JPH048089A - Interpolation device for offset sampling signal - Google Patents

Abstract

PURPOSE:To prevent occurrence of deterioration in a picture caused by loopback distortion by detecting correlation between a picture element group located around an interpolation picture element and plural other picture element groups based on each picture element data subject to n-value processing of each picture element group and selecting an interpolation means in response to the result of detection. CONSTITUTION:Each picture element data of group phi is binarized by a binarization circuit 10 and fed respectively to 1st-4th correlation calculation circuits 6, 7, 8, 9. Each picture element data of a group 1 is binarized by a binarizing circuit 12 and the result is fed to a 1st correlation calculation circuit 6 and the correlation between groups is calculated based on the picture element data of the groups phi, 1. The second-4th correlation calculation circuits 7-9 are implemented similarly. The calculation outputs P1-P4 by each correlation calculation circuit are fed to an interpolation means selection circuit 4 respectively. The above circuit 4 detects a group having the highest correlation with respect to the group phi based on the supplied output PN and outputs a selection signal selecting a prescribed interpolation method.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an interpolation device for offset-sampled image signals, etc.

(Prior Art) A method called offset sampling has been known as a method for compressing bandwidth or reducing the amount of information when recording and transmitting various information signals such as image signals.

In the two-dimensional case, this offset sampling refers to the horizontal direction (X direction) and vertical direction (X direction) as shown in Figure 5.
The sampling interval (Tx, Ty) is set to twice the pixel interval (Hx, Hy) in the original signal, and the vertically adjacent sampling points are separated from each other by half the sampling interval (T x / 2). By performing such offset sampling, the transmission band can widen the spatial frequency components in the horizontal or vertical direction relative to the spatial frequencies in the diagonal direction, as shown in FIG. In addition, when displaying or printing out the image signal that has been offset sampled and ringed as described above, the pixels between each sampling point (interpolated pixels) are interpolated from the adjacent pixels as shown in Figure 7. Interpolation processing is required. This interpolation process functions as a spatial filter that passes the frequency components in the shaded area shown in Figure 6 and blocks the frequency components in the area including the turning point A.This interpolation process is based on sampling theory. Above, it is positioned as a post-filter.

By the way, offset sampling as described above is
This is a very effective method if the pre-filter before sampling is incorrect or has been sufficiently band-limited, but if the pre-filter is not sufficiently applied due to hardware constraints, for example If a pre-filter is not applied sufficiently in order to widen the transmission band, etc., the problem of image deterioration due to generation of aliasing distortion will occur.

That is, if the prefilter is insufficient, the offset
When Sun and Bling are performed, the horizontal high-frequency components of the original signal are folded back as vertical high-frequency components as shown in Figure 8(a), and the vertical high-frequency components are folded back as shown in Figure 8(b). The high frequency components are folded back as high frequency components in the horizontal direction.

For this purpose, for example, an image including a straight line portion as shown in FIG. 9(a) is offset-sampled as shown in FIG. 9(b) without sufficiently band-limiting, and the sampling signal including this aliasing component is On the other hand, even if the interpolation processing described above is performed, the aliasing distortion cannot be sufficiently removed because the high frequency components in the original signal and the high frequency components due to aliasing cannot be separated. As shown in the figure, uneven shading and blurring occur in the straight line portions, resulting in significant image deterioration in the output image.

Therefore, in the past, a technique called an adaptive interpolation method was generally known as a technique for reducing the occurrence of aliasing distortion as described above.

This adaptive interpolation method is based on the detection results when performing interpolation processing on sampled image signals.
For example, unnecessary aliasing components are removed by selectively using spatial filters having characteristics as shown in FIGS. 10(a) to 10(C).

In this case, if the interpolated pixels are pixels forming a straight line in the horizontal direction, a filter having characteristics that allows more frequency components in the horizontal direction to pass, as shown in FIG. 10(a), should be used. Filters (removes) vertical frequency components including aliasing components.

(Problems to be Solved by the Invention) In the above-described adaptive interpolation method, determining which interpolation means to use determines the performance of the interpolation device.

