JPH0681270B2 - Television image filter - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a television image filter, and particularly to an improvement of a temporal filter that processes in the time direction.

SUMMARY OF THE INVENTION The present invention treats a component signal applied to a temporal filter that performs temporal processing on an image in a non-linear manner in a frequency domain and a level domain, so that a general image looks unnatural even when used in a preprocessing filter. It is designed to eliminate.

For example, the image input by the MUSE encoder has different noise states and enhancer amounts depending on the material. If these are folded back by sub-sampling, a reproduced image will have an unnatural feeling.

Therefore, the temporal filter is used to process in the time direction to remove this unnaturalness.

That is, the present invention newly introduces a non-linear process on the input side of the temporal filter in order to improve the ghost-like multi-line blurring that occurs in the image that works when the temporal filter is used. It also removes the instability of the filter operation.

[Prior Art] Conventionally, examples used as a preprocessing filter in image processing of television include a horizontal LPF (low-pass filter) and noise reduction for reducing noise. However, the horizontal LPF only limits the band and does not include the processing in the time direction, which causes an unnatural blurring effect.

As a method of moving image processing of an image signal, a temporal filter including a transversal or recursive filter in the frame direction of the image signal is often used. This is 1
One is used as a temporal low-pass filter that does not have frame correlation, such as noise reduction, and the other is two-dimensional (sometimes one-dimensional) high-frequency component signal dropping for the moving image part. The effect is enhanced by being used as an anti-aliasing filter that does not cause aliasing in three dimensions.

FIG. 4A shows a conventional image signal moving image processing circuit.
The input image signal Sin is high pass / low pass (HP / LP) filtered before being processed by the temporal filter 200.
The high-pass (HP) component signal S _HP is a temporal filter 2
After being processed by 00, the adder 300 adds the low-pass (LP) component signal S _LP to form an image signal output Sout.

[Problems to be Solved by the Invention] However, for example, a television signal, which represents an image signal, transmits an image in frame units, and therefore the image signal is sampled and transmitted in frame time. Of course, there is a folding component based on this, and when a moving image in a frame unit is large, such as a moving image, through a normal temporal filter, for example, a transversal filter having a frame memory, FIG. ) There is a drawback that so-called multi-line blur occurs as shown in the image signal output Sout of the waveform diagram of each part. In the case of a still image, there is no particular problem as shown in Fig. 4 (c).

As a measure against this, it is said that some effect can be obtained by increasing the order of the filter used, that is, the number of taps, but it is not definitive in fact, and increasing the number of taps is especially transversal. The filter method increases the number of required frame memories, resulting in an unnecessarily increased circuit scale.

As an attempt to improve the above-mentioned multi-line blur, there is Japanese Patent Application No. 62-170 “Image signal processing method” filed by the present inventors on January 6, 1987, and the effect is obtained. Furthermore, by appropriately processing the HP (high frequency) components to which the temporal filter is applied, the capacity (scale) of the frame memory that constitutes the temporal filter is suppressed, and the natural feeling of the image and the instability of operation are further enhanced. It needs to be removed.

That is, in order to reduce the bit length of the HP component to be applied with the temporal filter and reduce the capacity of the frame memory, the temporal filter is an integral in the time direction, so the information on the LSB side of the HP component, that is, the level of the image If the information is not valued, the S / N of the edge part will decrease and the image quality will deteriorate.

Also, the time constant (or tap coefficient) of the temporal filter determines how well the filter works,
When changed non-linearly, a threshold effect occurs, the time constant changes depending on the pattern of the movement of the image, the condition as a temporal filter changes, and the image becomes unstable.

Further, there is a problem that the time constant itself cannot be increased and the effect as a temporal filter is reduced.

Therefore, an object of the present invention is to solve the above-mentioned drawbacks and problems and appropriately perform image processing to suppress an increase in the capacity of the frame memory that constitutes the temporal filter, and obtain stable operation without impairing the natural feeling. The present invention is to provide a filter for a television image.

