 0 is updated in the second and subsequent lines). Further, a plurality of sub-tables are provided corresponding to respective positions of a plurality of reference data arranged on one line. Three bits of a sub index are added to specify the plurality of sub tables.
[0050]
Like the prediction table (upper 4 bits) 24, the prediction table (lower 4 bits) 25 is a set of a plurality of sub-tables in which a 4-bit 16 address is added with a sub-index of 3 bits as an example. The sub-table is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The data value of the lower 4 bits of the reference data is set to 16 addresses. The prediction data is stored in each of the addresses. For example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described. However, in the first line, all 0s are described since there is no past progress (this 0 is updated in the second and subsequent lines). Further, a plurality of sub-tables are provided corresponding to respective positions of a plurality of reference data arranged on one line. Three bits of a sub index are added to specify the plurality of sub tables.
[0051]
The other components are the same as those in the first embodiment, and the description is omitted.
Next, the principle of calculating the sub-index will be described with reference to the drawings.
FIG. 5 is an explanatory diagram of the sub-index of the matrix 9.
(1) indicates the first row of the dither matrix having the threshold number 9 in the main scanning direction, and the leading value is indicated by an asterisk.
(2) is a diagram in which the first row of the dither matrix is arranged without gaps in the main scanning direction and an encoding processing unit of 8 bits is superimposed thereon.
(3) is a diagram for explaining the specification of a sub-index and a sub-table.
[0052]
As can be seen from (2), a line from the bit string [1] to the bit string [9], which is processed by the coding processing unit number of 8 bits using the dither matrix having the threshold number of 9 in the main scanning direction is one line. In the specific example 2, nine sub-tables corresponding to respective positions from the bit string [1] to the bit string [9] are provided. Therefore, the sub-table corresponding to the bit string [1] is not updated with data other than the input data corresponding to the bit string [1].
[0053]
(3) shows the relationship between the sub-table and the sub-index. In the order from the leftmost on the figure, u · index = 0 as a sub-index of the sub-table corresponding to the bit string [1], u · index = 1 as a sub-index of the sub-table corresponding to the bit string [2], and the bit string [3 ] As a sub-index of the sub-table corresponding to u! Index = 2, u.-index = 3 as a sub-index of the sub-table corresponding to the bit string [4], and u. index = 4, u · index = 5 as a sub-index of the sub-table corresponding to the bit string [6], u · index = 6 as a sub-index of the sub-table corresponding to the bit string [7], and a sub-table corresponding to the bit string [8] U · index = 7 as a sub-index of the table, The u · index = 7 as sub-index of the sub table corresponding to column (9), added respectively.
[0054]
As can be seen from FIG. 5 (3), it is decomposed into upper 4 bits and lower 4 bits. In the sub-table of u · index = 0, the head value (star) of the dither matrix is arranged at the leftmost position. In the sub-table of u · index = 1, the head value (star) of the dither matrix is placed second. Similarly, the u-index = 4 sub-table is the third sub-table, the u-index = 3 sub-table is the fourth sub-table, the u-index = 4 sub-table is the fifth sub-table, and the u-index = 4 sub-table In the sub-table with u.index = 5, the sub-table with u.index = 5 is arranged sixth, in the sub-table with u.index = 6 seventh, and in the sub-table with u.index = 7 eighth.
[0055]
When the head value (star mark) of the dither matrix is not included, it is determined that the shift is the same as the previous one, and the previous sub-index is added. The sub-table corresponding to the bit string [9] in the figure corresponds to the sub-table. The buffer management unit 22 recognizes the position of the head value (star) of the dither matrix and the threshold value of the dither matrix at which the bit being currently processed is data sliced. Therefore, by recognizing the threshold value of the first position of the encoding processing unit or recognizing where the position of the star is located in the encoding processing unit, the corresponding sub-index (u Index) can be easily detected.
[0056]
<Operation of Specific Example 2>
FIG. 6 is a flowchart of the operation of the second embodiment.
The operation of the information coding apparatus according to the specific example 2 will be described according to steps S (2) 1 to S (2) 13 on the drawing.
[0057]
Step S (2) 1
All parts in the device are initialized and start operation.
