JPH05153562A - Correlation detection interpolation method and device - Google Patents

Abstract

PURPOSE:To decide an optimum interpolation direction, to prevent picture quality deterioration and to interpolate the resolution into a high resolution by outputting a corrected value selectively among calculated values for each interpolation line inputted from an arithmetic operation means. CONSTITUTION:Picture information from a picture input terminal is converted into an information string with 7 picture elements on a C line by horizontal delay means 2A-2F. Similarly, the picture information is converted into an information string with 7 picture elements on the C line by a vertical delay means 6 and horizontal delay means 1A-1F. The picture element information on the lines A, C is inputted to arithmetic operation means 4A-4G and an arithmetic mean value is calculated as an interpolated value for each interpolation line in a direction of an interpolation line 7 passing through the picture element to be interpolated. The arithmetic operation means 4A-4C calculate the interpolation value in the interpolation direction decreasing in a lower right direction, an arithmetic operation means 4D calculates that of the interpolation direction in the vertical direction, and the arithmetic operation means 4E-4G calculate that of the interpolation direction decreasing in a lower left direction. The interpolated values are inputted in a decision means 9, in which they are processed and any of which in the 7 kinds of interpolated values is to be used is decided, the result is passed selectively from a selection means 10 and outputted from a video output terminal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interpolation method and apparatus used when creating a frame signal from a field signal, for example, in the field of video and information such as televisions, videos, printers and copiers that handle gradation images. ..

[0002]

2. Description of the Related Art In recent years, the importance of pixel density conversion technology has increased with the digitization of video equipment. IDTV
(Improved Definition TV) and EDTV (Enhanced Defini
Non-interlacing is an important technology for broadcast TV, video signals, etc., because interlacing is performed from two fields into one frame. This non-interlacing can be easily performed by using the information one field before when there is frame correlation such as a still image, but the information one field before when there is no frame correlation such as a moving image. Is 1/60
Since it is the information at the time point a second before, it cannot be used and it is necessary to make one field by interpolating one field between scanning lines. Similarly, even in a hard copy device for video signals such as a video printer, the print engine unit has a configuration in which recording is performed with the number of pixels corresponding to the frame, and when the input video signal is a still image, the frame is recorded as it is and a moving image is recorded. In some cases, the information in one field is interpolated to be converted into the number of pixels corresponding to a frame for recording.

Conventionally, these field interpolations use a method called linear interpolation which takes an average of pixels of upper and lower scanning lines. This method aims at the smoothness of an image due to an increase in the number of pixels, because the resolution is abandoned due to the nature of interpolation that creates data for many pixels from a small amount of pixel information. The interpolated image becomes an image with a more blurred impression than the normal frame image.

Therefore, as another interpolation method for solving the problem of the linear interpolation, the statistical properties such as the continuity of the image are used, and the correlation detection is used to perform the interpolation in the image more than the linear interpolation. Interpolation methods have been studied for the purpose of obtaining the smoothness of the diagonal line of and the vertical resolution higher than linear interpolation. An interpolation method having this correlation detection will be described below with reference to FIG.

In FIG. 5, the A line and the C line are scanning lines which are continuously input in the same field, and the B line is a scanning line which is not input in this field and is a line to be interpolated. If the pixel to be interpolated is Bn in the B line, the differences Δ1, Δ2, Δ3 in the brightness levels in the three directions passing through Bn between the A and C lines are expressed by the following equation.

[0006] The value Bn to be interpolated is selected as follows. That is, according to this interpolation method, the level difference between the pixel An above the pixel Bn to be interpolated and the pixel Cn below, and the pixel An + 1 at the upper right.
And the level difference between the lower left pixel Cn-1 and the upper left pixel An-1
Is compared with the level difference between the lower right pixel Cn + 1 and the direction having the smallest pixel level difference is judged to have the highest image continuity, that is, the correlation, and the average value of the pixel set in that direction is interpolated. It is what ("Photo Industry" October 1989 PP107-108
Also, apply the same idea hierarchically and set the interpolation direction to 3
Some are enlarged to the left and right from the direction. (JP-A 2-17
7683).

[0007]

However, in the prior art, the absolute values of the pixel level differences in a plurality of interpolating directions such as downward rightward, downward leftward, and upward and downward are mutually compared to determine the correlation that determines the interpolation direction. However, since the smallest one among them is determined to have the highest correlation, and interpolation is performed in the direction of the interpolation line, there are the following problems.

