JPH0823440A - Image signal processing method and device - Google Patents

Abstract

PURPOSE:To improve quality of an image interpolated/magnified by classifying patterns of plural picture element data comprising plural adjacent columns in horizontal and vertical lines of an original image signal and controlling the size of the classification area based on the result of classification. CONSTITUTION:Plural picture element data in, e.g. 3X2 picture element areas in a 2nd area comprising plural adjacent columns in 7X2 picture element areas stored in a shift register 13 being a means storing, e.g. 7X2 picture elements in a 1st area are classified and the size of the 3X2 picture element area is controlled based on a pattern recognition result. Then interpolation is made depending on the result of recognition by using the plural picture element data in the 2nd area and a magnified picture element signal is formed by inserting interpolation picture element data between picture element data through the use of an interpolation arithmetic circuit 14 and an interpolation register 15. The register 13 conducts interpolation by using data of the 1st area read from line buffer memories 10a, 10b.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing method and apparatus suitable for, for example, a video printer, for forming an enlarged image signal by enlarging an original image from the original image signal.

[0002]

2. Description of the Related Art Conventionally, when printing an enlarged image in a video printer for obtaining a hard copy of a television image, for example, the number of pixels of the printed image is 1/2 or 1/4 of the number of printable pixels. In that case, the same pixel was printed twice or four times in succession. That is, in the case where the thermal head of the video printer has a head element corresponding to 2000 dots, for example, and the printed image has only 1000 dots in the head direction, the same data of 2 dots in the head direction is used. It is possible to print an enlarged image by printing side by side and similarly in the head feeding direction.

[0003]

The configuration for obtaining an enlarged image as described above can be realized very easily in terms of circuit. However, the enlarged image thus obtained has poor image quality in terms of gradation, resolution, jerkiness, and the like.

Therefore, the present invention has been made in view of such circumstances, and it is possible to obtain a sufficiently enlarged image in terms of gradation, resolution, and jerkiness, and the circuit is not complicated. An object is to provide an image signal processing method and device that can be realized.

[0005]

The present invention has been made in view of such circumstances, and is an image signal processing method for forming an enlarged image signal by enlarging an original image from the original image signal. The plurality of pixel data in a predetermined area consisting of a plurality of adjacent rows in the horizontal and vertical directions are subjected to pattern classification according to a predetermined table, and the predetermined classification is performed based on the pattern classification result. By adaptively controlling the size of the area, using the plural pixel data in the predetermined area in which the pattern is classified and the size is adaptively controlled,
An interpolation operation is performed according to the pattern classification result, and the data of the interpolation pixel by the interpolation operation is inserted between the data of each pixel of the original image signal to form an enlarged image signal.

Here, the predetermined area may be an area formed by a column of a plurality of adjacent pixels of the original image signal and a column of a plurality of interpolated pixels obtained by the interpolation calculation. Further, when adaptively controlling the size of the predetermined area, a predetermined table is set for the data of a plurality of pixels in a small area consisting of a plurality of adjacent horizontal and vertical rows of the original image signal. According to the pattern classification result of a plurality of pixels in the small area, an interpolation calculation is performed using data of a plurality of pixels in the small area, or the small area is centered. It is determined whether or not an interpolation calculation is performed using data of a plurality of pixels in a large area formed by a plurality of enlarged horizontal and vertical adjacent columns. Further, the small area is composed of, for example, data of 3 × 2 pixels in three rows in the horizontal direction and two rows in the vertical direction adjacent to the original image signal, and the large area is centered on, for example, 3 × 2 pixels in the small area. It is composed of data of 7 × 2 pixels, which are adjacent to each other and expanded in seven rows in the horizontal direction and two rows in the vertical direction.

Next, the image signal processing apparatus according to the present invention is a storage means for storing a plurality of pixel data in a first area consisting of a plurality of adjacent columns of horizontal and vertical columns of an original image signal. And the first stored in the storage means.
Pattern data according to a predetermined table for a plurality of pixel data in a second region consisting of a plurality of horizontal and vertical columns adjacent to each other in the region, and based on the pattern classification result. The size of the second region is adaptively controlled, and the plurality of pixel data in the second region in which the pattern is classified and the size is adaptively controlled is used to perform an interpolation calculation according to the pattern classification result. And an enlarged image signal forming means for forming an enlarged image signal by inserting the data of the interpolated pixel by the interpolation calculation between the data of the respective pixels of the original image signal.

Here, the predetermined area may be formed by a column composed of a plurality of pixels of the original image signal and a column composed of a plurality of adjacent interpolated pixels obtained by the interpolation calculation. Further, the enlarged image signal forming means interpolates using the data of a plurality of pixels in the second area based on a pattern classification result according to a predetermined table for the data of a plurality of pixels in the second area. It is determined whether the calculation is performed or the interpolation calculation using the data of a plurality of pixels in the first region is performed. further,
The plurality of pixel data in the first area is composed of 7 columns in the horizontal direction and 2 columns in the vertical direction which are adjacent to the original signal of the 1 field.
× 2 pixel data, and the second area is the first area
3 adjacent 3 rows in the horizontal direction and 2 rows in the vertical direction
It is data of × 2 pixels.

Further, the image signal processing device of the present invention stores w pieces (w is 2) each of which stores the line data of the original image signal.
(Above) and a line data supply means for outputting the same line data v times to supply to the w line buffers, and the storage means stores the data read from the line buffer, The enlarged image signal forming means performs an interpolation operation using the data extracted from the w number of line buffers and stored in the storage means, and interpolates the interpolation data by the arithmetic operation into the w number of line buffers. It is arranged to overwrite the part where the data for which the calculation has been completed is stored.

[0010]

According to the image signal processing method and apparatus of the present invention,
For a plurality of pixel data in a region consisting of a plurality of adjacent columns in the horizontal and vertical columns of the original image signal,
The pattern classification is performed according to a predetermined table, and the size of the pattern classification area is adaptively controlled based on the pattern classification result. When the pattern classification area can be reduced, this pattern classification can be easily performed. If the pattern classification area is small, the quality of the image obtained by the interpolation deteriorates. By increasing the pattern classification area, the quality of the interpolated and enlarged image is improved.

Further, since the interpolation data by the enlarged image signal forming means is written over the portion of the w line buffers where the data for which the interpolation calculation is completed is stored, a separate memory is provided for the interpolation data. No need.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described below with reference to the drawings.

As shown in FIG. 1, the image signal processing apparatus of the embodiment of the present invention for realizing the image signal processing method of the present invention has a plurality of adjacent horizontal and vertical columns of original image signals. Shift register 13 as storage means for storing 7 × 2 pixel data in the first region (7 × 2 pixel region in this embodiment) consisting of 7 × 2 pixels stored in the shift register 13. Tables 6 to 14, which will be described later, show a plurality of pixel data in a second region (3 × 2 pixel region in this embodiment) formed of a plurality of adjacent horizontal and vertical columns in the region. Pattern classification (hereinafter referred to as pattern recognition) is performed according to a predetermined table shown, and the above 3 × is performed based on the pattern recognition result.
The size of the two-pixel region is adaptively controlled, and the interpolation calculation according to the pattern recognition result is performed using the plural pixel data in the second region in which the pattern is classified and the size is adaptively controlled. Interpolation calculation circuit 1 as an enlarged image signal forming means for forming the enlarged image signal by inserting the data of the interpolated pixel by the interpolation operation between the data of each pixel of the original image signal.
4 and the interpolation value register 15.

Further, the image signal processing device of the present invention stores w pieces (where w is 2) each of which stores the line data of the original image signal.
As described above, in the present embodiment, for example, the line buffer memories 10a and 10b of 2 or 4) and the same line data 2 or 4 are stored.
The above two line buffer memories 10a, 1 are output once.
0b is provided with an input control circuit 6 as a line data supply means, the shift register 13 stores the data read from the line buffer memories 10a and 10b, and the enlarged image signal forming means uses the line Interpolation calculation is performed using the data of the first area extracted from the buffer memories 10a and 10b and stored in the shift register 13, and the interpolation data by the interpolation calculation is interpolated in the line buffer memories 10a and 10b. The part where the finished data of is stored is overwritten.

In this embodiment, a method applied to a video printer having a high-precision thermal head, for example, and enlarging a relatively small image has been described. That is, in the case where the thermal head is applied to a video printer including a head element corresponding to, for example, 2000 dots, when the printed image has only 1000 dots in the head direction. In addition, an image signal which is magnified 2 times or 4 times in the head direction is formed.