Step S (2) 2
The prediction unit 23 (FIG. 4) starts accepting data via the input processing unit 1 (FIG. 4). When input data for the first line is received, reference data is stored in the line buffer 7 (FIG. 4). Thereafter, the information encoding device starts normal encoding. The prediction unit 23 (FIG. 4) receives input data via the input processing unit 1 (FIG. 4). The reference data is received from the line buffer 7 (FIG. 4) via the buffer management unit 2 (FIG. 4). Here, [X] [Y] of the input data [X] [Y] and the reference data [X] [Y] on FIG. 6 indicate the positions of the data on the XY plane.
[0058]
Step S (2) 3
The sub-index bit calculation means of the prediction unit 23 (FIG. 4) calculates u · index from the data value of the reference data. This operation will be described again later using another drawing.
Step S (2) 4
It is determined whether or not the current process corresponds to mode 1, and if it corresponds to mode 1, the process proceeds to step S (2) 5, and if not, to step S (2) 6.
[0059]
Step S (2) 5
The prediction data calculation unit of the prediction unit 23 (FIG. 4) calculates a prediction table (upper 4 bits) 24 (FIG. 4) and a prediction table (lower 4 bits) based on the u · index calculated by the sub-index bit calculation unit. 25 (FIG. 4) to select a predetermined sub-table. The prediction data is calculated by searching this sub-table. This operation will be described again later using another drawing.
Step S (2) 6
Since this corresponds to prediction mode 2 (explained in the related art), the data value of the reference data is used as it is as prediction data.
[0060]
Step S (2) 7
The prediction unit 23 (FIG. 4) determines whether the input data and the prediction data match or does not match. If they match, the process proceeds to step S (2) 8. Goes to step S (2) 9.
Step S (2) 8
When the prediction unit 23 (FIG. 4) notifies the code generation unit 6 (FIG. 4) of the coincidence, the code generation unit 6 (FIG. 4) allocates a short code and transmits the code data.
[0061]
Step S (2) 9
When the prediction unit 23 (FIG. 4) notifies the code creation unit 6 (FIG. 4) of the mismatch, the code creation unit 6 (FIG. 4) adds the input data to the mismatched code and sends out the code data. Simultaneously, in the case of the prediction mode 1, the prediction unit 23 (FIG. 4) updates the data values of the prediction table (upper 4 bits) 24 (FIG. 4) and the prediction table (lower 4 bits) 25 with the data values of the input data.
[0062]
Step S (2) 10
The buffer management unit 22 (FIG. 4) controls the line buffer 7 (FIG. 4) to update the reference data with the input data.
The image information to be transmitted is processed over the entire XY plane by the return loop from step S (2) 11 to step S (2) 14 and step S (2) 2, and the flow ends.
[0063]
Next, step S (2) 3 will be described again with reference to another drawing.
The principle of the detection of the sub-index is already referred to by recognizing the threshold value of the first position of the coding processing unit, or by recognizing where the position of the star is located in the coding processing unit. FIG. 5 has already described that the corresponding sub-index (u · index) can be easily detected from the data.
[0064]
Here, an example of a flow in which the prediction unit 23 (FIG. 4) calculates the above step S (2) 3 will be described by actually applying the numerical values in FIG.
FIG. 7 is a flowchart of the sub-index bit calculation means.
In this flowchart, bcount is a counter that counts at which threshold of the dither matrix the bit being processed is sliced. However, in FIG. 5, the threshold value is represented by 1 to 9, but is represented by 0 to 8 for the convenience of arithmetic processing.
[0065]
As an example, assume that the prediction unit 23 (FIG. 4) has accepted the bit string [10] in FIG. As shown in the figure, the first data of the bit string [10] is sliced at the threshold 1 of the dither matrix. Therefore, bcount = 0. Hereinafter, u · index of the bit string [10] is calculated according to steps S (2) 15 to S (2) 20.
[0066]
Step S (2) 15
Add 8 to bcount. Therefore, bcount = 8.
Step S (2) 16
Compare bcount = 8 with the matrix length. Here, since bcount = 8 is smaller than the matrix length 9, the process proceeds to step S (2) 17.
[0067]
Step S (2) 17
It is determined whether or not bcount is 0. Here, since bcount = 8, the process proceeds to step S (2) 19.
Step S (2) 19
It is determined whether or not bcount is greater than 8. Here, since bcount = 8, the process proceeds to step S (2) 20.
[0068]
Step S (2) 20
The u-index is obtained as 8-bcount = u-index. Here, since bcount = 8, u · index = 0 is obtained, and the flow ends.