Therefore, when there is a high probability that a false determination will occur, interpolation processing based on the false determination not only reduces the original signal component (but also increases aliasing distortion, which is a problem). be.

(Means for Solving the Problems) The present invention has been made in view of the above-mentioned actual situation, and it is possible to accurately detect the "tendency" of an image around an interpolation pixel to prevent the occurrence of aliasing distortion. It is an object of the present invention to provide an interpolation device for offset sampling signals that can prevent such occurrences.

In order to achieve this object, the present invention includes a plurality of interpolation means having different interpolation methods, a pixel group consisting of pixels located around the interpolated pixel, and a plurality of pixel groups located around this pixel group. The degree of correlation with other pixel groups is converted to n-value for each pixel group (n is an integer of 2 or more)
Provided is an offset sampling signal interpolation device characterized by comprising a selection means for detecting based on each pixel data and selecting one of the outputs of each of the interpolation means according to the detection result. It is something.

(Function) According to the present invention configured as described above, the correlation between a pixel group made up of pixels located around an interpolated pixel and a plurality of other pixel groups located around this pixel group is achieved. By selecting the interpolation means accordingly, it is possible to perform interpolation processing based on the result of accurately determining the "tendency" around the interpolation pixel.

Thereby, it is possible to prevent the occurrence of aliasing distortion based on the conventional erroneous determination, and it is possible to prevent the occurrence of image deterioration due to the aliasing distortion.

Further, in the present invention, each pixel data to be supplied is
By converting the data into n-values and calculating the above-mentioned correlation degree based on this n-valued data, the arithmetic processing for calculating the correlation degree can be simplified, and the processing speed can be increased and hardware It is possible to reduce the scale.

(Embodiment) Hereinafter, a preferred embodiment of the offset/zumbling signal interpolation device according to the present invention will be described in detail with reference to FIGS. 1 to 4.

FIG. 1 is a block diagram showing the configuration of the interpolation device of this embodiment,
FIGS. 2(a) and 2(b) are layout diagrams of each pixel used to detect a "tendency" around the interpolation pixel in this embodiment.

In this embodiment, it is detected whether it is better to interpolate the interpolated pixel from pixels in the horizontal direction, from pixels in the vertical direction, or by using all surrounding pixels. Therefore, not only the adjacent image collection group φ consisting of four pixels A(φ) to D(ψ) adjacent to the interpolation pixel in the horizontal and vertical directions, but also the four pixels each including one pixel of each pixel in this group φ Using −4 to the peripheral pixel group 1 consisting of φ to 4, the correlation between each of these groups φ to 4 is determined (detected).

That is, A/D (analog-digital) not shown.
Each pixel data of the group φ of predetermined bits (for example, 12 bits) outputted from a converter etc. is supplied to a horizontal interpolation circuit 1, a vertical interpolation circuit 2, and a two-dimensional interpolation circuit 3, respectively. Ru. In the horizontal interpolation circuit 1, each pixel data (A, B,
CD), the value of the interpolated pixel is calculated by performing an interpolation calculation, for example, (B+D)/2. Similarly,
In the vertical interpolation circuit 2, based on each pixel data of the supplied group φ, for example, (A+C)/2
The value of the interpolated pixel is calculated by performing the following interpolation calculation. Furthermore, the two-dimensional interpolation circuit 3 calculates the value of the interpolated pixel by performing an interpolation operation p, for example (A+B+C+D)/4, based on each pixel data of the supplied group φ.

The interpolated pixel data calculated as described above are supplied to switches 5 whose switching is controlled by an interpolation means selection circuit 4, which will be described later.

Furthermore, each pixel data of the group φ is stored in the binarization circuit 1.
Binarized at 0, the first to fourth correlation calculation circuits 6
．． 7.8.9 respectively.

Here, each of the binarization circuits 10 divides each supplied pixel data into digits according to the threshold supplied from the threshold determination circuit 11.
The threshold value determination circuit 11 determines the threshold value according to the pixel data of the group φ, and applies this threshold data to the binarization circuit IO and other binarization circuits 12 and 13 described later. .14 and 15 respectively. Note that in this embodiment, the average value of all the pixel data within the group φ or the average value of the maximum value and minimum value of the pixel data within the group φ is used as the threshold value.