That is, in FIG. 1, a print data write clock is supplied to the terminal 1, 8-bit image data or print data from the frame memory is supplied to the terminal 3, and a signal indicating each color printing state is supplied to the terminal 4. A certain head active signal and a print timing pulse are supplied to the terminal 5. A print data request signal for the frame memory is output from the terminal 2. Of these terminals 1 to 5, terminals 1, 2 and 3 are connected via a frame memory and an interface circuit (not shown),
The terminals 4 and 5 are connected to the input control circuit 6 via a CPU (central processing unit) (not shown) and an interface circuit. The memory in the previous stage of the input control circuit 6 may be a field memory.

The input control circuit 6 controls fetching of data from the frame memory to the line buffer memories 10a and 10b of the next stage, as described later.

The memories 10a and 10b are 1-line buffer memories for holding image data, which are print data supplied to the terminal 3 via the input control circuit 6, and are, for example, static RAM (SRA).
M).

The work table transfer control circuit 9 takes out data from the memories 10a and 10b. That is, the work table transfer control circuit 9
Is a circuit for performing address control for taking out image data, which is used in the image enlargement and which is necessary for an interpolation calculation described later, from the memories 10a and 10b. The work table transfer control circuit 9 causes the memories 10a and 10b to operate.
The necessary data (7 × 2 pixel data in the case of the present embodiment) is taken out from and is transferred to a work table composed of the shift register 13, for example.

Shift register 13 of the work table
Of the image data transferred to, the pixel data necessary for the interpolation calculation to be described later is sent to the interpolation calculation circuit 14, where pattern recognition using the image data transferred to the shift register 13 is performed and interpolation is performed. Perform interpolation calculation to calculate the value.

The interpolation value obtained by the interpolation calculation circuit 14 is temporarily held in the interpolation value register 15. That is, since the interpolation value register 15 uses the past interpolation pixel as data when performing the interpolation calculation of the next pixel, it is a data register for temporarily holding the data of the past interpolation pixel.

The data of the interpolated pixel held in the interpolated value register 15 is overwritten in the storage area of the memory 10a or 10b where the pixel data for which the interpolation calculation is completed is stored for the reason described later. .

After that, the pixel data used for the interpolation calculation and the data of the overwritten interpolation pixel are sequentially read from the memories 10a and 10b and sent to the transfer control circuit 7. The transfer control circuit 7 controls the timing at which data is transferred to the structure for printing of the next stage via the terminal 8.

Here, in the image processing apparatus of this embodiment, the thermal head has two scanning lines in each of the main scanning direction and the sub-scanning direction.
It enables 4x image enlargement. In order to realize it using the line buffer memories 10a and 10b for two lines, the data input form is determined as follows.

That is, at the time of double enlargement in the main scanning direction,
As shown in Table 1, the sequence in the input line data is 2
Input line data that has the same data dot by dot and is arranged. Further, even when the image is magnified four times in the main scanning direction, as shown in Table 2, the line data in which the lines in the input line data are arranged in the same data every 4 dots is input.
Further, these Table 1 and Table 2 show 2 in the main scanning direction.
It also shows the relationship between the input form at the time of double or quadruple enlargement and the interpolated value used to form the enlarged image.

[0026]

[Table 1]

[0027]

[Table 2]

As can be seen from Table 1 above, in the case of double enlargement in the main scanning direction, for example, the data in one line of the original image is A, B, C, D, E, F, ... At one time, as input data, data (A, A, B, B, C, C, D, D,
E, E, F, F, ...) are used to interpolate the data A and B, B and C, C and D, D and E, E and F, ... (A + B) / 2. , (B + C) / 2, (C + D) /
Output data obtained by interpolating 2, (D + E) / 2, (E + F) / 2, ... Between the respective data of A, B, C, D, E, F ,. By doing so, image enlargement twice in the main scanning direction is realized.

Further, as can be seen from Table 2 above, in the case of 4 times enlargement in the main direction, the data (A, B, C, D, E, F, ...) Within one line of the original image. On the other hand, as input data, data (A, A, A, A, B, B, B, B, C,
C, C, C, ...) and data A and B, respectively.
(3A + B) / 4, (A
+ B) / 2, (A + 3B) / 4, (3B + C) / 4,
(B + C) / 2, (B + 3C) / 4, ...
By interpolating between the respective data of A, B, C, D, E, F, ... In the line, the output data is used to realize image enlargement of 4 times in the main scanning direction.

On the other hand, as shown in Table 3, 2 in the sub-scanning direction
In the case of double expansion, the order of the input line data is continuously input so that two lines are the same line, and in the case of four times expansion, four lines are continuously input so that they are the same line. To do. In addition, Table 3 also shows the relationship between the input form at the time of magnification of 2 or 4 times in the sub-scanning direction and the interpolated value used for forming the magnified image.

[0031]

[Table 3]

As can be seen from Table 3, in the case of double enlargement in the sub-scanning direction, the original image line order is, for example, I ₀ , I ₁ , I ₂ , I ₃ , I ₄ ,. In this case, the input line data is data (In (I ₀ ), In (I ₀ ), In) in which the same line is arranged every two lines.
(I ₁ ), In (I ₁ ), ..., Using the lines In (I ₀ ) and In (I ₁ ), In (I ₁ ) and In, respectively.
_{(I 2), In (I} 2) and _{In (I 3), In (} I 3)
And In (I ₄ ), In (I ₄ ) and In (I ₅ ), ...
Interpolation calculation described later is performed for the obtained interpolation values (I ₀ I ₁ ), (I ₁ I ₂ ), (I ₂ I ₃ ), and (I _{3
I 4), (I 4 I} 5), the a ... Each line I _0, I
Output line data (Out (I ₀ ), Out (I ₀ ), Out) obtained by interpolating between ₁ , I ₂ , I ₃ , ...
(I ₀ I ₁ ), Out (I ₁ ), Out (I ₁ I ₂ ),
Out (I ₂ ), ..., That realizes twice the image enlargement in the sub-scanning direction.

As can be seen from Table 3, sub-scanning
In the case of 4 times magnification in the direction, the order of the lines of the original image is
I₀, I₁, I₂, I₃, I_FourWhen ...
As the force line data, the same line is not 4 lines.
Data (In (I₀), In (I₀), In
(I₀), In (I₀), In (I₁), In
(I₁), In (I₁), In (I₁), In
(I₂), ...) are used for each line In
(I₀) And In (I₁), In (I₁) And In
(I₂), ...
Is the interpolated value (I₀I₁)₁, (I₀I₁) ₂,
(I₀I₁)₃, (I₁I₂)₁, (I₁I₂)₂,
(I₁I₂)₃, ... Above each line I₀, I₁, I
₂, Output line data (O
ut (I₀), Out (I₀), Out (I₀), Ou
t (I₀), Out (I₀I₁)₁, Out (I
₀I₁)₂, Out (I₀I₁)₃, Out (I₁),
Out (I₁I₂)₁, Out (I₁I₂)₂, Out
(I₁I₂)₃, ...)
The image is magnified 4 times.

Next, a method of actually using the buffer memories 10a and 10b for two lines to realize image enlargement of 2 or 4 times in the main scanning direction and the sub scanning direction shown in Table 3 will be described. .

Two-line buffer memories 10a and 10b
Whether or not the image can be magnified twice or four times in the main-scanning direction and the sub-scanning direction depends mainly on how to enlarge in the sub-scanning direction. Therefore,
First, a method of enlarging in the sub-scanning direction will be described. In addition,
A method for actually realizing the image enlargement twice or four times in the main scanning direction will be described later.

First, a description will be given starting from the case where the magnification is doubled in the sub-scanning direction.
It As shown in Table 4 corresponding to Table 3 above, for example, input
Line data (I₀) And (I₁) Is entered twice each
Suppose In addition, (I in Table 4₀I₁) Is (I₀)When
(I₁Interpolation line between₁I₂) Is (I₁)
And (I₂Interpolation line between₂I₃) Is
(I ₂) And (I₃) Shows the interpolation line. Also,
Fig. 2 shows the print pulse when the image is magnified twice in the sub-scanning direction.
And the print data write clock and the image of each line.
Image data, data stored in memories 10a and 10b
Data, the transfer data of the transfer control circuit 7, and the output from the output terminal 8.
The force data is shown.

[0037]

[Table 4]

In Table 4 and FIG. 2, the input line
Data In (I₀) Is input twice and each is stored in memory 1
0a and 10b, then input line data In
(I ₁) Is input to the memory 10a.