When the pixel level difference is large in any of the interpolation directions, it is judged that there is no correlation originally and linear interpolation should be performed, but a pixel level difference which is as small as possible is selected among them. Therefore, there is a problem that the correlation is erroneously detected, and interpolation is performed by using the average value of pixels in the wrong direction as an interpolation value, and the pixel becomes noise, which causes deterioration of image quality.

Further, even if the pixel level difference is small in any of the interpolation line directions and there is a correlation in all directions, the linear interpolation should be performed while judging that there is an inherent correlation, and Among them, the pixel level difference is selected to be as small as possible, so that there is a problem in that erroneous detection similarly occurs and image quality is deteriorated by noise. FIG. 6 is an example of an image in which noise is generated due to interpolation. Solid circles are input pixels, and dashed circles are interpolated pixels. Two vertical black line images are input vertically with 1 pixel spacing,
When the pixel of Bn is obtained by interpolation, the one having the smallest pixel level difference in the three directions shown in the figure is selected.
In this example, the pixel level differences in the three directions are small, but the pixel levels slightly vary, so if the level difference in either the left or right diagonal direction is the smallest, the diagonal interpolation is selected and Bn There is a problem in that black is selected as the interpolation value, resulting in interpolation noise.

Further, for example, when the level difference between the pixels in both the right-down direction and the left-down direction is smaller than the level difference between the pixels in the vertical direction, that is, there is a contradiction in the correlation judgment due to the pixel level difference, Regardless of whether or not it should be determined that there is essentially no correlation, and vertical linear interpolation should be performed, the conventional example has a problem that either one is selected and image quality is similarly deteriorated.

Generally, in the correlation interpolation method, if the correlation is detected from the property of the continuity of the image and the interpolation direction can be appropriately detected, the smoothness of the diagonal line in the image and the vertical resolution can be improved. Although it is a feature, if there is an error in the detection of the correlation, there is a problem that it becomes noise and deteriorates the image quality. For this reason, how accurately the correlation can be determined is extremely important.

In particular, in order to obtain an image as smooth as a frame signal from an actual field signal, it is necessary to improve even diagonal lines that are substantially horizontal, so at least 7 directions are required as interpolation directions. When interpolation is performed in the direction that is closest to the horizontal among the seven directions, it is possible to improve diagonal lines that are substantially horizontal, but if interpolation is performed erroneously, the pixels to be interpolated are shifted horizontally. There is a problem in that the pixels are separated by 6 pixels, which causes large noise and deterioration of the horizontal resolution. Therefore, when the number of pixels to be referred to for determining the correlation is large, the closer to the horizontal direction, the more accurate the correlation detection is required.

In view of the above problems, it is an object of the present invention to provide an interpolating device equipped with correlation detection which prevents deterioration of image quality due to erroneous correlation judgment and further improves interpolating performance in a direction close to horizontal.

[0014]

In order to solve the above problems, the correlation interpolation detection method of the present invention determines the pixel to be interpolated.
B (0), the pixel directly above it is A (0), the pixel below is C (0),
The left and right 7 pixels including A (0) are A (-3), A (-2), A (-1), A
(0), A (1), A (2), A (3), and 7 pixels on the left and right including C (0)
(-3), C (-2), C (-1), C (0), C (1), C (2), C (3)
A (-3) which is a set of pixels on the interpolation line in 7 directions passing through (0)
And C (3), A (-2) and C (2), A (-1) and C (1), A (0) and C
(0), A (1) and C (-1), A (2) and C (-2), A (3) and C (-3)
For A (-1) and C (1), A (0) and C (0), A (1)
And C (-1), which has a three-direction interpolation line, and A (1) and C (-1), A (2) and C (-2), A (3) and C
A second area having a three-direction interpolation line consisting of (-3) and A (-3) and C (3), A (-2) and C (2), A (-1) and C (
1) is a method of interpolating B (0) by dividing into a third area having a three-direction interpolation line consisting of
For each interpolation line in the area, the step of comparing the difference between the two pixel levels with a predetermined reference set value to detect the presence or absence of correlation, and in the first area in the downward left direction (A (1) -C (-1) direction), the second
Similarly, the presence or absence of correlation is detected for the interpolation lines in the three directions in the area, and for the interpolation line for which correlation is detected, the arithmetic mean of the two pixel levels is calculated as the value of B (0). If there is a contradiction in the detection of correlation, A (1) −C (-
1) The step of setting the arithmetic average of pixel levels in the direction B (0) to the value of B (0), and the downward rightward direction (A (-
1) −C (1) direction), the presence or absence of the correlation is similarly detected for the interpolation lines in the three directions of the third region, and the interpolation line for which the correlation is detected is detected. Is
The arithmetic mean of the two pixel levels is taken as the value of B (0),
When there is a contradiction in the detection of the correlation, the step of setting the arithmetic mean of the pixel levels in the A (-1) -C (1) direction as the value of B (0), Otherwise, A (0)
It is characterized in that the value of the pixel value B (0) to be interpolated is determined and interpolated from the step of setting the arithmetic average of the pixel level in the −C (0) direction as the value of B (0).