On the other hand, the first correlation calculation circuit 6 includes the group 1.
Each pixel data is binarized and supplied to a binarization circuit 12, and this first correlation calculation circuit 6 converts the supplied pixel data of group φ (2f normalized data) and pixel data of group 1. The degree of correlation between these groups is calculated based on (binary IHI data). Similarly, the second to fourth
Each pixel data of group 2 and group 4 is binarized and supplied to the tenth correlation calculation circuits 7 to 9, respectively, and the second correlation degree is calculated. A calculation circuit 7 calculates the correlation between group φ and group 2, a third correlation calculation circuit 8 calculates the correlation between group φ and group L-3, and a fourth correlation calculation circuit 9 calculates the correlation between group φ and group 2. calculates the degree of correlation between group φ and group 4. The calculated outputs P(1) to P(4) from each of these correlation degree calculation circuits are supplied to each of the interpolation means selection circuits 4.

In this embodiment, the correlation degree P (N) is calculated by each of the correlation degree calculation circuits 7 to 9 as follows: P(N)=A(ψ)■
A(N) + 8(φ) [Yes] B(N) 10C(φ)
■C(N) + D(φ)CD (Nl(N=1.2
．． 3.4) It is carried out based on the formula:

Furthermore, in the above equation, the calculation of X and Y follows the calculation rules as shown in the table below.

The correlation degree calculation output P(N) as described above is supplied to the interpolation means selection circuit 4, and this interpolation means selection circuit 4 selects the group φ that has the highest correlation degree based on the supplied output P(N). The group is detected, and a selection signal for selecting a predetermined interpolation method is output in accordance with the detection result. That is, as shown in FIG. 3, first the supplied calculation output P(
Step 1): From N), detect those with a high degree of correlation in the horizontal direction and those with a high degree of correlation in the vertical direction.
, it is determined whether these are equal or not (step 2).

As a result, if they are equal, it is determined that it is impossible to determine whether it is better to interpolate the interpolated pixels starting from the pixels in the horizontal direction or from the pixels in the vertical direction, and the output of the two-dimensional interpolation path 3 is switched to the switch. A selection signal such as that outputted from 5 is output (step 3).

In addition, if they are not equal, determine the magnitude of the horizontal correlation degree and vertical correlation degree (Step 4), and if the horizontal correlation degree is large, it is better to interpolate the interpolation pixel from the horizontal pixel. In step 5), if the correlation in the vertical direction is large, it is better to interpolate from pixels in the vertical direction. It is determined that the vertical interpolation circuit 2 is suitable, and a selection signal is outputted to cause the output of the vertical interpolation circuit 2 to be outputted from the switch 5 (step 6).

In this way, in this embodiment, since the degree of correlation between the adjacent pixel group including the pixels adjacent to the interpolated pixel and the surrounding pixel group is calculated, it is better to interpolate the interpolated pixel starting from the pixels in the horizontal direction. , it is possible to accurately determine whether it is better to perform interpolation starting from pixels in the vertical direction or using all surrounding pixels.

Then, based on such accurate judgment, the optimal pixel is selected to be used in the calculation when the interpolated pixel is calculated from the data of the adjacent pixel, that is, the selection of interpolation circuits 1 to 3 is performed. Interpolation processing can be performed.

Therefore, in this embodiment, aliasing distortion due to the above-mentioned erroneous determination hardly occurs, thereby making it possible to prevent image deterioration due to aliasing distortion.

Further, as a result of simulation experiments conducted by the present applicant, it was found that the probability of occurrence of an erroneous determination according to this embodiment is extremely low compared to the conventional method.

Furthermore, in the above embodiment, by binarizing each pixel data supplied to each correlation degree calculation circuit,
Arithmetic processing in these circuits can be simplified,
As a result, processing speed can be increased and hardware can be downsized.

