JPH0323789A - Movement adaptive processor - Google Patents

Abstract

PURPOSE:To attain accurate movement detection with simple constitution by converting a movement detection signal into the movement detection signal with an interpolation signal mixing movement detection sensitivity characteristic and an edge detection correction movement detection sensitivity characteristic. CONSTITUTION:A movement detection circuit 14 takes inter-frame difference from Y0 and Y2 to obtain a movement detection signal with one nonlinear movement detection sensitivity characteristic being the synthesis of the movement detection sensitivity characteristic for interpolation signal mixing and the movement detection sensitivity characteristic for edge detection correction. The movement detection signal is fed to a control circuit 22 comprising of a field memory 20 and a maximum value selection circuit 21, expanded temporarily and sent to a conversion circuit 27. Then the signal is converted into a movement detection signal according to an edge discrimination signal from a movement component elimination circuit 26 and the converted movement detection signal is fed to a mixing circuit 17. Moreover, the movement detection signal outputted from the control circuit 22 is fed to the movement component elimination circuit 26 and the movement component is eliminated from the inter-field edge detection signal. Thus, accurate movement is detected with simple constitution.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) This invention detects moving parts and still parts of a television image, and controls the processing of a television signal according to the detection results. The present invention relates to an adaptive processing device.

(Prior art) In recent years, in order to improve the quality and definition of television images, a completely new method of high-definition television (HDT) has been developed.
In parallel with the research and development of V), research and development of image quality improvement using digital signal processing technology is also active in the current system. Typical of these are the progressive scan (non-interlaced) conversion method and the adaptive Y/C conversion method, but both methods require a motion detection circuit to detect moving and still parts of the image. The processing method is switched depending on the output of this motion detection circuit.

The non-interlaced conversion method will be explained below as an example, but the image quality can also be improved using a similar method in the adaptive Y/C separation method.

The non-interlace conversion method eliminates interference such as line flicker and line crawling that are problems with current methods by converting television signals transmitted in interlace (interlaced scanning) to non-interlace (sequential scanning). This is a method to do so. Specifically, with reference to FIG. 3, the principle of converting 525 interlaced images into 525 non-interlaced images using an image memory in the NTSC system is as follows.

First, for a still image, the current field is set to "0", and the data of the field "-1" one field before is stored in the image memory, and as shown by arrow a in Figure 3, the current field is The data of "0" and the data of the field "-1" one field before are output at the same time. That is, interfield interpolation is performed in which calculations are performed only in the time axis direction t. With this, both still images and moving images,
It is possible to convert 525 interlaced images into 525 non-interlaced images, and remove interlace disturbances such as line flicker and large screen flicker.

Here, FIG. 4 shows a conventional motion adaptive non-interlace conversion device constructed based on the above principle.

The input terminal 1l receives a luminance signal Y of the interlaced television signal. is supplied. This luminance signal Y
. is delayed by one field by the first field memory l2, and then further delayed by one field (total of 2 fields-1 frame) by the second field memory l3.

Input terminal! The luminance signal Y supplied to l. and the luminance signal Y2 output from the field memory l3 is 1, respectively.
The signal is supplied to a motion detection circuit 28 that can detect the difference between frames, and is used to generate a motion detection signal. Further, the luminance signal Y0 and the luminance signal Y outputted from the first field memory l2 are supplied to an interfield interpolation signal generation circuit l5 and used to generate an interfield interpolation signal. Furthermore, a luminance signal Y. is the interline interpolation signal generation circuit l6
and used to generate an inter-line interpolation signal.

Both the inter-field interpolation signal and the inter-line interpolation signal are supplied to the mixing circuit 17 and mixed at a ratio according to the motion detection signal. The output of the mixing circuit 17 is supplied to the non-interlaced conversion circuit 1B, and a luminance signal Y is generated. The signal is converted into a non-interlaced signal based on the signal, and is output from the output terminal 19.

Here, in order to adapt the non-interlaced conversion to the motion of the image, a motion detection circuit 28 takes the difference between one frame and generates a motion detection signal. However, it is not always possible to detect motion simply by taking the difference between one frame.

Now, in the 3fI6 diagram, the horizontal axis is the horizontal position of the image,
If the vertical axis is the level, the motion detection signal obtained by simply taking the difference between one frame (the difference between the "0" field and the "-2" field) is the one that detects motion in the α and γ parts. However, the portion of β (corresponding to the Cr-lJ field) is not detected even though there is movement. In this case, the β portion of the motion detection signal has no movement, that is, it is processed as a still image, resulting in a problem of two-line blur on the screen.

Therefore, conventionally, as shown in FIG. 4, a control circuit 22 consisting of a field memory 2o and a maximum value selection circuit 2l is interposed between the motion detection circuit 2B and the mixing circuit I7, and the current motion detection signal and one field The higher level of the previous motion detection signal is output. In this way, the motion detection signal can be tensorally expanded as shown in FIG. 6, and loss of the portion β can be prevented.

In addition, if the motion detection sensitivity is increased to reduce the problem of two-line blurring, flickering occurs at the edges of still images due to moving image judgment due to noise, etc. lower the detection sensitivity. It is effective to switch the motion detection sensitivity between the edge portion of the image and the flat portion other than the edge portion of the image using a motion detection signal that has been tensorally expanded.