[Means for Solving Problems] In order to achieve such an object, according to the present invention,
The HP component values the LSB side, that is, without losing the information on the LSB side, and preventing overflow and underflow processing from increasing the bit length, and the time constant of the temporal filter itself is a linear operation. Try to do

That is, the present invention limits the bit length of the separated high-frequency component signal to a predetermined level range by separating the image signal into a low-frequency component signal and a high-frequency component signal. An overflow / underflow circuit to suppress, a high-frequency component signal output from the overflow / underflow circuit, a component signal obtained by passing the high-frequency component signal through a temporal filter with a constant time constant having a low-pass filter characteristic, and two components A selection means for selecting the component signal having the smallest absolute value among the component signals of the average value of the signals for each pixel, and a synthesis means for synthesizing the output from the selection means and the low-frequency component signal It is characterized by

[Operation] According to the present invention, the bit length of the HP component of the image can be limited to a predetermined length and input to the temporal filter, the operation of the temporal filter itself is stable, and the time constant is large. It is possible to improve the effectiveness of the filter.

This makes it possible to reduce the capacity of the hardware that constitutes the temporal filter, and
The instability of the temporal filter caused by the difference in S / N and excessive enhancer can be removed, and the natural feeling can be increased.

Embodiments Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

In FIG. 1, 1 is an LPF (low pass filter) and 2 is a subtractor. An overflow / underflow circuit 3 converts the HP (high frequency) component signal of the image to a predetermined level range to limit the bit length. That is, the bit on the LSB side of the HP component that affects the accuracy of the image is passed as it is to prevent the image quality from being deteriorated, and by limiting the bit length, the capacity of the frame memory forming the temporal filter described later is suppressed.

Reference numeral 4 is a temporal filter having a low-pass filter characteristic and constant time constant, 5 is a time constant circuit, and 6 is a frame memory. The temporal filter 4 is a time constant circuit 5, a frame memory 6, a subtracter 101 and an adder 102 which will be described later. Composed of and.

7 is a minimum absolute value selection circuit, which has three input signals c, d,
Each absolute value of e is detected, and the input signal corresponding to the minimum value among the detected values is selected and output for each pixel. 8, 102 and 103 are adders, and 101 is a subtractor. Also, a, b, c, d, e
And f are the signals of the correspondingly described portions.

Next, the operation of each part will be described with respect to the present embodiment shown in FIG.

LPF1 is a filter for creating an HP component to be subjected to a temporal filter in the input signal, and the HP component b is obtained by subtracting the LP component a from the all-pass (all-pass) component in the subtractor 2.
To get Overflow / underflow circuit 3
It is input to the temporal filter 4 while preventing the bit length from increasing. The temporal filter 4 is a recursive filter using the frame memory 6, and the time constant circuit 5
Determines its characteristics. The time constant is linearly changed with respect to the input, but is linear in this embodiment. The HP component f added by the adder 8 to the LP component a is the pixel of the HP component e with the temporal filter 4 and the HP component c without the temporal filter 4, and the half component d of the sum of the two. It is selected by the absolute value minimum selection circuit 7 which selects the one having the smallest absolute value for each.

Here, the three signals c, d, and e will be described in detail. That is, when the image signal input is a still image, the signals c, d, and e are all the same signal as shown in FIG. 5 (b), and therefore the image signal input and the image signal output are completely equal. When the image signal input is a moving image, the signal with the smallest absolute value among the signals c, d and e is used as the output signal f of the absolute value minimum selection circuit 7 as shown in FIG. That is, since the signal with the least high frequency components is used, it operates as an anti-licing filter. As a result, the HP component in which the multi-line blur is suppressed is added to the LP component by the adder 8 to obtain the image signal output.

Note that the signal d may be unnecessary for the purpose of simply removing only the above-described multi-line blur, but this is necessary for the following reason. That is, although the polarities of the signals c and e are completely different depending on the type of moving image of the image signal input, the signals may have very close absolute values. At this time, | c | ＞ |
It becomes e | or | c | <| e |, but when the signals are selectively switched according to this, the signals themselves are significantly different from each other, so a large discontinuity occurs in the image and a large disturbance occurs. The image quality is significantly degraded. In order to prevent this, if the signal d is included in the options, in such a case, the signal d becomes a signal extremely close to zero, and the signal d is always selected, so that this problem does not occur.

Next, FIG. 2 is a block diagram showing the configuration of the second embodiment of the present invention.

In FIG. 2, the same parts as those in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. Reference numeral 9 is a non-linear (non-linear compression) circuit, which non-linearly compresses the output HP component signal g of the overflow / underflow circuit 3 and outputs it.
Reference numeral 12 is an LPF / HPF, which is composed of the LPF 1 and the subtracter 2 shown in the first figure, and outputs the LP component signal a and the HP component signal b, respectively.