Incidentally, in the above embodiment, each group is composed of four pixels, but the present invention is not limited to this. For example, as shown in FIG. 4, each group is composed of more pixels (in this case, 12 pixels), and more accurate determination can be made.

Note that in the above embodiment, two-dimensional offset and
Although the case of sampling has been described, it goes without saying that the present invention may be applied to an interpolation device when offset sampling is performed on a three-dimensional image such as a moving image.

Furthermore, in the above-described embodiments, each of the interpolation circuits 1 to 3 uses only the pixels adjacent to the interpolation pixel for calculation to obtain the interpolation pixel, but it is also possible to use surrounding pixels for this calculation.

Furthermore, in the above-mentioned embodiment, each pixel data was binarized, but in the present invention, βR is not determined for 2 (ii) conversion, but multi-value conversion (
Of course, it is also possible to perform n-value conversion), and the larger the value of n, the more accurate correlation degree detection can be performed.

(Effects of the Invention) As is clear from the above description, according to the present invention, it is possible to select an optimal interpolation means from a plurality of interpolation means for calculating and interpolating interpolation pixels from an offset sampled signal. , since the decision is made based on the correlation between multiple pixel groups consisting of pixels located around this interpolated pixel, it is better to interpolate the interpolated pixel from pixels in the horizontal direction or from pixels in the vertical direction. It is possible to accurately determine whether it is better to perform interpolation or to interpolate using all surrounding pixels.

Therefore, according to the present invention, it is possible to prevent the occurrence of aliasing distortion due to the erroneous determination of the "tendency" around the interpolated pixel as described above, and thereby it is possible to prevent image deterioration due to aliasing distortion.

Further, in the present invention, each pixel data to be supplied is
By converting the data into n-values and calculating the above-mentioned correlation degree based on this n-valued data, the arithmetic processing for calculating the correlation degree can be simplified, thereby speeding up the processing and reducing the hardware cost. It is possible to reduce the size.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment according to the present invention,
2(a) and 2(b) are pixel arrangement diagrams showing the relationship between interpolated pixels and adjacent pixel groups and surrounding pixel groups, FIG. 3 is a flowchart showing the operation of the embodiment shown in FIG. 1, and FIG. 4 is a pixel arrangement diagram showing other relationships between interpolated pixels and adjacent pixel groups and peripheral pixel groups, Figure 5 is a diagram showing a two-dimensional offset sampling structure, and Figure 6 is a diagram of the band that can be transmitted by offset sampling. Figure 7 is a diagram showing the spatial frequency spectrum; Figure 7 is a diagram schematically showing the interpolation operation for restoring the original signal from sampling data;
Figures (a) and (b) are diagrams showing the spatial frequency spectrum of aliasing components, Figure 9 is a diagram showing an example of occurrence of aliasing distortion, and Figure 10 is a diagram showing an example of occurrence of aliasing distortion.
Figures (a) to (c) are characteristic diagrams of the interpolation circuit. 1--Horizontal interpolation circuit, 2--Vertical interpolation circuit, 3--Two-dimensional interpolation circuit, 4--Interpolation means selection circuit (selection means), 5--Switch, 6.7, 8 .9...Correlation degree calculation circuit, 10.12.1
3.14.15...Binarization circuit, 11...Threshold value determination circuit. 1 5' / 251 pH ｡F =] 13th position off e7 7cJ) The 6th ward offse'...to'Tiburi)fu'Jsu, 4 yen can be sent [pan-shaped bread] ni 苧仄■ (to) Akira 8 (Treasury) ha) <b ＞O ・ ・ ・ −ゝfχ

Claims (1)

[Claims] A plurality of interpolation means with different interpolation methods are used to calculate the degree of correlation between a pixel group consisting of pixels located around the interpolated pixel and a plurality of other pixel groups located around this pixel group for each pixel group. It is characterized by comprising a selection means for detecting based on each pixel data converted into n-value (n is an integer of 2 or more) and selecting one of the outputs of each of the above-mentioned interpolation means according to the detection result. An interpolator for offset sampling signals.