JPH02312381A - Method of interpolating scanning line - Google Patents

Abstract

PURPOSE:To set a picture to be smooth by interpolating a scanning line in a field not only from an oblique direction but also from upper and lower directions by detecting the degree of change in picture elements in respective directions and selecting the element least in change so as to decide from which direction the scanning line is interpolated. CONSTITUTION:Picture elements (x) which come to be the object of interpolation are picture elements (a)-(e) on an upper scanning line and picture elements (f)-(g) on a lower scanning line. The degree of the change in the picture elements in plural pairs of picture elements of point symmetry which are in point symmetry positions is detected by setting the picture element to be interpolated on the scanning line to be interpolated in the picture element on the upper scanning line and the lower scanning line, both of which are positioned above and below the scanning line to be interpolated, as a center. Then, the interpolation value of the picture element to be interpolated is decided by using the picture element value of plural pairs of picture elements of point symmetry whose detection value is the smallest. Thus, a smooth oblique edge is obtained and stepwise distortion is improved.

Description

[Detailed Description of the Invention] [Industrial Field of Application] The present invention is a method for converting interlaced television signals into non-interlacing and converting the number of scanning lines in equipment that handles images such as television receivers. Concerning line interpolation methods.

[Conventional technology]

Currently, the color television standard system commonly used in television broadcasting is interlaced, and in contrast to non-interlaced, in which all scanning lines are sent in one field, every other scanning line is thinned out. It has become a thing. In this case, the spatiotemporal frequency band that can be transmitted is
As shown in FIG. 4, it has a rhombic shape in the temporal frequency (f) and vertical frequency (ν) regions, and has a transmission band that is more suitable for viewing angle characteristics. However, the spatiotemporal frequency band of the original signal is not necessarily appropriately limited, and the band limitation by the receiver or vision is not sufficient, so the image contains many aliased components.

Therefore, the vertical high-frequency component becomes a temporal high-frequency aliasing component, which obstructs viewing as line flicker.

This also increases the vertical resolution to 7
Only about 0% can be obtained.

A motion adaptive scanning line interpolation method is available as a means to solve such problems. This solves the above problem in static areas by making the image displayed on the monitor non-interlaced, and scanning lines thinned out by interlacing are captured between frames when the image is still. It is. An example of this is the "signal processing circuit" (Japanese Patent Laid-Open No. 61-3
No. 2681). This is an interpolated value that is obtained by adding the pixels before and after one frame in a still image, and by adding the pixels above and below within a field in a moving image and dividing the result by "2".

As a result, in a still image, the same non-interlaced signal before thinning out is reproduced by interpolation between frames, and a flicker-free image with high vertical resolution can be obtained. On the other hand, in a moving part of the image, intra-field processing is performed by motion detection, so that 2ff11 etc. do not occur.

By the way, such scanning line interpolation is used not only for non-interlacing, but also when converting the number of scanning lines from NTSC to HDTV, etc., or when one field of a moving image is used as a still image, by interpolating one field to create a frame. It is also used when forming images.

On the other hand, in efficiency encoding of moving images, there is a method of predicting one field from the other and encoding the prediction residual. Therefore, it is conceivable to use the field interpolation signal as the pre-4-1 signal in such predictive coding as well. In this case, by performing more appropriate prediction, the preliminary full residual can be reduced and a high data compression rate can be obtained.

[Problem to be solved by the invention]

However, in the above-mentioned conventional blood complementation method, processing of the moving region involves interpolation from the upper and lower pixels, so depending on the shape of the image, the vertical resolution in the moving region may not be improved, and the sharpness may decrease due to interpolation. Ta.

Furthermore, ``interlaced signals usually have a fourth
There is no spatiotemporal band limitation as shown in the figure, and the image of one field simply has scanning lines thinned out one by one, and includes aliasing components. Therefore, in the case of field inner sleeve, the diagonal edge is the fourth
The image becomes step-like as shown in Figure (a), resulting in significant visual deterioration.

Furthermore, if the transmission side is band-limited by spatiotemporal frequency, the resolution of the step-like distortion will decrease, resulting in rounded edges.