Here, in order to obtain a double enlarged image in the sub-scanning direction, the line data (I ₁ ) stored in the memory 10a is sent to the structure of the next stage, and the line data desired next is ( The line data is between I ₀ ) and (I ₁ ). Therefore, in order to calculate the line data, interpolation calculation is performed between the line data (I ₀ ) and (I ₁ ),
Interpolation line data Ho (I ₀ I ₁ ) must be calculated. The process proceeds as follows.

The line data (I₀) Is now entered
Line data (I₁) Required to perform interpolation calculation with
When the number of pixels is complete, it is stored in the memory 10b.
Line data (I₀) And memory 10a
Line data (I₁) And the interpolation calculation are started.
Interpolation line data Ho (I
₀I ₁) Is generated, and this line data is stored in the memory 10a.
Will be overwritten. Then, overwrite the memory 10a.
Line data Ho (I₀I₁) Is transferred to the next stage
Made with data.

The line data (I ₁ ) to be output next is the third input line data (I ₁ ).
Is output without performing arithmetic processing. From this point onward, two lines are input for each of I ₂ , I ₃ , ..., By performing the same processing as above,
Image enlargement in the sub-scanning direction is realized. However, as can be seen from Table 4 and FIG. 2, only the first two lines (I ₀ ) are output as they are because the other party line (I ₁ ) for which the interpolation calculation is performed is not yet input.

Next, quadruple enlargement in the sub-scanning direction will be described. As shown in Table 5 corresponding to Table 3 above, it is assumed that the input line data (I ₀ ) and (I ₁ ) are input four times, respectively. In Table 4, (I ₀ I ₁ ) ₁ , (I
₀ I ₁ ) ₂ and (I ₀ I ₁ ) ₃ indicate interpolation lines between (I ₀ ) and (I ₁ ). In FIG. 3, the print pulse, the print data write clock, the image data of each line, and the memory 1 when the image is magnified four times in the sub-scanning direction.
The data stored in 0a and 10b, the transfer data of the transfer control circuit 7, and the output data from the output terminal 8 are shown.

[0043]

[Table 5]

In Table 5 and FIG. 3, the input line data In (I ₀ ) is input four times, and the data is stored in the memory 1 respectively.
It is assumed that the input line data In (I ₁ ) is input to the memory 10a after being input to and output from 0a and 10b.

Here, an enlarged image of 4 times in the sub-scanning direction is obtained.
To obtain the line data I₀And I ₁Line day between
(I₀I₁)₁, (I₀I₁)₂, (I₀I₁)₃so
is there. The line data (I₀I₁)₁To calculate
First, line data (I₀I ₁)₂And then
Line data (I₀) And (I₀I₁)₂Interpolated between
Line data (I₀I₁)₁Must be calculated
Must. The process proceeds as follows.

That is, when the number of pixels necessary for performing the interpolation calculation between the line data (I ₀ ) and the line data (I ₁ ) that is just input is complete, the line stored in the memory 10b is stored. The interpolation calculation is started between the data (I ₀ ) and the line data (I ₁ ) input to the memory 10a. By this interpolation calculation, the interpolation line data Ho (I
₀ I ₁ ) ₂ is generated, and this interpolation line data is overwritten in the memory 10a. At this time, what is needed is not the above line data Ho (I ₀ I ₁ ) ₂ but the line data (I ₀ I ₁ ) _1, so that the line data Ho (I ₀ I ₁ ) ₂ stored in the memory 10a is further continued. When the number of pixels required for the calculation becomes equal between the line data (I ₀ ) held in the memory 10b and the line data (I ₀ ), the interpolation calculation is restarted. As a result, the interpolation line data Ho (I ₀ I ₁ )
_{When 1} is generated, the generated Ho (I ₀ I ₁ ) ₁
Each pixel data of is sequentially overwritten again in the memory 10a. After that, the line data (I ₀ I ₁ ) ₁ stored in the memory 10a is used as the transfer data to the next stage.

As described above, in the case of the 4-fold enlargement in the sub-scanning direction, the memory 1 is used to calculate one line to be output.
0a is overwritten twice. That is, for the memory 10a, the interpolated line data Ho (I ₀ I ₁ ) ₂ is placed on the line data (I ₀ ) and the interpolated line data Ho (I ₀ I ₁ ) ₂ is placed again on the interpolated line data Ho (I ₀ I ₁ ) _2. Ho
(I ₀ I ₁ ) ₁ is overwritten.

Further, the line data Ho (I ₀ I ₁ )
Line data Ho must be output after ₁
(I ₀ I ₁ ) ₂ and the line data Ho (I ₀ I ₁ ) ₃ that must be output next is (I ₀ I ₁ ) ₃ .
How these are made is explained below.

The line data (I
_{When 1} ) is input to the memory 10a, interpolation calculation is started with the line data (I ₀ ) stored in the memory 10b, and interpolated line data Ho (I ₀ I ₁ ) ₂ is generated. The interpolation line data Ho (I ₀ I ₁ ) ₂ is overwritten in the memory 10b.

Then, the sixth line is repeated again.
Data (I₁) Is input to the memory 10a,
Interpolation line data Ho (I₀I
₁) ₂And line data stored in the memory 10a
(I₁) And an interpolation calculation is performed between
In Data Ho (I₀I₁)₃Is generated. And
This interpolation line data Ho (I₀I₁)₃Is the above memory
10b is overwritten.

In this way, the interpolation line data Ho (I ₀
I ₁ ) ₂ and Ho (I ₀ I ₁ ) ₃ are sequentially generated and output.

Next, the line data (I
₁ ), so the 7th line, line data (I
Output without performing arithmetic processing on ₁ ). From this point onward, I ₂ , I ₃ , ... Are input by 4 lines each, and therefore, by performing the same processing as described above, it is possible to realize image enlargement of 4 times in the sub-scanning direction. Even in the case of the 4 × enlargement in the sub-scanning direction, as can be seen from Table 5 and FIG. 3, only the first 4 lines (I ₀ ) have the other line (I ₁ ) on which the interpolation is performed. Since it is not input, it is output as it is.

On the other hand, the enlargement in the main scanning direction is performed when transferring to the next stage after the enlargement in the sub scanning direction is completed.

That is, as shown in Table 1, in the case of double enlargement in the main scanning direction, every other dot is used, and the pixels in between are replaced with the average value of 2 dots. Further, as shown in Table 2, in the case of 4 times enlargement in the main scanning direction, two dots every three dots are used and three pixels in between are replaced by (3A + B) / 4 and (A + 3B) / 4.

Here, when the image is enlarged in the main scanning direction, the simple average value as described above is adopted because the main scanning direction and the scanning line direction are the same. Further, the image data is obtained by sampling an analog signal, and the simple average is adopted because the frequency characteristics of the analog signal are not high. The outline will be described with reference to FIG.

That is, in FIG. 4, it is assumed that, for example, an analog signal is sampled at the position of ◯ in the drawing. To double this, the sampling value at the ● position in the figure is required. When trying to infer this value, it can be seen that the analog signal frequency characteristics are not high, so that a value close to the average value can be obtained. Therefore, when enlarging in the main scanning direction, the above-mentioned simple average should be used. I have to.
It should be noted that in the case of an analog signal, it takes about 220 ns even for a sharp rise, and sampling is performed at 60 or 7
It takes about 0 ns.

Next, the interpolation calculation used for the image enlargement in the sub-scanning direction will be described. In the following explanation,
An example is given in which the image is doubled in the sub-scanning direction.

Whether the input control circuit 6 is a frame memory
The input library read from the memory and sent to the memories 10a and 10b.
The sequence of in data is as follows. In addition, the following
I indicates the i-th line of the original input signal. Also,
A_j, B_j, C_j, D_j・ ・ ・ ・ ・ ・, X_j, Y _j, Z_jIs the original input signal
Each pixel value in any one line is shown.