In the correlation detecting / interpolating device of the present invention, the pixel to be interpolated is B (0), the pixel directly above it is A (0), the pixel immediately below is C (0), and the above A (0). The left and right 7 pixels including A (-3), A (-
2), A (-1), A (0), A (1), A (2), A (3), and the left and right 7 pixels including C (0) are C (-3), C (- 2), C (-1), C (0), C (1), C (2), C (3), which is a set of pixels on an interpolation line in 7 directions passing through B (0). (-3) and C (3), A (-2) and C (2), A (-1) and C (1)
, A (0) and C (0), A (1) and C (-1), A (2) and C (-2), A
(3) and C (-3), A (-1) and C (1), A (0) and C
(0), A (1) and C (-1), the first area having a three-direction interpolation line and A (1) and C (-1), A (2) and C (-2)
, A (3) and C (-3), a second area having a three-direction interpolation line, A (-3) and C (3), A (-2) and C (2),
A device for interpolating B (0) by dividing into a third region having interpolation lines in three directions consisting of A (-1) and C (1), for each interpolation line in the seven directions. The arithmetic means for taking the arithmetic mean of the two pixel levels to find the pixel level of the pixel B (0) to be interpolated, and the difference between the two pixel levels for each interpolation line in the 7 directions A correlation value detecting means for detecting a correlation value; a setting means for outputting a reference setting value for determining the presence or absence of correlation with respect to the correlation value output by the correlation value detecting means; In the second area and the third area, the correlation value output from the correlation value detecting means and the reference setting value are compared in magnitude to generate a binarized signal indicating the presence or absence of correlation in each direction. A comparing means for outputting each area, and a determining means for determining a direction to be interpolated by the output of the comparing means. Selecting means for selectively passing the output of the computing means corresponding to the output of the deciding means, wherein the deciding means has respective binarized signals for the first region, the second region and the third region. Of the interpolating line in the A (1) -C (-1) direction (downward leftward) in which the binarized signal is composed of first, second and third decoding circuits for decoding the The output of the second decoding circuit is prioritized over the output of the first decoding circuit, and the binarized signal is A (-1)-
When it is indicated that the interpolation line in the C (1) direction (downward to the right) has a correlation, the output of the third decoding circuit is prioritized over the output of the first decoding circuit, and the binarized signal is output. In other cases, the output of the first decoding circuit is prioritized and the control signal of the selecting means is output.

Further, the setting means outputs reference setting values of two levels of strength and weakness, and the comparing means uses the weak setting reference value in the first area, and the second area and the third area. The strong setting reference value may be used in the region.

[0017]

According to the present invention having the above-mentioned structure, the comparing means is arranged to correlate the interpolation lines for each of the first area, the second area and the third area based on the correlation value from the correlation value detecting means. Determine the presence or absence of sex. The determining means determines the interpolation direction based on the correlation in the second region when the first region has only the correlation in the downward left direction,
When the first region has only the correlation in the downward-sloping direction, the interpolation direction is determined based on the correlation in the third region, and in the first region otherwise, the region is determined. The interpolation direction is determined based on the correlation of. According to the output of the determining means, the selecting means selects and outputs a value to be interpolated from the calculated values for each interpolation line input from the calculating means.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A correlation detecting / interpolating device according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of a correlation detection / interpolation device according to an embodiment of the present invention. In FIG. 1, reference numerals 1A to 1F denote horizontal delay means for delaying one pixel, and pixel data of a scanning line (A line) above a pixel to be interpolated is sequentially shifted.