The second embodiment of the present invention, compared to the first illustrated embodiment,
The difference is that a non-linear (non-linear compression) circuit 9 is added, and the HP component signal g limited to a predetermined level range by the overflow / underflow circuit 3 is further compressed according to the level. This acts as an inverse enhancer that suppresses excessive enhancement that causes an unnatural feeling as an image.

FIG. 3 is a block diagram showing the configuration of the third embodiment of the present invention.

In FIG. 3, the same parts as those in FIG. 2 are designated by the same reference numerals, and the description thereof will be omitted. Reference numeral 10 is a threshold circuit, which determines when the input signal of the temporal filter 4 exceeds a predetermined threshold value. A selection circuit 11 adaptively switches the output of the temporal filter 4 and the output of the absolute minimum selection circuit 7 in accordance with the output of the threshold circuit 10.

That is, the HP component selected by the absolute value minimum selection circuit 7 may be a signal after temporal filtering or a signal not subjected to temporal filtering, that is, a through signal, depending on the movement of the image in the image signal. When such a change occurs at a large edge portion of the image, the change causes blurring of the edge, which makes the image unstable. Therefore, when the absolute value of the HP component exceeds a predetermined value (threshold value of the threshold circuit 10) (that is, a large edge portion), not the output of the absolute value minimum selection circuit 7 but the output e of the temporal filter 4 as it is. By giving it to the adder 8, the absolute value minimum selection circuit 7
It is possible to avoid the instability of the image due to the change of the blur of the edge of the image due to the above change based on the selected output of.

[Effects of the Invention] As is apparent from the above, according to the present invention, the temporal filter is used to improve the image quality without losing the natural feeling of the image and stably operating with respect to the moving image. It is possible to put the television image filter that is significantly improved into practical use.

Further, the circuit can be simplified and the hardware scale can be reduced, which is extremely effective in terms of economy.

Further, according to the present invention, the application range as a preprocessing filter is wide.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of the present invention, FIG. 2 is a block diagram showing a configuration of a second embodiment of the present invention, and FIG. 3 is a configuration of a third embodiment of the present invention. 4A is a block diagram showing a configuration relating to a conventional image processing method, FIGS. 4B and 4C are waveform diagrams of respective parts in the case of a moving image and a still image, and FIG. 3A and 3B are waveform diagrams of respective parts in the case of a moving image and a still image according to an embodiment of the present invention. 1 ... LPF, 2,101 ... subtractor, 3 ... overflow / underflow circuit, 4 ... temporal filter, 5 ... time constant circuit, 6 ... frame memory, 7 ... absolute value minimum selection circuit, 8,102,103 ... … Adder, 9 …… Nonlinear compression circuit, 10 …… Threshold circuit, 11 …… Selection circuit, 12 …… LPF / HPF.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Seiichi Koshi 1-10-11 Kinuta, Setagaya-ku, Tokyo Inside the broadcasting technology laboratory of Japan Broadcasting Corporation (72) Yoshiaki Kakuro 1-10 Kinuta, Setagaya-ku, Tokyo No. 11 within Japan Broadcasting Corporation Broadcasting Technology Laboratory (56) References Japanese Patent Laid-Open No. 63-169192 (JP, A)

Claims (3)

[Claims] 1. A separation means for separating an image signal into a low-frequency component signal and a high-frequency component signal, and a bit length of the high-frequency component signal separated by the separation means is limited to thereby generate the high-frequency component signal. An overflow / underflow circuit for suppressing a predetermined level range, a high-frequency component signal output from the overflow / underflow circuit, and a high-frequency component signal obtained by passing the high-frequency component signal through a temporal filter having a low-pass filter characteristic and a constant time constant. Selection signal for selecting, for each pixel, the component signal having the smallest absolute value from among the component signal and the component signal that is the average value of the two component signals, and the output from the selection device and the low-frequency component signal. And a synthesizing means for synthesizing a television image filter. 2. The television image filter according to claim 1, wherein the high frequency component signal applied to the temporal filter is non-linearly compressed. 3. The television image filter according to claim 1 or 2, wherein when the high frequency component signal applied to the temporal filter exceeds a predetermined threshold value, the selecting means. And a component signal from the temporal filter is supplied to the synthesizing means instead of the output from the television image filter.