Input line Image line data of original signal ········ n-2nd line A _i-1 B _i-1 C _i-1 D _i-1 ... X _i-1 Y _i-1 Z _i-1 nth line A _i-1 B _i-1 C _i-1 D _i-1 ... X _i-1 Y _i-1 Z _i-1 nth line A _i B _i C _i D _i・ ・ ・ X _i Y _i Z _i nth line A _i B _i C _i D _i・ ・ ・ X _i Y _i Z _i n + 2 line A _{i + 1} B _{i + 1} C _{i + 1} D _{i + 1} ...・ X _{i + 1} Y _{i + 1} Z _{i +} 1th n + 3 line A _{i + 1} B _{i + 1} C _{i + 1} D _{i + 1} ... X _{i + 1} Y _{i + 1} Z _{i + 1} ...・ ・ ・

That is, for example, the memories 10a, 10
When the (n-2) th line is input as an input line to one of the memories b, the n-1th line, which is the same data, is subsequently input to the other memory. Similarly, when the data of the nth line is input to one memory, the same data of the (n + 1) th line is input to the other memory, and when the n + 2th data is input to one memory, the other memory is input. In order that the data of the (n + 3) th line which is the same data is input to the memory 10a,
The same image data is input to 10b.

Here, it is assumed that, for example, the data of the (n-1) th line is already held in the memory 10a and the data of the nth line is transferred to the memory 10b.

When the image data of the n-th line is input to the memory 10b and the number of data equal to or larger than the number of data which can be interpolated starts to be stored in the memory 10b, the work table transfer control circuit 9 transfers the data required for the operation to the memory 10a. 10b to the work table shift register 13.

[0063] That is, the shift register 13 consists of two shift registers 13a and 13b, the n-1 pixel data registers a _m line from the memory 10a on one of the shift register 13a, a _{m + 1} , A _{m + 2} , a
_{m + 3, a m + 4} , a m + 5, a m + 6 are stored are shifted sequentially to the other shift register 13b the n-th pixel data registers b _m of the line from the memory 10b to, b
_The data are sequentially shifted and stored in _{m + 1} , b _{m + 2} , b _{m + 3} , b _{m + 4} , b _{m + 5} , b _{m + 6} .

When a predetermined number of pixel data of the (n−1) th line and the nth line are prepared in the shift register 13, each pixel data is output from the shift register 13 and sent to the interpolation calculation circuit 14.

In the interpolation calculation circuit 14, pattern recognition described later is performed using the supplied pixel data of the (n-1) th line and the nth line, and a line between the (n-1) th line and the nth line. The interpolated value of is calculated.

The data of this interpolation value is transferred to the memory 10b and overwritten on the portion which is no longer necessary for the interpolation calculation.

Next, the interpolation method used in the interpolation calculation circuit 14 of this embodiment will be described below.

Here, the line buffer memory 10a
And 10b to output the registers a _m , a _{m + 1} , a _{m + 2} , a _{m + 3} , a _{m + 4} , a _{m + 5} , a of the shift register 13 a.
_{m + 6} and each register b _m , b of the shift register 13b
_The data (data of the existing line) stored in _{m + 1} , b _{m + 2} , b _{m + 3} , b _{m + 4} , b _{m + 5} , b _{m + 6} is shifted as shown in FIG. 5, for example. In the register 13a, A _i , B
_i , C _i , D _i , E _i , F _i , and G _i become the data, and in the shift register 13b, A _{i + 1} , B _{i + 1} ,
It is assumed that the data are C _{i + 1} , D _{i + 1} , E _{i + 1} , F _{i + 1} , and G _{i + 1} . In addition, these shift registers 13
Assuming that the data of the interpolated pixel calculated using the data stored in a and 13b is X, X _k in the drawing of FIG. 5 represents the pixel currently interpolated, and X _k-2 is two pixels before X _k. the interpolation pixel, X _k-1 is the interpolation pixel one pixel before the X _k, X _{k + 1} represents the pixel to be interpolated in the following X _k.

Here, in the interpolation calculation circuit 14 of the present embodiment, when the pixel x _k is obtained by the interpolation calculation, each of the 7 pixels as shown in FIG. 5 stored in the two shift registers 13a and 13b (that is, The pattern recognition is performed using the data of each of the three pixels (that is, 3 × 2 pixels) in the two existing lines surrounded by the broken line in FIG. 5 out of the 7 × 2 pixel data.

At the time of this pattern recognition, first, the magnitude relationship between adjacent pixels in the vertical and horizontal directions is obtained. That is, when the pixel X _k is obtained by interpolation calculation, C _i , D _i , E _{i of the} pixel data stored in the shift register 13a and C _{i + 1} , of the pixel data stored in the shift register 13b, Using D _{i + 1} and E _{i + 1} , the size relation of the pixels adjacent in the vertical and horizontal directions is obtained. However, the magnitude relation between D _i and D _{i + 1} is excluded.

At this time, there are 6 combinations of adjacent pixels. Considering 3 combinations of equal sign and inequality sign for each of them, the size of the above 3 × 2 pixel area surrounded by the broken line in FIG. 5 is considered. Combination by relationship (ie pattern)
There are 3 ⁶ types (= 729 types). The interpolation calculation circuit 14 recognizes the 729 patterns and determines whether or not to further expand the recognition area.

At this point, if it is known that the recognized pattern is a pattern in which the amount of information does not change much even if the recognition area is further expanded, or a pattern that cannot be recognized in the recognition area of 7 × 2 pixels at the maximum. , The interpolation value x _k is _obtained from the 3 × 2 pixel recognition area.

On the other hand, if the pattern is judged to be capable of more accurate pattern recognition by further expanding the recognition area of 3 × 2 pixels, the recognition area further expanded than the recognition area of 3 × 2 pixels. (In this case, the maximum is 7 ×
Pattern recognition is performed by (it can be expanded to a recognition area of 2 pixels).

The algorithm of the above interpolation calculation will be schematically described with reference to FIG. In FIG. 6, step S1
First, pattern recognition is performed in the 3 × 2 pixel area. In the next step S2, it is judged whether or not the recognition area should be further expanded. If it is judged that the recognition area should be further expanded (Yes), the processing advances to step S3, and if it is judged not to be expanded (no), the processing advances to step S4.

In step S4, Tables 2 to 10 described later are used.
Pattern recognition is performed in the 3 × 2 pixel area by referring to the table shown in FIG. 3, and in the next step S6, the optimum one is selected from the following three formulas to determine the interpolation calculation. X _k = (C _i + E _{i + 1} ) / 2 X _k = (D _i + D _{i + 1} ) / 2 X _k = (E _i + C _{i + 1} ) / 2

Further, in step S3, Table 6 to
Pattern recognition is performed in the 7 × 2 pixel area with reference to the table shown in Table 14, and in the next step S5, the optimum one is selected from the following eight expressions to determine the interpolation calculation. X _k = (A _i + G _{i + 1} ) / 2 X _k = (B _i + F _{i + 1} ) / 2 X _k = (C _i + E _{i + 1} ) / 2 X _k = (D _i + D _{i + 1} ). / 2 X _k = (D _i + D _{i + 1} + 2X _k-1 ) / 4 X _k = (E _i + C _{i + 1} ) / 2 X _k = (F _i + B _{i + 1} ) / 2 X _k = (G _i + A _{i + 1} ) / 2

That is, in this embodiment, as described above,
When the pattern recognition area has a pattern in which the amount of information does not change much even if the area is widened, an interpolation value is obtained from the recognition area of 3 × 2 pixels by referring to the tables shown in Tables 6 to 14 described later. If the pattern is determined to be more accurate by expanding the pattern recognition area, Table 6 to be described later
By obtaining the interpolation value from the 7 × 2 pixel area with reference to the table shown in Table 14, the scale of the pattern recognition circuit is prevented from expanding while keeping the image quality after interpolation high.

For example, if the entire area of 7 × 2 pixels is set as the pattern recognition area from the beginning, the combination (that is, the pattern) of the magnitude relation between the adjacent pixels is 3 ^18.
(= 387420489), and it is very difficult to determine the interpolation calculation method for each pattern by performing pattern recognition, and it seems that the circuit scale becomes too large. In the embodiment, as described above, first, the recognition is performed in the area of 3 × 2 pixels,
After that, by adaptively expanding the area to a maximum of 7 × 2 pixels, the number of combinations to be considered is reduced. Even when the area is expanded to a 7 × 2 pixel area, the equations for interpolation calculation are only the above eight equations. Is suppressed from increasing. In the circuit of this embodiment, the number of combinations (patterns) is reduced to about 920.

The above-described interpolation calculation will be described below with reference to a more specific example. In the example of FIG. 5, when obtaining the interpolated pixel X _i , the pixel C _i inside the broken line in FIG.
It is determined whether the interpolation direction is diagonal or vertical based on the magnitude relation between the six data D _i , E _i and the six data C _{i + 1} , D _{i + 1} , E _{i + 1} . Tables 6 to 14 show tables showing the interpolation directions determined by this magnitude relationship. These tables show the number of combinations of the above 6 pixels, 3 ⁶ =
It corresponds to 729.