[0069]
In FIG. 5C, u · index of the prediction table of the upper 4 bits in which the first data is sliced by the threshold 1 of the dither matrix is 0, which matches the above calculation result.
Next, step S (2) 5 will be described again with reference to another drawing.
[0070]
FIG. 8 is a flowchart of the prediction data calculation means.
The prediction data calculation means of the prediction unit 23 (FIG. 4) calculates prediction data according to steps S (2) 21 to S (2) 24.
Here, in the figure, Est.Index1 is an index indicating the address of the prediction table (upper 4 bits) 24, and Est.Index2 is an index indicating the address of the prediction table (lower 4 bits) 25. “Refer” indicates reference data.
[0071]
Step S (2) 21
The address of the prediction table (upper 4 bits) is obtained by shifting the 3 bits of u · index to the left and shifting the reference data to the right by 4 bits to obtain an OR.
Step S (2) 22
By shifting the 3 bits of u-index leftward and masking the upper 4 bits of the reference data and performing an OR operation, the address of the prediction table (lower 4 bits) 25 is obtained.
[0072]
Step S (2) 23
The table is searched from the addresses obtained in steps S (2) 21 and S (2) 22.
Step S (2) 24
The flow ends with the value retrieved in step S (2) 23 as prediction data.
[0073]
The above description has been limited to the case where three bits of the sub-index are added to the prediction table. However, the invention is not limited to this example. That is, when one line is long, there is a case where it is not enough to add the sub-index of 3 bits. In that case, the number of bits may be further increased.
[0074]
Although the flowchart of the sub-index bit calculating means has been described with reference to FIG. 7, the sub-index bit calculating means of the present invention is not limited to this example. That is, similar to calculating by recognizing the threshold value of the first position of the encoding processing unit from the reference data, or by recognizing where the position of the star is located in the encoding processing unit. If it is a method, you may calculate using another algorithm.
[0075]
<Effect of Specific Example 2>
As described above, by providing a plurality of sub-tables to which different sub-indexes are added, overwriting (updating) of the prediction table can be avoided, and the prediction accuracy can be improved.
[0076]
<Specific example 3>
In the specific example 2, in order to avoid overwriting of the prediction table, a plurality of sub tables are provided by adding a sub index. In this case, the number of memories increased significantly. For example, in FIG. 4, 3 bits are added to a sub index in order to provide eight sub tables, and the prediction table is increased to 128 addresses. The specific example 3 aims at not increasing the address even if a plurality of sub tables are provided.
[0077]
FIG. 9 is a block diagram of the configuration of the third embodiment.
As shown in FIG. 9, the information encoding apparatus according to the third embodiment includes an input processing unit 1, a buffer management unit 22, a prediction unit 33, a prediction table (upper 4 bits) 34, a prediction table (lower 4 bits) 35, a code generation unit 6, And a line buffer 7. Hereinafter, only a portion different from the specific example 1 or the specific example 2 will be described.
[0078]
When receiving the input data from the input processing unit 1, the prediction unit 33 receives the reference data from the line buffer 7 via the buffer management unit 22. This is a part for calculating prediction data using the reference data, a prediction table (higher 4 bits) 34 and a prediction table (lower 4 bits) 35 described later. Further, it is a part for sending a prediction result (prediction match or prediction mismatch) to the code creation unit 6. Further, in the case of the prediction mode 1, when the prediction does not match, the data value of the prediction table is updated with the data value described in the input data. A reading area determining means for determining a reading area to be described later from the reference data when the reference data is received, a prediction data reading means for reading the prediction data using the reading area as a predetermined sub-table, and a prediction table when the prediction does not match. And a prediction table updating means for updating the information.
[0079]
The prediction table (upper 4 bits) 34 is, for example, a table of 11 bits and 16 addresses. The prediction data is stored in these 11 bits. There are provided a plurality of sub-tables each consisting of four consecutive bits from a predetermined bit position in the 11 bits. That is, a plurality of sub-tables are included in 11 bits specified by one address. The predetermined bit position required to specify this sub-table is determined by the reading area determining means of the prediction unit 33. The address is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The 16 upper 4-bit data values are set as the address. In a plurality of sub-tables at the same address, for example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described. However, in the first line, all 0s are described since there is no past progress (this 0 is updated in the second and subsequent lines). Further, a plurality of sub tables are provided corresponding to a plurality of reference data arranged on one line.