SUMMARY OF THE INVENTION Therefore, it is an object of the present invention to provide a scanning line interpolation method that performs interpolation between fields in accordance with the shape of an image.

[Means to solve the problem]

The scanning line interpolation method of the present invention detects the degree of change between pixels in each of a plurality of pixel sets consisting of pixels located in a predetermined direction from the interpolated symmetrical pixel, and detects the degree of change between the pixels among the plurality of pixel sets. is characterized in that the interpolation value of the interpolation target pixel is determined using the pixel value of the smallest pixel.

[For production]

According to the present invention, since the lines connecting the pixels of each pixel group have different directions, determining the interpolation value by looking at the degree of change between the pixels and using the one with the smallest change is a By finding out in what direction the edge of the image is changing, it is possible to determine the interpolation value so that the pixel to be interpolated is placed on the line of that edge, and it is possible to perform interpolation according to the shape of the image.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a conceptual diagram of an interpolation method according to an embodiment of the present invention.

First, in this embodiment, from the lower scanning line located above and below the interpolation symmetrical scanning line and the pixels of the lower scanning line that are located at points symmetrical positions with respect to the interpolation symmetrical pixel of the interpolation symmetrical scanning line, Detect the degree of change between pixels in each of a plurality of point-symmetric pixel pairs, and use the pixel value of the one whose detected value is the smallest among the point-symmetric pixel pairs to determine the interpolated value of the captured symmetric pixel. do.

The situation is shown in FIG. The pixels X to be interpolated are pixels a, b, c, and d of the lower scanning line.

e, and the pixels of the lower scanning line f, g, h, t,
j and changes in each direction 1a-jl, 1b-it, 1c-
hl.

Find ld-gl and 1e-fl, and when 1a-jl is the smallest among the changes in 5 directions, (a+j)/2, Ib-
When 1f is the minimum, use (b+i)/2, when 1c-hl is the minimum, use (c+h)/2, when ld-gl is the minimum, use (d+g)/2, and when Ie-fl is the minimum, use (e+f). /2 is determined as the interpolation value.

For example, in the case of FIG. 3, since ld-gl is minimum at the edge portion, (d+g)/2 becomes the interpolation value of the interpolation target pixel X. As a result, smooth diagonal edges can be obtained as shown in FIG. 2(b). The interpolation value depends on the shape of the edge; for edges that extend vertically, it is (c + h)/2, and for edges that are close to the horizontal direction, it is (a + j) / 2 or (e + f).
/2 becomes the interpolated value of the interpolated symmetrical pixel X.

Therefore, whereas conventionally, (c + h) / 2 was used as the fixed interpolation value, appropriate interpolation is performed according to the direction of image change, and the step-like problem that has been a problem in the past is solved. Distortion is improved. This means that the vertical resolution of the moving area is improved.

FIG. 2 is a block diagram of the apparatus used to implement such an interpolation method.

In this figure, a signal input from an interpolation signal input terminal 10 is delayed by a time corresponding to one pixel (sample interval T) in pixel delay devices 12, 14, 16, and 18, and then
The delay device 20 delays one horizontal scanning line (H) by 4 pixels, and the pixel delay devices 22, 24, 26, 2
8, each pixel is delayed by one pixel.

The input signal at the interpolation signal input terminal 10 is the pixel j in FIG.
, the outputs of pixel delay devices 12.14 and 16.18 are t,
h, g, f, the output of the delay device 20 is e, and the output of the pixel delay devices 22, 24, 26.28 is d, c,
This applies to b and a, respectively. These respective outputs are converted to (a+j)/2. by adders 30, 32, 34, 36.
(b+f)/2. (c+h)/2, (d+g)/2
．． (e+f)/2 is obtained. On the other hand, subtractors 40, 4
2,44,46.48 (a j), (b-i),
(c-h), (d-g), (e-f) are obtained, and 1a-j is obtained using an absolute value converter 50°52.56, 58.60.
f.

1b-it, 1c-hl, I d-g I.

It is converted to Ie-fl.