[0080]

[Table 6]

[0081]

[Table 7]

[0082]

[Table 8]

[0083]

[Table 9]

[0084]

[Table 10]

[0085]

[Table 11]

[0086]

[Table 12]

[0087]

[Table 13]

[0088]

[Table 14]

According to these tables, the interpolated pixel X is determined as follows according to the blank mark, the / mark and the \ mark in each table by the pattern recognition of the 3 × 2 pixel area. No mark X = (D _i + D _{i + 1} ) / 2 / mark X = (E _i + C _{i + 1} ) / 2 \ mark X = (C _i + E _{i + 1} ) / 2

However, D is placed beside or above the / and \ marks in the table.
_{When i} > C _{i + 1} , D _{i + 1} > C _i , D _i ≦ E _{i + 1} , ..., It means that the diagonal interpolation is performed only when the condition is satisfied. . Also on the table ~
The portion of indicates a portion where an interpolated image is obtained by pattern recognition of a 7 × 2 pixel area described later.

Further, the rows and columns of the tables in Tables 6 to 14 are respectively numbered from 0 to 26, where the row number is q and the column number is r, and each matrix of the table is coordinated. When expressed as (r, q), interpolation is performed from the data of 7 × 2 pixels for the coordinate portion on the table indicated by the following.

(0,12), (9,12) (14,2), (14,11) (14,23), (14,26) (21,12), (24,12)

(17,14), (26,14) (12,15), (12,24) (12,0), (12,3) (2,14), (5,14)

(26,0), (0,26)

(7, 13), (8, 14), (12,
6), (13, 7), (12, 19), (13, 2)
0), (18,12), (19,13)

(0,23), (9,26), (17,
0), (26,3) (0,17), (3,26), (23,0), (26,
9)

The interpolation algorithm in the area where the pattern recognition is performed by 7 × 2 pixels shown in the coordinate portions of these tables will be described below.

First, the interpolation of the coordinate part in the table
As an example, for example, C_i<D_i<E _i> E_{i + 1}= D_{i + 1}
= C_{i + 1}<C_iIn case of, the large and small size of the flowchart of FIG.
Do another and interpolate. In other words, search for smaller data.
Interpolation is performed. The above C_i<D_i<E_i
> E_{i + 1}= D_{i + 1}= C_{i + 1}<C_iIs (r,
In this example, q) = (24, 12).

That is, in FIG. 7, step S
B in 10_i<C_iAnd E_{i + 1}= F _{i + 1}Whether or not
The judgment is made, and if the judgment is NO, step S11.
If the answer is yes, then go to step S12.

In step S11, X _k = (C _i + E
_{i + 1} ) / 2 is calculated to obtain the interpolated value. Step S
In step 12, if it is determined that A _i <B _i , the process proceeds to step S13 if the result is NO, and to step S14 if the result is YES.

In step S13, X _k = (B _i + F
_{i + 1} ) / 2 is calculated to obtain the interpolated value, and step S1
In step 4, X _k = (A _i + G _{i + 1} ) / 2 is calculated to obtain an interpolated value.

Next, in the interpolation of the portion of, for example, C _i >
When D _i > E _i <E _{i + 1} = D _{i + 1} = C _{i + 1} > C _i , FIG.
Interpolation is performed by determining the size of the flow chart. That is, interpolation is performed by performing the process of searching for the larger data. Note that this C _i > D _i > E _i <E _{i + 1} = D _{i + 1} = C _{i + 1} >
C _i is an example corresponding to the coordinates (r, q) = (2, 14) in Table 9.

That is, in FIG. 8, step S
B at 20_i> C_iAnd E_{i + 1}= F _{i + 1}Whether or not
If the judgment is made and the judgment is NO, step S21.
If the answer is yes, then go to step S22.

In step S21, X _k = (C _i + E
_{i + 1} ) / 2 is calculated to obtain the interpolated value. Step S
At 22, it is judged whether A _i > B _i . If it is judged as no, the process proceeds to step S23, and if it is judged as yes, the process proceeds to step S24.

In step S23, X _k = (B _i + F
_{i + 1} ) / 2 is calculated to obtain the interpolated value, and step S2
In step 4, X _k = (A _i + G _{i + 1} ) / 2 is calculated to obtain an interpolated value.

Here, in the interpolation calculation of the portions of and, both upper and lower limits of the interpolation value X _k are set. That is, D _i <X _k <D _{i + 1} or D _i > X _k > D
_{When the} upper and lower limits are set like _{i + 1} and the interpolation result X _k satisfies this relationship, the interpolation value X _k is adopted as it is.
When this relationship is not satisfied, the innermost diagonal interpolation value is set to X _k . That is, X _k = (E
Let _i + C _{i + 1} ) / 2 be the interpolation result, and X _k = (C _i + E _{i + 1} ) / 2 be the interpolation result at the time of the interpolation calculation of the portion.

Next, in the interpolation calculation of the portion of, for example, C
_i<D_i<E_i> E_{i + 1}<D_{i + 1}<C_{i + 1}> C_itime,
C_iAnd E_{i + 1}Is there a valley of magnitude relation with
_iAnd C_{i + 1}I don't know if there is a ridge of magnitude relation with.
Therefore, interpolation is performed under the following conditions. The interpolation calculation
C above in_i<D_i<E_i> E_{i + 1}<D_{i + 1}<C
_{i + 1}> C_iIn the case of a pattern in which all signs of
Is C_iAnd E_{i + 1}Ridge between, E_iAnd C_{i + 1}Between the valley
Like Also, C_i<D_i<E_i> E_{i + 1}<D_{i + 1}
<C_{i + 1}> C_iOf (r, q) = (26,0) in Table 8
It is an example corresponding to coordinates.

At this time, the judgment as to whether it is a ridge or a valley in the above-mentioned magnitude relation is made based on the following criteria. For example, if the magnitude relationship of A _{i to} E _i is as follows, it is assumed that there is a valley in B _i or C _i .

A _i > B _i <C _i <D _i <E _i , or A
_i > B _i = C _i <D _i <E _i , or A _i > B _i > C _i
<D _i <E _i

Therefore, if there is no magnitude relation between B _i and C _i, and if the magnitude relation between A _i and B _i is A _i > B _i , it is judged that there is a valley.

Here, as the first condition, A _i > B _i ,
And F _{i + 1} <G _{i + 1} , it is determined whether or not there is a valley in B _i or C _i and whether or not there is a valley in E _{i + 1} or F _{i + 1.} When there is _one, Q ₁ = 1 and when there is no valley in either one or both, Q ₁ = 0.

As the second condition, A _{i + 1} <
When B _{i + 1} and F _i > G _i , it is determined whether or not there is a ridge at B _{i + 1} or C _{i + 1} and whether or not there is a ridge at E _i or F _i. If there is a ridge, Q ₂ = 1; if there is no ridge on either or both, then Q ₂ = 0.

At this point, the interpolation method is first determined as follows. If Q ₁ = 1 and Q ₂ = 1 then X _k = (D _i + D _{i + 1} ) /
If 2 Q ₁ = 1 and Q ₂ = 0, then X _k = (C _i + E _{i + 1} ) /
If 2 Q ₁ = 0 and Q ₂ = 1 then X _k = (E _i + C _{i + 1} ) /
2 If Q ₁ = 0 and Q ₂ = 0, the next judgment is made.

Next, as the third condition, when A _i > B _i or F _{i + 1} <G _{i + 1} , there is a valley in B _i or C _i or E _{i + 1} or F _{i + 1.} If there is a valley on either side, set Q ₃ = 1. If there is no valley on either side, Q
₃ = 0.

As the fourth upper limit, A _{i + 1} <
When B _{i + 1} or F _i > G _i , it is determined whether or not there is a ridge in B _{i + 1} or C _{i + 1} or E _i or F _{i. 4} = 1 and when none have ridges, set Q ₄ = 0.

As the fifth condition, C _i = E _{i + 1} , or C
_In the judgment of _i = F _{i + 1} , or B _i = E _{i + 1} , or B _i = F _{i + 1} , when yes, Q ₅ = 1 and when no, Q _{5
Set 5} = 0.

As the sixth condition, E _i = C _{i + 1} , or E
_In the judgment of _i = B _{i + 1} , or F _i = C _{i + 1} , or F _i = B _{i + 1} , when yes, Q ₆ = 1 and when no, Q
₆ = 0.