[0080]
The prediction table (lower 4 bits) 35 is, for example, a table of 11 bits and 16 addresses. The prediction data is stored in these 11 bits. There are provided a plurality of sub-tables each consisting of four consecutive bits from a predetermined bit position in the 11 bits. That is, a plurality of sub-tables are included in 11 bits specified by one address. The predetermined bit position required to specify this sub-table is determined by the reading area determining means of the prediction unit 33. The address is configured as follows. One byte of reference data is decomposed into upper 4 bits and lower 4 bits. The 16 data values of the lower 4 bits are set as the address. In a plurality of sub-tables at the same address, for example, when there is the latest progress in which the data value X has changed to the data value Y in the next line, the data value Y is described. However, in the first line, all 0s are described since there is no past progress (this 0 is updated in the second and subsequent lines). Further, a plurality of sub tables are provided corresponding to a plurality of reference data arranged on one line.
[0081]
The other components are the same as those of the first or second embodiment, and thus the description is omitted. Next, the basic principle of the reading area determining means will be described with reference to the drawings.
FIG. 10 is an explanatory diagram of a reading area of the matrix 9.
(1) indicates the first row of the dither matrix having the threshold number 9 in the main scanning direction, and the leading value is indicated by an asterisk.
(2) is a diagram in which the first row of the dither matrix is arranged without gaps in the main scanning direction, and an encoding processing unit of 8 bits is superimposed thereon.
FIG. 3C is a diagram illustrating the specification of a reading area and a sub table.
[0082]
As can be seen from (2), a line from the bit string [1] to the bit string [9], which is processed by the coding processing unit number of 8 bits using the dither matrix having the threshold number of 9 in the main scanning direction is one line. In the specific example 3, nine sub tables corresponding to the bit strings [1] to [9] are provided. These nine sub-tables correspond to nine stages of bit strings from top to bottom of (3). Therefore, the sub-table corresponding to the bit string [1] is not updated with data other than the input data corresponding to the bit string [1].
[0083]
(3) indicates the relationship between the sub-table and the reading area and the sub-index for specifying the sub-table. From the top of the figure, v · index = 0 is determined as a sub-index of the sub-table corresponding to the bit string [1]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [1], both the upper 4 bits and the lower 4 bits are arranged from the eighth bit position in the prediction table. Furthermore, the first threshold of the dither matrix is at the first position.
[0084]
As the sub-index of the sub-table corresponding to the bit string [2], v.index = 1 is determined. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [2], both the upper 4 bits and the lower 4 bits are arranged from the bit position 7 in the prediction table. Furthermore, the first threshold of the dither matrix is at the second position.
[0085]
It is determined that v · index = 2 as a sub-index of the sub-table corresponding to the bit string [3]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [3], both the upper 4 bits and the lower 4 bits are arranged from the sixth bit position in the prediction table. Further, the first threshold of the dither matrix is at the third position.
[0086]
It is determined that v.index = 3 as a sub-index of the sub-table corresponding to the bit string [4]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [4], both the upper 4 bits and the lower 4 bits are arranged from the fifth bit position in the prediction table. Furthermore, the first threshold of the dither matrix is at the fourth position.
[0087]
It is determined that v · index = 4 as a sub-index of the sub-table corresponding to the bit string [5]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [5], both the upper 4 bits and the lower 4 bits are arranged from the fourth bit position in the prediction table. Furthermore, the first threshold of the dither matrix is at the fifth position.
[0088]
It is determined that v.index = 5 as a sub-index of the sub-table corresponding to the bit string [6]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [6], both the upper 4 bits and the lower 4 bits are arranged from the third bit position in the prediction table. Furthermore, the first threshold of the dither matrix is at the sixth position.
[0089]
As the sub-index of the sub-table corresponding to the bit string [7], v.index = 6 is determined. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [7], both the upper 4 bits and the lower 4 bits are arranged from the second bit position in the prediction table. Further, the first threshold of the dither matrix is at the seventh position.
[0090]
As the sub-index of the sub-table corresponding to the bit string [8], v.index = 7 is determined. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [8], both the upper 4 bits and the lower 4 bits are arranged from the first bit position in the prediction table. Further, the first threshold of the dither matrix is at the eighth position.
[0091]
It is determined that v.index = 7 as a sub-index of the sub-table corresponding to the bit string [9]. Here, it is divided into upper 4 bits and lower 4 bits. In the bit string [9], both the upper 4 bits and the lower 4 bits are arranged from the first bit position in the prediction table. Furthermore, the first threshold of the dither matrix is not included.