The outputs of the adders 30, 32, 34, 36.38 are divided by 2 and input to the selector 60, and the outputs of the adders 30, 32, 34, 36.
The output of 6.58 is input to the minimum value detector 62. The minimum value detector 62 determines which one is the minimum among the five types of input signals and sends this information to the selector 60. The selector 60 selects one of the five types of signals input based on the information, and outputs it from the interpolated value output terminal 64.

In motion-adaptive scanning line interpolation, the intra-field interpolated values thus created and the inter-frame interpolated values are mixed while changing the mixing ratio by motion area detection to form an interpolated scanning line. Furthermore, in non-interlacing, a non-interlaced signal can be obtained by time-compressing interpolated scanning lines and originally existing scanning lines to alternately form scanning lines.

In the above embodiment, a plurality of point symmetric pixels centered around the interpolation symmetric pixel are used as the interpolation drawing set, but the present invention is not limited to this.

That is, in the case of FIG. 1, first, whether a point-symmetric relationship is obtained between the pixels of the upper scanning line and the pixels of the lower scanning line with the interpolation symmetric pixel X as the center, or in the case of field offset sampling, the sample Offsets between the scan lines of points prevent such point symmetry relationships. For example, even if a point-symmetrical relationship cannot be obtained between pixels a to e on the upper scanning line and pixels f to j on the lower scanning line, changes can be detected using these pixels in the same way as in the above embodiment. and can be used for interpolation.

Further, in the above embodiment, the interpolation of the interpolation symmetrical scanning line is performed using the scanning lines located above and below it, but it may be performed using the scanning line on either side.

In this case, for example, two scanning lines located above the interpolation symmetrical scanning line may be used, and changes may be detected from two pixel values located in the same direction with respect to the interpolation symmetrical pixel.

〔Effect of the invention〕

As explained above, according to the present invention, scanning line interpolation within a field is performed not only in the vertical direction but also in the diagonal direction, and the direction from which interpolation is performed is determined by detecting the degree of change in pixels in each direction. However, by selecting the one with the least amount of change, the step-like distortion of diagonal edges, which has been a problem in the past, is eliminated, resulting in a smooth image. This means that the vertical resolution of the moving area is improved. As a result, motion-adaptive scan line interpolation improves image quality even in video processing, and seamlessly integrates with still processing.

This method is useful not only for non-interlacing, but also for forming a single field when a moving image is frozen in a field,
It can also be used as an inter-field prediction method in converting the number of scanning lines and in high-efficiency encoding of images.

When applied to this inter-field prediction, as explained using the example in Figure 3, conventionally, the edge part becomes step-like, so two pixels (hatched pixels) must be encoded per scanning line. However, the method of the present invention eliminates the need for encoding. Therefore, this greatly contributes to improving the compression ratio.

Furthermore, since there is no increase in field memory or line memory compared to the conventional example, it is easy to implement it into an LSI.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of an interpolation method according to an embodiment of the present invention, and FIG.
The figure is a block diagram of a device used for implementing the method, FIG. 3 is a comparison diagram of interpolation results between the conventional method and the method of the present invention, and FIG. 4 is a diagram of spatio-temporal characteristics of an interlace signal. 10... Interpolation signal input end, 12, 14, 16° 18.
22, 24, 26.28... Pixel (1T) delay device, 2
0...(iH-47) Delay device, 30,32°34.3
6.38... Adder for calculating interpolated value candidates, 40.4
2.44, 46.48...Subtractor for determining the degree of change between pixels in a pixel pair, 50.52, 54, 56.
58... Absolute value converter, 60... Interpolated value selector, 6
2: Minimum value detector, interpolation value output terminal, X: Interpolation target pixel, a-j: Pixel forming the interpolation pair pixel group. Applicant's agent Mr. Sato (α) Conventional example (b) Hontefan Figure 3 is shown in Figure 4

Claims (1)

[Claims] In intra-field interpolation of scanning lines, the degree of change between pixels is detected in each of a plurality of pixel sets each consisting of pixels located in a predetermined direction from the interpolation target pixel, and the degree of change between the pixels is detected, A scanning line interpolation method characterized in that the interpolation value of the interpolation target pixel is determined using the pixel value of which the detected value is the smallest.