From the above conditions, the interpolation method is finally determined as follows. That is, Q ₃ = 1 and Q ₄ = 1 and Q
_{If 5} = X _K and Q ₆ = X _K, then X _K = (D _i + D _{i + 1} ) /
2. If Q ₃ = 1, Q ₄ = 0, Q ₅ = X _K , Q ₆ = 1 then X _K = (E _i + C _{i + 1} ) / 2, and Q ₃ = 0, Q ₄
= 1 and Q ₅ = 1 and Q ₆ = X _K , X _K = (C _i + E
_{i + 1} ) / 2, Q ₃ = 0, Q ₄ = 0, Q ₅ = X _K , Q
_{If 6} = X _K, then X _K = (D _i + D _{i + 1} ) / 2. However, Q ₁ = Q ₂ = 0.

Next, in the interpolation calculation of the portion, the pixel to the left of the interpolated pixel X _k is X _k-1 , for example, C _i <D _i > E
_{When i} > E _{i + 1} = D _{i + 1} = C _{i + 1} <C _i , interpolation is performed under the following conditions. Note that C _i <D _i > E _i > E _{i + 1} = D _{i + 1}
= C _{i + 1} <C _i is (r, q) = (18,1) in Table 11.
This is an example corresponding to the coordinates of 2).

For example, as the condition, C _{i + 1} = X _k-1
If yes, Q ₇ = 1 and if no, Q ₇ = 0. The interpolation method at this time is X _k = (D _i + D _{i + 1} ) / 2 if Q ₇ = 0 and X if Q ₇ = 1.
_{Let k} = (D _i + C _{i + 1} + 2X _k−1 ) / 4.

Next, in the interpolation of the part, D _i and D _{i + 1}
The interpolation pixel between an X _k, for example, C _i = D _i <E _i
When> E _{i + 1} <D _{i + 1} <C _{i + 1} > C _i , interpolation is performed under the following conditions. Note that C _i = D _i <E _i > E _{i + 1} <D _{i + 1} <C
_{i + 1} > C _i is an example corresponding to the coordinates (r, q) = (17,0) in Table 7.

For example, as the condition, C _i = E _{i + 1} ,
Or C _i = F _{i + 1} , or B _i = E _{i + 1} , or B _i = F
_In the judgment of _{i + 1} , if yes, Q ₈ = 1 and if no, Q ₈ = 0. The interpolation method at this time is Q ₈ =
If 1, then X _k = (C _i + E _{i + 1} ) / 2, and if Q ₈ = 0, then X _k = (D _i + D _{i + 1} ) / 2.

Further, as a condition at the time of each size discrimination described above, the size discrimination of two data is performed under the following conditions. That is, if one data is Y and the other data is W, then Y = W if INT ((Y-W) / L) = 0 and INT ((Y-W) / L)> 0. If there is Y> W, if INT ((Y−W) / L) <0, then Y <W. Note that INT is a built-in function.

Here, the range of ± (L-1) depending on L is regarded as an equal sign. For L, use 1 or 4 or 8 or 16. The value of L is changed according to the SN ratio required by the image, and the smaller the value becomes, the more sensitive the value becomes.
The reference to the area of × 2 pixels is reduced.

The following effects can be obtained by using the algorithm of the interpolation calculation of the above-mentioned parts (1) to (3).

First, by using the interpolation calculation algorithm of the portions (1) and (2), when two areas have diagonal boundaries, interpolation can be performed so that the boundaries become smooth.

For example, as a first specific example of the image before and after the interpolation in FIG. 9, the effect of the interpolation calculation of the above portion will be described. Further, each cell of FIG. 9 corresponds to each pixel data of FIG. 5, and the pixel data of each cell of FIG. 9 is also an example of pixel data stored in each register of the shift register 13 of FIG. . Even in the interpolation calculation of the above portion, the same effect can be obtained by performing the interpolation calculation according to the tables of Tables 6 to 14.

As a first specific example of this, as shown in FIG.
There is an image of the existing line (image before interpolation) as shown
Suppose At this time, each image of the existing line in FIG.
The color density of the element is expressed in hexadecimal notation.
In each pixel A_i, B_i, C _i, D_i, E_i, F_i, G
_iThe actual data of color density is (10) (50) (70)
(90) (F0) (F0) (F0), which is the existing lower side
Each pixel A in the line_{i + 1}, B_{i + 1}, C_{i + 1}, D_{i + 1}, E
_{i + 1}, F_{i + 1}, G_{i + 1}The actual data of the color density of (1
0) (10) (10) (10) (10) (10) (1
0). In addition, (10) represents a dark color,
The color becomes lighter as it goes to (F0).

That is, the positional relationship between the interpolated pixel X _k shown in FIG. 5 and the pixels of each existing line is as follows: existing line (10) (50) (70) (90) (F0) (F0) ( F0) Interpolation line X _k Existing line (10) (10) (10) (10) (10) (10) (10) If the interpolation line X _k is 3 × 2 pixels above and below the existing line C _i , D _i , E _i and C _{i + 1} ,
Regarding the magnitude relationship between D _{i + 1} and E _{i + 1} , the three pixels in the upper existing line are (70) (90) (F0) in order from the left, and the three pixels in the lower existing line are (10) (10). ) Since it is (10), (70) <(90) <(F0)>
(10) = (10) = (10) <(70).

In this case, the pattern recognition area is enlarged,
Recognize the size relationship of the remaining pixels. That is, in FIG.
In case of (a), A_i, B_i, C_iAnd E_i, F_i, G
_i, And A_{i + 1}, B_{i + 1}, C_{i + 1}And E_{i + 1}, F_{i + 1}, G
_{i + 1}The size relationship of_i<B _i<C_iAnd E_i= F_i= G
_i, And A_{i + 1}= B_{i + 1}= C_{i + 1}And E_{i + 1}= F_{i + 1}= G
_{i + 1}Then, from the matrix of 7 × 2 pixels, X_k= (A
_i+ G_{i + 1}) / 2 interpolation formula is determined.

That is, the example of FIG. 9A is an example corresponding to the coordinates (r, q) = (24, 12) in Table 11 as an example of the interpolation calculation of the above-mentioned portion, C _i <D _i <E _i > E _{i + 1} = D _{i + 1} = C _{i + 1} <C _i
Has become. Therefore, the size of the above-described flowchart of FIG. 7 is determined and interpolation is performed.

Specifically, toward the smaller data
Find the pattern, B_i<C_iAnd E _{i + 1}= F_{i + 1}Is
It is judged whether or not (step S10 in FIG. 7)
In the example of 9 (a), B_iIs (50) and C_iBut (7
E in 0)_{i + 1}And F_{i + 1}Were both (10)
Therefore, (50) <(70) and (10) = (10)
Because it is judged as yes, further A_i<B_iThe judgment of
Perform (step S12 of FIG. 7). In this judgment, the figure
In the example of 9 (a), A_iBut in (10) B_iIs (50)
Therefore, in step S12 of FIG.
Determined, and therefore X in step S14 of FIG._k=
(A_i+ G_{i + 1}) / 2 interpolation calculation is performed. This allows
X_k= (10 + 10) / 2 and the interpolation value (this
If it is (10)).

The interpolation pixel X in FIG._kInterpolation pixel other than X
_k-3, X_k-2, X_k-1, X_{k + 1}, X _{k + 2}, X_k-3Against
The interpolation calculation according to the tables of Tables 6 to 14 above.
By performing this, the image before interpolation in FIG.
You can get the interpolated image as shown in (b).
Swell.

That is, if the interpolation calculation of this embodiment is performed, it is possible to obtain an enlarged image in which the image of the diagonal line of the image before interpolation is more smoothly interpolated. Further, according to the interpolation calculation of the present embodiment, it seems appropriate that the interpolated value is between the pixel values of the upper and lower existing lines, so that the value of the interpolated value is limited to the range of the upper and lower pixel values. Therefore, the noise can be reduced. For example, in the case of a color image, noise such as cyan and magenta is reduced at the edge of the image.

Next, by using the interpolation calculation algorithm of the portions (1) and (2), the image of the diagonal line of the image before interpolation can be interpolated more smoothly.

For example, as a second specific example of the image before and after the interpolation in FIG. 10, the effect of the interpolation calculation of the above portion will be described. Further, FIG. 10 also shows the same as FIG. 9 described above. Even in the interpolation calculation of the above portion, the same effect can be obtained by performing the interpolation calculation according to the tables of Tables 6 to 14.