[0092]
Attention should be paid to the following points. In all the bit strings except the bit string [9] described in the last stage, the threshold value of the dither matrix is the same in all the vertical directions of the table. This means that 9 sub-tables having different read areas can be included in the same address 11 bits. Therefore, if the read area is moved based on the sub-index, data of a desired sub-table can be easily read. Also, the bit string [9] can be handled in the same manner because the data shift from the bit string [8] is the same as the other bit strings, so that it is not overwritten with the bit string [8].
[0093]
<Operation of Specific Example 3>
FIG. 11 is a flowchart of the operation of the third embodiment.
The operation of the information coding apparatus according to the third embodiment will be described with reference to the drawings.
However, the operation is different from the operation of the specific example 2 only in three steps of v · index calculation in step S (3) 1, prediction data reading in step S (3) 2, and prediction mismatch processing in step S (3) 3. . Since all other steps are the same as those in the specific example 2, only the above three steps will be described in detail with reference to other drawings, and description of the other steps will be omitted.
[0094]
FIG. 12 is a flowchart of the reading area determining means. This is a flow corresponding to the calculation of v · index in step S (3) 1.
In the figure, Est · index1 is the index (address) of the prediction data of the upper 4 bits, and Est · index2 is the index (address) of the prediction data of the lower 4 bits. Bmasc is information indicating the position of the reading area.
[0095]
Step S (3) 1-1
The prediction unit 33 (FIG. 9) decomposes the reference data into upper 4 bits and lower 4 bits, and stores the upper 4 bits in Est.index1 and the lower 4 bits in Est.index2.
[0096]
Step S (3) 1-2
The prediction unit 33 (FIG. 9) determines the sub-index (v · index) by recognizing, from the reference data, the position of the first threshold of the dither matrix in the bit string. The value of this v-index is stored in bmasc, and the flow ends. In this step, a reading area for searching a predetermined sub-table from the reference data is determined.
[0097]
FIG. 13 is a flowchart of the prediction data reading means. This is a flow corresponding to the prediction data reading in step S (3) 2. In the figure, temp1 and temp2 are portions that temporarily hold data.
Step S (3) 2-1
The prediction unit 33 (FIG. 9) reads 4-bit data from the 11 bits specified by the Est.index1 using the read area (bmasc) obtained in step S (3) 1-2 (FIG. 12). And temporarily store it in temp1. Similarly, 4-bit data is read out of 11 bits specified by Est.index2 and temporarily stored in temp2.
[0098]
Step S (3) 2-2
The prediction unit 33 (FIG. 9) shifts the data temporarily stored in temp1 to the right by the value of v · index (from 0 to 10). Similarly, the data temporarily stored in temp2 is shifted to the right by the value of v.index (from 0 to 10).
[0099]
Step S (3) 2-3
The data of temp1 is shifted left by 4 bits, and the data of temp2 is stored as it is.
Step S (3) 2-4
The flow ends with the data of temp1 as the prediction data of the upper 4 bits and the data of temp2 as the prediction data of the lower 4 bits.
[0100]
FIG. 14 is a flowchart of the prediction table updating means. This is a flow corresponding to the prediction mismatch processing in step S (3) 3. In the figure, current means input data.
Step S (3) 3-1
The prediction unit 33 (FIG. 9) reads 4-bit data from the 11 bits specified by the Est.index1 using the read area (bmasc) obtained in step S (3) 1-2 (FIG. 12). And temporarily store it in temp1. Similarly, 4-bit data is read out of 11 bits specified by Est.index2 and temporarily stored in temp2.
[0101]
Step S (3) 3-2
The prediction unit 33 (FIG. 9) shifts and reads the upper 4 bits of the input data to the right. This data is shifted left by v · index. The logical sum of this data and the data stored in temp1 in step S (3) 3-1 is calculated and stored again in temp1. Similarly, the lower 4 bits of the input data are taken, and this data is left shifted by v · index. The logical sum of this data and the data stored in temp2 in step S (3) 3-1 is calculated and stored again in temp2.
[0102]
Step S (3) 3-3
The prediction unit 33 (FIG. 9) overwrites the data stored in temp1 on the original position read in step S (3) 3-1 to finish updating the upper 4 bits. Similarly, the data stored in temp2 is overwritten on the original position read in step S (3) 3-1, and the updating of the lower 4 bits is completed.