Reference numerals 2A to 2F denote horizontal delay means for delaying one pixel, and the pixel data of the scanning line (C line) below the pixel to be interpolated are sequentially shifted. 3A-3
F is a horizontal delay unit that delays by three pixels to match the position of the pixel to be interpolated. Reference numerals 4A to 4G are arithmetic means for performing arithmetic average calculation of the pixels on the A line and the pixels on the C line.

Reference numerals 5A to 5F are subtraction means for detecting the correlation value by subtracting the pixels on the A line and the pixels on the C line to calculate pixel level differences for calculating pixel level differences in various directions. Reference numeral 6 is a vertical delay means for delaying one scanning line period in order to match the timing so that the A line is output at the same time as the C line.

7A and 7C are absolute value comparing means for comparing the absolute value of the input from 5A to 5C and 5E to 5G with the set value k1 and binarizing 1 if it is larger than the set value and 0 if it is smaller than the set value. Is. 7B is an absolute value comparison means for comparing the absolute value of the input from 5C to 5E with a set value, k0, and binarizing 1 if it is larger than the set value and 0 if it is smaller than the set value.

Reference numeral 8A is a setting means for outputting a predetermined set value k0, which is input to the absolute value comparing means 7B. 8B
Is a setting means for outputting a predetermined set value k1, which is input to the absolute value comparing means 7A and 7C. Reference numeral 9 is a deciding means for deciding an optimum interpolation line based on the binary signal group.

Reference numeral 10 is a selection means for selectively passing one of a plurality of inputs according to a selection signal. FIG. 4 is an example of a detailed circuit diagram of the determining means 9 and the selecting means 10 of FIG.
Reference numerals 91, 92, and 93 are 3-bit input decoders, and
Each output of the book becomes active when the input corresponds to a combination of 3 bit numbers written to the right of the symbols in the figure.

94A, 94B, 94C, 94D, 94E
Are OR gates having 6 inputs, 4 inputs, 3 inputs, 3 inputs and 4 inputs, respectively. 95A to 95F is a 2-input AND
It is a gate. 100A to 100G are analog switches, the outputs of which are wired or connected.

The AND gates 95A, 95B and 95C are
Decoder 9 only when decoder 91 selects (1,0,0)
3 and the OR gates 94D and 94E give the judgment results to the selection means 10, and AND gates 95D, 95E and 95F.
Serves to provide the selection means 10 with the judgment result of the decoder 92 and the OR gates 94B and 94C only when the decoder 91 selects (0,0,1). When the decoder selects any other than the above, the judgment result by the decoder 91 and the OR gate 94A is given to the selecting means 10. The outputs of these determining means 9 are given to the corresponding analog switches 100A to 100G of the selecting means 10, and the interpolation value of the corresponding interpolation line is output from the image output terminal.

The operation of the correlation detecting / interpolating apparatus of the present embodiment constructed as above will be described with reference to the drawings.
The image information input from the image input terminal of FIG. 1 is processed by the horizontal delaying means 2A to 2F into 7 pixels C (-3), C (-2), on the C line.
Converted to the information sequence of C (-1), C (0), C (1), C (2), C (3). Similarly, 7 pixels A (-3), A (-2), A (-1), A (0), A (1), A (on the A line are formed by the vertical delay means 6 and the horizontal delay means 1A to 1F. 2), A
Converted to the information string in (3).

FIG. 2 shows pixel information (A and C with subscripts in the figure) of the input A line and C line centered on the pixel B (0) to be interpolated (B with subscript in the figure). It is a conceptual diagram of an interpolation operation, showing the positional relationship of pixel information to be interpolated, the interpolation lines in seven directions, and the interpolation values (H with a subscript in the drawing). The pixel information on the A line and the C line is input to the calculating means 4A to 4G in FIG. The calculation means 4A to 4G calculate an arithmetic mean as an interpolation value for each interpolation line in the direction of the interpolation line 7 passing through the pixel b (0) to be interpolated. The computing means 4A, 4B, 4C calculate the interpolated values H (-3), H (-2), and H (-1) in the interpolating directions of 18 degrees, 27 degrees, and 45 degrees to the right in FIG. 2, respectively. The calculation means 4D calculates the interpolation value H (0) in the vertical direction. Further, the calculation means 4E, 4F, 4G respectively use the interpolated values H in the interpolating directions of 45 °, 27 °, and 18 ° to the left in FIG.
Calculate (-3), H (-2), and H (-1). These interpolated values are input to the deciding means 10, and by the processing of the deciding means 9, which will be described later, which of these seven types of interpolating values should be used is decided, and the selecting means 10 makes a selective pass through the image output terminal. Is output from.