As the second specific example, FIG.
There is an image of the existing line (image before interpolation) as shown in
Suppose At this time, the existing line of FIG.
The color density of each pixel is expressed in hexadecimal notation.
Each pixel A of the existing line_i, B_i, C_i, D_i, E_i,
F_i, G_iThe actual data of color density of (10) (10)
(10) (70) (F0) (70) (10) and below
Each pixel A of the existing line on the side _{i + 1}, B_{i + 1}, C_{i + 1}, D
_{i + 1}, E_{i + 1}, F_{i + 1}, G_{i + 1}Actual data of color density of
Is (10) (70) (F0) (70) (10) (10)
Let (10).

That is, the positional relationship between the interpolated pixel X _k shown in FIG. 5 and the pixels on each existing line is as follows: (10) (10) (10) (70) (F0) (70) ( 10) Interpolation line X _k existing line (10) (70) (F0) (70) (10) (10) (10) If this is the interpolation pixel X _k , the upper and lower existing lines 3 × 2 pixels C _i , D _i , E _i and C _{i + 1} ,
The magnitude relationship between D _{i + 1} and E _{i + 1} is (10) (70) (F0) from the left in the upper three existing lines, and (F0) (70) from the lower three existing lines. ) (10), so (10) <(70) <(F0)>
(10) <(70) <(F0)> (10).

In this case, the example of FIG.
As an example of the interpolation calculation of the above-mentioned part,
This is an example corresponding to the coordinates of (r, q) = (26, 0),
C_i<D _i<E_i> E_{i + 1}<D_{i + 1}<C_{i + 1}> C_iTona
ing. Therefore, as described in the interpolation of the part above
As described above, the interpolation calculation is performed according to each condition.

Specifically, the size of the pixels of each existing line
The clerk determines whether it is a ridge or a valley. Of this FIG.
In the example of (a), each pixel value of the upper existing line is (1
0) (10) (10) (70) (F0) (70) (1
0) and therefore A_i= B_i= C_i<D_i<E
_i> F_i> G_iNext, E_iIs a ridge. Also, the lower side
The pixel values of the existing line of (10) (70) (F0)
(70) (10) (10) (10)
A_{i + 1}<B_{i + 1}<C_{i + 1}> D_{i + 1}> E_{i + 1}= F_{i + 1}
= G _{i + 1}And C_{i + 1}Is a ridge.

That is, under the above-mentioned first condition, since there is no valley in B _i or C _i of the upper existing line and E _{i + 1} or F _{i + 1} of the lower existing line, Q ₁ = 0, Further, under the condition of FIG. 5 described above, Q ₂ = 1 indicating that both the upper and lower existing lines B _{i + 1} or C _{i + 1} and E _i or F _i have a ridge, and therefore, As described above, as the interpolation method at the time point, since Q ₁ = 0 and Q ₂ = 1 are used, the interpolation calculation of X _k = (E _i + C _{i + 1} ) / 2 is performed.

As a result, the calculation of X _k = (F0 + F0) / 2 is performed to obtain the interpolated value ((F0) in this case).

In the example of FIG. 10, since the interpolation calculation expression can be obtained up to the second condition, the third condition to the sixth condition are not judged.

[0144] interpolation pixel X _k-3 other than the interpolation pixel X _k of FIG _{10, X k-2, X} k-1, X k + 1, X k + 2, X k-3
In contrast, by performing the interpolation calculation according to the tables in Tables 6 to 14, the interpolated image as shown in FIG. 10B can be obtained from the image before interpolation in FIG. Will be able to.

Next, the effect of the algorithm of the interpolation calculation of the above portion will be described with reference to FIG. 11 as a third concrete example. 11B and 11C are also shown similarly to FIGS. 9A and 9B described above. Also, this figure 1
(A) of 1 shows the entire image 100, an example image 102 therein, and the image area 101 stored in the shift register 13. That is, in FIG.
For example, an example of performing interpolation on the vertex portion of the triangular image 100 will be described.

In the third concrete example, as shown in FIG.
There is an existing line image (image before interpolation) as shown
Will be. At this time, the existing line of FIG.
The color density of each pixel of
In each pixel A_i, B_i, C _i, D_i, E_i, F_i, G
_iThe actual data of color density is (F0) (F0) (F0)
(F0) (F0) (F0) (F0), which is the lower existing
Each pixel A in the line_{i + 1}, B_{i + 1}, C_{i + 1}, D_{i + 1}, E
_{i + 1}, F_{i + 1}, G_{i + 1}The actual data of color density of (F
0) (F0) (70) (10) (70) (F0) (F
0).

That is, the positional relationship between the interpolated pixel X _k shown in FIG. 5 and the pixels of each existing line is the existing line (F0) (F0) (F0) (F0) (F0) (F0) ( F0) Interpolation line X _k Existing line (F0) (F0) (70) (10) (70) (F0) (F0) If the interpolation pixel X _k is 3 × 2 pixels above and below the existing line C _i , D _i , E _i and C _{i + 1} ,
Regarding the magnitude relationship between D _{i + 1} and E _{i + 1} , the three pixels in the upper existing line are (F0) (F0) (F0) in order from the left, and the three pixels in the lower existing line are (70) (10). ) (70), so (F0) = (F0) = (F0)>
(70)> (10) <(70) <(F0).

That is, in the example of FIG. 11B, as an example of the interpolation calculation of the above-mentioned portion, C _i = D _i =
It shows the case of E _i > E _{i + 1} > D _{i + 1} <C _{i + 1} <C _i , and corresponds to the coordinates (r, q) = (12, 6) in Table 7 above. Therefore, the interpolation calculation is performed in accordance with each condition as described in the above-mentioned interpolation of the part.

[0149] In this case, pattern recognition region to expand, and the interpolating pixel X _k-1 and C _i for one place to the left of the previous is a need in the interpolation operation of the interpolation pixel X _k, C _i = X _{k -1} judgment is made. In the example of FIG. 11B, since C _i = X _k−1 , the above Q ₇ = 0, and the interpolation calculation at this time is X _k = (D _i + D _{i + 1} + 2X as described above. _k-1 )
It becomes / 4.

[0150] interpolation pixel X _k-3 other than the interpolation pixel X _k of FIG _{11, X k-2, X} k-1, X k + 1, X k + 2, X k-3
In contrast, by performing the interpolation calculation according to the tables in Tables 6 to 14, the interpolated image as shown in FIG. 11C is obtained from the image before interpolation in FIG. 11B. Will be able to. That is, when only simple interpolation is performed as in the conventional interpolation method, the apex of (a) of FIG. 11 is extended and further emphasized so that it looks like a so-called mustache. According to the interpolation calculation of the present embodiment, the density at the extension of the apex is blurred and thinned so that the whisker-like image becomes inconspicuous.

Next, a specific example of performing the interpolation calculation of 3 × 2 pixels according to the tables of Tables 6 to 14 will be described.

That is, as a fourth specific example, FIG.
Image of existing line as shown in (a) (image before interpolation)
Is present. Note that FIG. 12 is similar to FIG. 9 described above.
It is shown in. At this time, the existing line of FIG.
Represents the color density of each pixel in hexadecimal notation, and
Each pixel A on the existing line_i, B_i, C_i, D_i, E_i, F
_i, G_iThe actual data of color density is (F0) (90)
(60) (10) (60) (90) (F0) and below
Each pixel A of the existing line on the side_{i + 1}, B_{i + 1}, C_{i + 1}, D
_{i + 1}, E_{i + 1}, F_{i + 1}, G_{i + 1}Actual data of color density of
Is (F0) (90) (60) (10) (60) (90)
(F0).

That is, the interpolated image represented in correspondence with FIG.
Element X_kAnd the pixel relationship between each existing line and the existing line (F0) (90) (60) (10) (60) (90) (F0) Interpolation line X_k If the existing line is (F0) (90) (60) (10) (60) (90) (F0), this interpolation pixel X_kUp and down against
3 x 2 pixels C of existing line_i, D_i, E_iAnd C_{i + 1},
D_{i + 1}, E_{i + 1}The size relationship is the 3 lines of the existing line above
Element is (60) (10) (60) from left to right
3 pixels of the existing line of (60) (10) (60)
From that, C_i> D_i<E_i= E _{i + 1}> D_{i + 1}
<C_{i + 1}= C_iIt has become.