[0103]
The above description has been limited to the case where the number of threshold values of the dither matrix in the main scanning direction is 9 and the encoding processing unit is 8 bits. However, the invention is not limited to this example. Even if the number of thresholds of the dither matrix and the number of bits of the encoding processing unit are different, it is possible to perform processing on the same principle.
[0104]
<Effect of Specific Example 3>
As described above, by including a plurality of sub-tables in 11 bits (an example) of the same address, the effect of the specific example 2 can be obtained without increasing the size of the memory.
[Brief description of the drawings]
FIG. 1 is an explanatory diagram of learning time (No. 1).
FIG. 2 is a block diagram of a configuration of a specific example 1.
FIG. 3 is an explanatory diagram (part 2) of learning time.
FIG. 4 is a block diagram of a configuration of a specific example 2.
FIG. 5 is an explanatory diagram of a sub-index of a matrix 9;
FIG. 6 is a flowchart of an operation of a specific example 2.
FIG. 7 is a flowchart of a sub-index bit calculation unit.
FIG. 8 is a flowchart of a prediction data calculation unit.
FIG. 9 is a block diagram of a configuration of a specific example 3.
FIG. 10 is an explanatory diagram of a reading area of a matrix 9;
FIG. 11 is a flowchart of an operation of a specific example 3.
FIG. 12 is a flowchart of a reading area determining unit.
FIG. 13 is a flowchart of prediction data reading means.
FIG. 14 is a flowchart of a prediction table updating unit.
[Explanation of symbols]
1-16 Threshold of dither matrix
S (1) 1 to S (1) 10 steps
[1] to [10] Bit string (encoding processing unit)

Claims (4)

An information encoding method for binarizing multi-valued image data using a dither matrix, encoding the image data in units of a predetermined number of bits, and transmitting the image,
The threshold number in the main scanning direction of the dither matrix is A, the predetermined number of bits is B, and the greatest common divisor of A and B is C,
When A ≧ B, select the above A where B = C holds,
An information encoding method comprising: selecting A which is a divisor of B when A <B. By accepting input data consisting of a bit string of the number of encoding processing units obtained by binarizing multi-valued image data using a dither matrix and reference data consisting of the bit string located one line before the input data, The data value indicated by the reference data is referred to from a predetermined prediction table that indicates, as predicted data, a data value that should have changed after one line in the past, and the predicted data is compared with the input data. An information encoding device including a prediction unit that determines whether the prediction matches and does not match, and updates the prediction data with the input data when the mismatch occurs.
The one line includes a plurality of the bit strings,
The prediction table includes a plurality of sub-tables to which sub-index bits respectively corresponding to the plurality of bit strings are added,
The prediction unit includes:
Sub-index bit calculation means for calculating the sub-index bit from a position in the one line of the reference data,
Prediction data calculation means for calculating the prediction data from the predetermined sub-table specified by the sub-index bit,
The information encoding apparatus according to claim 1, wherein, when said mismatch is found, said prediction data is updated with said input data only in a sub-table designated by said calculated sub-index bit. The information encoding device according to claim 2,
The sub-index bits are
An information encoding device comprising a plurality of bits. Input data consisting of a bit string of the number of coding processing units obtained by binarizing multi-valued image data using a dither matrix, and a reference consisting of a bit string of the number of coding processing units located one line before the input data. Data, and based on the reference data, refer to the prediction data from a predetermined prediction table that indicates, as prediction data, a data value that the data value indicated by the reference data should change one line later in the past. An information encoding device including a prediction unit that compares the prediction data with the input data to determine whether the prediction matches or does not match, encodes the data, and when the prediction does not match, updates the prediction data with the input data. hand,
The one line is composed of a plurality of bit strings of the number of the encoding processing units,
The prediction table is composed of a bit string longer than the bit string of the number of encoding units,
A plurality of sub-tables respectively corresponding to a plurality of the bit strings of the plurality of encoding processing units in which the read area is specified by changing the position of the first bit in the prediction table,
The prediction unit includes:
Reading area determining means for specifying the reading area from the reference data;
Prediction data reading means for reading the prediction data from the sub-table specified by the reading area determination means, and a prediction for updating the prediction data with the input data only in a bit range specified by the determined reading area when there is a mismatch. An information encoding device comprising: a table updating unit.

Images