Pixel information on the A and C lines is shown in FIG.
Is also input to the subtraction means 5A to 5G. Subtraction means 5A-
The 5G subtracts the pixel on the A line and the pixel on the C line for each correlation line 7 direction to calculate the pixel level difference. The subtraction means 5A, 5B, and 5C calculate pixel level differences in the interpolating directions of 18 degrees, 27 degrees, and 45 degrees to the right in FIG. The subtracting means 5D calculates the pixel level difference in the vertical direction. Also, subtraction means 5E, 5F, 5G
Are respectively 45 °, 27 °, and 1 ° to the left in FIG.
The pixel level difference in the interpolation direction of 8 degrees is calculated. The calculated seven pixel level differences are 5A to 5C and 5C to 5
E, 5E to 5G are divided into three groups, and are input to absolute value comparing means 7A to 7C, respectively.

The outputs of the subtraction means 5C, 5D, 5E are converted into positive values by absolute value calculation by the absolute value comparison means 7B, and are compared with a predetermined set value k0 output by the setting means 8A to set values. 1 if smaller than k0, 0 if larger
Is converted into a 1-bit binarized signal (D (-1), D (0), D (1)) representing the presence or absence of correlation. This binarized signal (D (-1), D (0), D (1)) indicates the presence / absence of correlation with respect to the correlation lines in the three directions in the first region consisting of the six central pixels in FIG. means.

The outputs of the subtraction means 5A, 5B, 5C are converted into positive values by absolute value calculation by the absolute value comparison means 7A, and at the same time, a predetermined set value k1 (k1
<K0) and similarly binarized signals (E (-3), E (-2), E
(-1)). This binary signal (E (-3), E (-2), E
(-1)) is A (-3), A (-2), A (-1), C (1), C (2), C (3) in Fig. 2.
Means the presence or absence of correlation with respect to the correlation lines in the three directions in the second region consisting of 6 pixels.

The outputs of the subtraction means 5E, 5F, 5G are converted into positive values by absolute value calculation by the absolute value comparison means 7C, and are compared with a predetermined set value k1 output by the setting means 8B to be binarized. Converted to signals (E (1), E (2), E (3)). The binarized signals (E (1), E (2), E (3)) are A (1), A (2), A (3), C in FIG.
(3), C (-2), C (-1) means the presence / absence of correlation with respect to the correlation lines in the three directions in the second region consisting of 6 pixels.

The deciding means 9 determines the set of these binarized signals (D
(-1), D (0), D (1)) (E (1), E (2), E (3)) (E (-3), E (-2), E (-
Based on 1)), which of the interpolated values H (-3) to H (3) is optimum as the interpolated value is determined. FIG. 3 is a chart showing the characteristics of the determining means 9, which is the first value obtained using the set value k0.
Of the binarized signal (D (-1), D (0), D (1)) in the region of and the binarized signal (E (1), in the second region obtained by using the set value k1. E (2), E (3)) and the binarized signals (E (-3), E (-2), E (-1)) in the third region are used for the interpolation direction. <Number> in the figure indicates the interpolation direction.

Binarized output (D (-1), D in the first region
The determination of the interpolation direction for (0), D (1)) will be described.
(0,0,0) means that the pixel level difference exceeds the predetermined set value k0 and has no significant correlation in any of the three interpolation directions of vertical, 45 degrees to the right, and 45 degrees to the left. Means that vertical direction H (0) is selected as the interpolation direction, and (1,1,1,
Since (1) has a correlation in all three directions, and as a result, no significant directionality can be found, H (0) is also selected, and (1,
0,1) means that there is a correlation in the direction of 45 degrees to the right and a correlation in the direction of 45 degrees to the left. Since there is a contradiction in the correlation detection, it is regarded as an erroneous detection. (0)
Select. Also, (0,1,0) has a vertical correlation
Select H (0).

At (0,0,1), since the pixel level difference in the downward left direction exceeds the predetermined set value k0, H (1) is set in this area.
It can be considered that the correlation of the directions is high, but there is a possibility that the direction to be interpolated is a shallower angle to the lower left direction.
It is determined in consideration of (E (1), E (2), E (3)) in the second area described later. Similarly for (1,0,0), the correlation in the H (-1) direction is high in this region, but there is a possibility that the direction to be interpolated is a shallower angle to the lower right direction. (E (-3), E (-2), E (-1)) are taken into consideration when making a decision.