In this case, the pattern recognition area is not expanded, and the interpolation formula is determined from the matrix table of 3 × 2 pixels in Tables 6 to 14. Therefore, from the tables of Table 6 to Table 14, the above C _i > D _i <E _i = E _{i + 1} > D _{i + 1} <C
The interpolation calculation formula in the case of _{i + 1} = C _i is X _k = (D _i + D
_{i + 1} ) / 2 is determined, whereby _Xk = (10 + 1
0) / 2 = 10 is calculated.

For interpolation pixels X _k-3 , X _k-2 , X _k-1 , X _{k + 1} , X _{k + 2} , X _k-3 other than the above interpolation pixel X _{k in} FIG. By performing the interpolation calculation according to the tables of Table 6 to Table 14, the image before the interpolation in FIG.
It is possible to obtain an image after the interpolation as shown in (b) of 2.

The size of the recognition area when the pattern recognition area is enlarged can be made larger than the 7 × 2 pixel area. For example, when expanding the area in the direction of increasing the number of lines, for example, 7 × 4 pixels,
7 × 6 pixels can be taken as an example, and in this case, better interpolation can be performed. For example, it is an L-shaped area.

In the interpolation calculation method as described above, pattern classification is performed in accordance with a predetermined table, and the size of the area for pattern classification is adaptively controlled based on the pattern classification result. When the area of classification can be made small, this pattern classification can be easily performed. When the area of pattern classification is small, the quality of the image obtained by interpolation deteriorates. It is possible to improve quality. Therefore, by using this interpolation operation when enlarging an image, the circuit scale can be reduced, and even if the image is a slanted image, a sufficiently enlarged image is obtained in terms of gradation, resolution, and jerkiness. It is possible to obtain Furthermore, the image quality in still image output can be expected to be significantly improved.

[0158]

According to the image signal processing method and apparatus of the present invention, a plurality of pixel data in a region consisting of a plurality of columns of the original image signal in the horizontal and vertical directions, which are adjacent to each other, are predetermined. Since the pattern classification is performed according to the table in Table 1 and the size of the pattern classification area is adaptively controlled based on the pattern classification result, pattern classification can be easily performed when the classification area can be reduced. As a result, the amount of interpolation calculation is reduced, and if the pattern classification area is enlarged, the quality of the image that is interpolated and enlarged can be improved. Therefore, by using this interpolation calculation when enlarging an image, it is possible to reduce the circuit scale and prevent the cost from increasing, and even for an oblique image, gradation, resolution,
It is possible to obtain a sufficiently enlarged image in terms of jerkiness. In addition, the image quality in still image output can be significantly improved.

Further, the interpolation data by the enlarged image signal forming means is overwritten on the portion of the w line buffers in which the data for which the interpolation operation has been completed is stored, so that it is not necessary to provide a separate memory for the interpolation data. It is possible to reduce the circuit scale. Further, according to the present invention, it is possible to obtain a sufficiently enlarged image in terms of gradation, resolution, and jerkiness even with an oblique image.

[Brief description of drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of an image signal processing device according to an embodiment of the present invention.

FIG. 2 is a timing chart for explaining the flow of line data when performing double enlargement in the sub-scanning direction.

FIG. 3 is a timing chart for explaining the flow of line data when performing 4 times enlargement in the sub-scanning direction.

FIG. 4 is a diagram used for explaining the reason for adopting a simple average when enlarging in the seed scanning direction.

FIG. 5 is a diagram for explaining data of a work table.

FIG. 6 is a flowchart of an interpolation algorithm.

FIG. 7 is a flowchart showing an example of an interpolation algorithm when performing interpolation calculation by pattern recognition of 7 × 2 pixels.

FIG. 8 is a flowchart showing another example of an interpolation algorithm when performing interpolation calculation by pattern recognition of 7 × 2 pixels.

FIG. 9 is a diagram showing an image of a first specific example for explaining an interpolation calculation by pattern recognition of 7 × 2 pixels.

FIG. 10 is a diagram showing an image of a second specific example for explaining an interpolation calculation by pattern recognition of 7 × 2 pixels.

FIG. 11 is a diagram showing an image of a third specific example for explaining an interpolation calculation by pattern recognition of 7 × 2 pixels.

FIG. 12 is a diagram showing an image of a fourth specific example for explaining an interpolation calculation by pattern recognition of 3 × 2 pixels.

[Explanation of symbols]

 1 Input Control Circuit 7 Transfer Control Circuit 9 Work Table Transfer Control Circuit 10a, 10b Line Buffer Memory 13 Shift Register 14 Interpolation Calculation Circuit 15 Interpolation Register

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H04N 5/91

Claims (9)

[Claims] 1. An image signal processing method for forming an enlarged image signal by enlarging an original image from an original image signal, the method comprising: Pattern classification is performed on a plurality of pixel data in accordance with a predetermined table, the size of the predetermined area is adaptively controlled based on the pattern classification result, and the pattern classification is performed to adaptively control the size. Interpolation calculation according to the pattern classification result is performed using the multiple pixel data in the specified area, and the data of the interpolation pixel by the interpolation calculation is inserted between the data of each pixel of the original image signal to obtain the enlarged image signal. An image signal processing method, comprising: 2. An area formed by a column composed of a plurality of adjacent pixels of the original image signal and a column composed of a plurality of interpolated pixels obtained by the interpolation calculation is used as the predetermined region. 1. The image signal processing method described in 1. 3. In the adaptive control of the size of the predetermined area, the data of a plurality of pixels in a small area consisting of a plurality of adjacent horizontal and vertical rows of the original image signal, Pattern classification is performed according to a predetermined table, and based on the pattern classification result of a plurality of pixels in the small area, interpolation calculation is performed using data of a plurality of pixels in the small area, or the small area is calculated. 2. The image signal processing according to claim 1, wherein it is determined whether to perform an interpolation calculation using data of a plurality of pixels in a large area consisting of a plurality of horizontally and vertically adjacent columns that are enlarged with respect to the center. Method. 4. The small area comprises data of 3 × 2 pixels in three horizontal rows and two vertical rows adjacent to the original image signal, and the large area includes 3 × 2 pixels of the small area. 7x2 with 7 horizontal rows and 2 vertical rows adjacent to each other and expanded centering
The image signal processing method according to claim 3, wherein the image signal processing method comprises pixel data. 5. An image signal processing apparatus for forming an enlarged image signal by enlarging an original image from the original image signal, wherein the first and second horizontal and vertical columns of the original image signal are each composed of a plurality of adjacent columns. A storage unit for storing a plurality of pixel data in the area, and a second row composed of a plurality of adjacent horizontal and vertical columns in the first area stored in the storage unit.
Pattern classification is performed on a plurality of pixel data in the area according to a predetermined table, and the size of the second area is adaptively controlled based on the pattern classification result to classify the pattern. Interpolation calculation according to the pattern classification result is performed using a plurality of pixel data in the second area whose size is adaptively controlled, and the data of the interpolation pixel by the interpolation calculation is set between the data of each pixel of the original image signal. And an enlarged image signal forming means for inserting the image signal into an enlarged image signal to form an enlarged image signal. 6. An area formed by a column composed of a plurality of adjacent pixels of the original image signal and a column composed of a plurality of interpolated pixels obtained by the interpolation calculation is used as the predetermined region. 5. The image signal processing device according to item 5. 7. The enlarged image signal forming means is the second
Based on the pattern classification result according to a predetermined table for the data of a plurality of pixels in the area, the interpolation calculation using the data of the plurality of pixels in the second area, or
The image signal processing apparatus according to claim 5, wherein it is determined whether or not an interpolation calculation using data of a plurality of pixels in the first area is performed. 8. The plurality of pixel data in the first region is data of 7 × 2 pixels in seven horizontal columns and two vertical columns adjacent to each other in the original signal of the one field, and the second region is 6. The image signal processing apparatus according to claim 5, wherein the data is data of 3 × 2 pixels in adjacent three horizontal rows and two vertical rows in the first area. 9. W line buffers for respectively storing line data of the original image signal (w is 2 or more), and line data supply means for outputting the same line data v times and supplying the same to the w line buffers. And the storage means stores the data read from the line buffer, and the enlarged image signal forming means stores the w
Interpolation operation is performed using the data of the first area extracted from the line buffers and stored in the storage means, and the interpolation data of the interpolation operation is completed for the w line buffers. The image signal processing apparatus according to claim 5, wherein the portion in which the data is stored is overwritten.