Further, since (0,1,1) and (1,1,0) have no contradiction in correlation but indicate an interpolation direction between 45 degrees and vertical, when the interpolation performance is improved. May be treated the same as (0,0,1) and (1,0,0), but in the present embodiment, H (0,0) is treated in the same way as (0,1,0) in order to reduce interpolation errors. )
Is selected. Next, in the second area, (E (1), E
The determination of the interpolation direction for the state (2), E (3)) will be described.

Since the set value k1 for determining the correlation in the second area is k1 <k0, (D (-1), D (0), D (1)) =
Even if it is (0,0,1), E (1) does not always become 1. Therefore, also in this case, there are eight combinations shown in FIG. (E (1), E (2), E (3)) = (0,0,0) has no significant correlation with any of the three interpolation directions at the set value k1, so the interpolation direction As the previous judgment result, H (1) is selected as, and (1,0,1) has a contradiction in the correlation detection, so H (1) is selected in the same manner as the non-correlation.

However, since (1,1,1) has a correlation in all three directions and as a result no significant directionality can be found, H (2) is selected together with the previous judgment. In addition, significant correlations are found (0,0,1), (0,1,0), (1,0,0).
), The result of this judgment is prioritized and H (3), H (2), H
Select (1). Since (0,1,1) and (1,1,0) mean that there is no contradiction in the correlation but the correlation direction is between 7 interpolation lines, (0,0,1) and (0 , 1,0), but in the present embodiment, in order to reduce interpolation errors, they are treated as being closer to vertical (0,1,0) and (1,0,0). , H (2) and H (1) are selected.

(E (-3), E (-2), E (-1) in the third region
The determination of the interpolation direction with respect to the above condition will be described.
Since the set value k1 for determining the correlation is k1 <k0, (D (-
Even if 1), D (0), D (1)) = (1,0,0), E (-1) does not always become 1, and in this case as well, there are eight patterns shown in FIG. There are combinations.

(E (-3), E (-2), E (-1)) = (0,0,0) is
Since the set value k1 has no significant correlation with any of the three types of interpolation directions, it is the previous determination result as the interpolation direction.
H (-1) is selected, and at (1,0,1), since there is a contradiction in the correlation detection, H (-1) is selected as in the case of no correlation. However,
(1,1,1) has a correlation in all three directions, and as a result, no significant directionality can be found, but H (-2) is selected together with the previous judgment.

A significant correlation is also found (0,0,1
), (0,1,0) and (1,0,0), the decision result of this time is prioritized and H (-1), H (-2) and H (-3) are selected. (0,1,
1) and (1,1,0) are treated as the same as (0,1,0) and (1,0,0) because the correlation is consistent but the correlation direction means between interpolation lines. However, in the present embodiment, it is closer to vertical (0,0,
1) and (0,1,0) are treated as H (-1) and H
(-2) is selected.

Next, the operation of the circuit of FIG. 4 embodying the above characteristics will be described. Binary data (D (-1), from the comparison means 7B,
D (0), D (1)) is decoded by the decoder 91, and the binary data (E (1), E (2), E) from the comparison means 7A, 7C is decoded.
(3)), (E (-3), E (-2), E (-1)) are decoders 92, 9
3 is decoded. AND gates 95A and 95
B and 95C give the judgment result by the decoder 93 and the OR gates 94D and 94E to the selecting means 10 only when the decoder 91 selects (1,0,0), and the AND gates 95D and 9C.
5E and 95F function to give the selection result to the selection means 10 only when the decoder 91 selects (0,0,1) and the judgment result by the decoder 92 and the OR gates 94B and 94C. When the decoder selects any other than the above, the judgment result by the decoder 91 and the OR gate 94A is given to the selecting means 10. When the outputs of the determining means 9 are given to the analog switches 100A to 100G of the selecting means 10, the interpolation value of the corresponding interpolation line is output from the image output terminal.

For example, (D (-1), D (0), D (1)) is (1,0,0
), The output (1,0,0) of the decoder 91 becomes active and AND gates 95A, 95B, 95C become active, and at this time (E (-3), E (-2), E (-1) ) Is (1, 1, 0)
In this case, the OR gate 94D becomes active, so AN
The buffer gate 100B is activated by passing through the D gate 95B, and the interpolation value H (-2) is output to the image output terminal.

To summarize the operation of the present embodiment described above, first, using the information of 6 pixels in the first area, the level difference of the pixels of the interpolation lines in three directions including the pixel to be interpolated is detected. , Binarization is performed by comparing with the loose setting value k0, and 3 is used by using the combination of the binarization output for these interpolation lines in three directions.
The optimum interpolation line is determined from among the different interpolation directions.
Next, when the interpolation direction in the lower left direction is selected in the first area, the optimum interpolation line is selected from the three kinds of interpolation directions using the strict setting value k1 for the 6 pixels in the second area. If the lower right interpolation direction is selected in the first area, the optimum interpolation line is similarly selected from the three kinds of interpolation directions for the 6 pixels in the third area. To do. With respect to one interpolation line determined from the interpolation lines in seven directions, linear interpolation is performed in that direction.

In this embodiment, seven pixels are arranged on the upper and lower scanning lines for seven interpolation directions. However, for example, for every five pixels, interpolation in five directions is performed.
This can be easily done by setting the interpolation lines in the two directions in the second and third areas and determining the second stage. Although the present embodiment has been described as monochrome information, a method of detecting correlation using a luminance signal or a G signal (in the case of RGB signal) and interpolating in the same direction with respect to other signals is suitable for color information. is there.

Although the present embodiment has been described with the hardware configuration, it can be realized by using substantially the same software routine.

[0046]

As described above, according to the present invention, the correlation that determines the interpolation direction can be independently determined for each interpolation line, and the optimum interpolation direction can be determined according to all cases of those combinations. Since it can be determined, it has an excellent effect in the following points as compared with the method of comparing the pixel level differences in each direction, which is the conventional method.

When the pixel level difference is large in any of the interpolation directions, it can be clearly determined that there is no correlation, so that there is an effect that the correlation detection error caused by it can be prevented, and the noise due to the correlation detection error is eliminated. Image quality deterioration can be prevented. In addition, even if the pixel level difference is small in any interpolation line direction and there is correlation in all directions,
Since it can be clearly determined that there is a correlation in all directions, there is an effect that a correlation detection error due to it can be prevented, and similarly, image quality deterioration can be prevented.

Further, since the correlation information for the shallow angle of 45 degrees or less is detected by using the highly reliable correlation information of 6 pixels close to the pixel to be interpolated, the detection error for the shallow angle with high probability of detection error is prevented. it can. When these effects are combined, there is an effect that noise is not generated, diagonal lines in the image are smooth, and interpolation with high resolution can be performed.

[Brief description of drawings]

FIG. 1 is a block diagram of a correlation detection / interpolation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing the positional relationship between the pixel information of the input A line and C line and the pixel information to be interpolated centering on the pixel B (0) to be interpolated in the same embodiment; FIG. 7 is a conceptual diagram of an interpolation operation, showing an interpolation value (H with a subscript in the drawing).

FIG. 3 is a chart diagram for explaining a correlation detection operation in the embodiment.

FIG. 4 is a detailed circuit diagram of a determining means and a selecting means of the correlation detecting / interpolating device in the embodiment.

FIG. 5 is an explanatory diagram of an interpolation method having correlation detection of a conventional example.

FIG. 6 is an explanatory diagram of an example of an image in which interpolation noise of a conventional example occurs.

[Explanation of symbols]

 1A to 1F Horizontal delay means 2A to 2F Horizontal delay means 3A to 3G Horizontal delay means 4A to 4G Calculation means 5A to 5G Subtraction means 6 Vertical delay means 7A to 7C Absolute value comparison means 8A to 8B Setting means 9 Determining means 10 Selection means 91-93 decoder 94A-94E OR gate 95A-95F AND gate 100A-100G analog switch

Claims (4)

[Claims] 1. A pixel to be interpolated is B (0), a pixel immediately above it is A (0), a pixel directly below is C (0), and 7 pixels on the left and right including the A (0) are A (0). -3), A (-2), A (-1), A
(0), A (1), A (2), A (3), 7 pixels on the left and right including C (0) are C (-3), C (-2), C (-1), C
(0), C (1), C (2), C (3), which is a set of pixels on an interpolation line in 7 directions passing through B (0).
A (-3) and C (3), A (-2) and C (2), A (-1) and C (1), A
(0) and C (0), A (1) and C (-1), A (2) and C (-2), A (3) and C (-3), The first area having a three-direction interpolation line consisting of C (1), A (0) and C (0), and A (1) and C (-1) and the above A (
1) and C (-1), A (2) and C (-2), A (3) and a second area with a three-direction interpolation line consisting of C (-3) and A (-3)
And C (3), A (-2) and C (2), A (-1) and C (1) 3
Divided into a third region having a direction interpolation line,
A method of interpolating B (0), which comprises comparing a difference between two pixel levels with a predetermined reference set value for each interpolation line in the first area, and detecting the presence or absence of correlation. If there is a correlation in the downward left direction (A (1) -C (-1) direction) in the first region, the presence / absence of correlation in the three-direction interpolation lines in the second region is the same. For the interpolation line in which the correlation is detected, the arithmetic mean of the two pixel levels is used as the value of B (0), and if there is a contradiction in the detection of the correlation, A ( 1) a step of setting the arithmetic average of pixel levels in the -C (-1) direction to the value of B (0), and a downward rightward direction in the first region (A (-1) -C (1) direction ), The presence or absence of the correlation is similarly detected for the interpolation lines in the three directions of the third region, and the interpolation line for which the correlation is detected is detected. , The arithmetic mean of the two pixel levels is taken as the value of B (0), and if there is a contradiction in the detection of correlation, the arithmetic mean of the pixel levels in the A (-1) -C (1) direction To the value of B (0), and A (0) -C (0) in the first region except the above.
A correlation detection interpolation method characterized in that the value of a pixel B (0) to be interpolated is determined and interpolated from the step of setting the arithmetic mean of pixel levels in the direction to the value of B (0). 2. The setting reference value used in the step of detecting the presence or absence of the correlation in the first area is the setting used in the step of detecting the presence or absence of the correlation in the second area and the third area. The correlation detection method according to claim 1, wherein the correlation detection value is larger than the reference value. 3. A pixel to be interpolated is B (0), a pixel directly above it is A (0), a pixel directly below is C (0), and 7 pixels on the left and right including the A (0) are A (0). -3), A (-2), A (-1), A
(0), A (1), A (2), A (3), 7 pixels on the left and right including C (0) are C (-3), C (-2), C (-1), C
(0), C (1), C (2), C (3), which is a set of pixels on an interpolation line in 7 directions passing through B (0).
A (-3) and C (3), A (-2) and C (2), A (-1) and C (1), A
(0) and C (0), A (1) and C (-1), A (2) and C (-2), A (3) and C (-3), The first area having a three-direction interpolation line consisting of C (1), A (0) and C (0), and A (1) and C (-1) and the above A (
1) and C (-1), A (2) and C (-2), A (3) and a second area with a three-direction interpolation line consisting of C (-3) and A (-3)
And C (3), A (-2) and C (2), A (-1) and C (1) 3
Divided into a third region having a direction interpolation line,
A device for interpolating B (0), which calculates an arithmetic mean of two pixel levels for each interpolation line in the seven directions to obtain a pixel level of a pixel B (0) to be interpolated. Means, a correlation value detecting means for detecting a correlation value by taking a difference between two pixel levels for each of the 7-direction interpolation lines, and a correlation value output by the correlation value detecting means for the presence or absence of correlation Setting means for outputting a reference set value for determining, and the first area, the second area and the third area,
Correlation value output from the correlation value detection means, the size of the reference set value is compared, a binarized signal indicating the presence or absence of correlation for each direction, a comparison means for outputting for each region, The determining means determines the direction to be interpolated by the output of the comparing means, and the selecting means for selectively passing the output of the computing means corresponding to the output of the determining means, and the determining means includes the first region. , The second area and the third
The first, second and third decoding circuits for decoding the respective binarized signals for the area of (1) and controlling the selecting means, wherein the binarized signals are in the A (1) -C (-1) direction. When it is indicated that the correlation of the interpolation line is (downward left), the output of the second decoding circuit is prioritized over the output of the first decoding circuit, and the binarized signal is A (-1 ) −C (1) direction (downward to the right) indicates that there is correlation between the interpolation lines, the output of the third decoding circuit is prioritized over the output of the first decoding circuit, and the binary value A correlation detecting / interpolating device, characterized in that, when the converted signal is other than that, the output of the first decoding circuit is prioritized and the control signal of the selecting means is output. 4. The setting means outputs reference setting values in two levels of strength and weakness, and the comparing means uses the weak setting reference values in the first area, the second area and the third area. 4. The correlation detecting / interpolating device according to claim 3, wherein the strong setting reference value is used in the area.