Here, FIG. 5(b) shows motion detection sensitivity characteristics for flat areas. This is an example in which the motion detection signal is expressed in 2 bits, and when the inter-frame difference is close to "0", the motion detection signal also becomes "0", and the image is determined to be a completely still image.

As the inter-frame difference increases, the motion detection signal also increases, and when the inter-frame difference exceeds a certain value, the motion detection signal becomes "3" and it is determined that the video is a complete moving image.

FIG. 5(c) shows motion detection sensitivity characteristics for edge portions, and the sensitivity for moving image determination is low. FIG. 5(a) shows an additive combination of the motion detection sensitivity characteristics for flat areas and the motion detection sensitivity characteristics for edge areas. In other words, the motion detection signal from rOJ to "3" is the sensitivity characteristic for the flat part, and from "4" to "7" is the sensitivity characteristic for the edge part, and both can be distinguished by looking at the value of the MSB bit. be. By making a determination based on such a principle, it is possible to convert two 2-bit signals into one 3-bit signal, thereby reducing the number of signals.

The motion detection signal obtained with the sensitivity characteristics shown in FIG. 5(a) is supplied to the control circuit 22 shown in FIG. 4, and is tensorally expanded. This protects against missing motion detection. The motion detection signal output from the control circuit 22 is
The signal is supplied to a conversion circuit 27 constituting the motion determination circuit 23 (the motion determination circuit 23 includes a motion removal circuit 26 and a conversion circuit 27).

The conversion circuit 27 is configured as shown in FIG.
The motion detection signal shown in FIG. 5(a) is converted into the characteristics shown in FIG. 5(C) (for edge portions) or FIG. 5(b) (for flat portions) according to the edge determination signal. ing. However, the edge determination signal output from the motion component removal circuit 26 is
It is assumed that it is "1" when an edge portion is detected, and "0" otherwise.

In FIG. 7, AND circuits G and ~G2 each input the lower two bits of the motion detection signal separately, output the input bit signal as it is when the common gate signal is "1", and the common gate signal is When the bit signal is "0", the input bit signal is cut off and all outputs are set to "0". The common gate signal is obtained by inverting the edge determination signal by an inverting circuit G, and then ORing the resulting signal and the MSB bit signal of the motion detection signal. , and generate it by performing a logical sum.

That is, the gate signal output from the OR circuit G becomes "1" when the motion detection signal is "3" or less (MSB-0) and the edge determination signal is "0", and the gate signal is output from the AND circuits 01 to G2.
outputs the lower two bits of the motion detection signal. Here, when an edge is detected and the edge determination signal becomes "1", the common gate signal becomes "0", and the AND circuits 61 to G
Force the output of 2 to ``0''. Further, when the motion detection signal is "4" or more (MSB-1) and the edge determination signal is rOJ, it becomes "1" and causes the AND circuits G to G2 to output the lower two bits of the motion detection signal. Even if an edge is detected here and the edge determination signal becomes "1", the common gate signal remains "1", and the AND circuit G l- 0
The output of 2 remains unchanged.

On the other hand, the OR circuits G and ~G4 are each an AND circuit 61
The bit signals output from ~G2 are input separately, and when the common gate signal is rOJ, the input bit signals are output as they are, and when the common gate signal is "1", the outputs are all "1". The common gate signal is generated by inputting a signal obtained by inverting the edge determination signal by the inverting circuit G and the MSB bit signal of the motion detection signal to the AND circuit G7, and performing a logical product.

That is, the gate signal output from the AND circuit G7 becomes "0" when the motion detection signal is "3" or less (MSB-0) and the edge determination signal is "0", and the gate signal is output from the AND circuit c, -G.
Each bit signal from 2 is output as is. here,
Even if an edge is detected and the edge determination signal becomes "1", the common gate signal remains "0" and the AND circuit 01
The output of ~G2 is directly derived from OR circuits G and ~G4. Also, the motion detection signal is "4" or more (MSB-1)
When the edge judgment signal is “0”, the common gate signal is “1”
'', and the outputs of OR circuits G, to G4 are all set to ``1''. When an edge is detected here and the edge determination signal becomes "1", the common gate signal becomes "0", and the AND circuit G
The output of l-02 is directly derived from the OR circuits 61 to G4.

As can be seen from the above, the conversion circuit 27 converts the motion detection signal, which has been expanded by 1 bit and has the characteristic shown in FIG. For the characteristics shown in Figure 5 (c),
In the flat portion, the signal is converted to the characteristic shown in FIG. 5(b) and is supplied to the mixing circuit 17 as the final motion detection signal for interpolation signal mixing. In the mixing circuit 17, when the value of the motion detection signal is K, the inter-field interpolation signal is mixed at a ratio of (1-(K/3)) and the inter-line interpolation signal is mixed at a ratio of (K/3).

Next, edge determination will be explained. As shown in FIG. 4, the edge detection circuit 24 is supplied with the luminance signal yo and performs intra-field edge detection. As shown in Figure 3, take the difference between L2 and L1 delayed by one horizontal period,
If the difference exceeds a predetermined value, an intra-field edge detection signal is output.

Here, the detected edges are 52574[
Since only components up to cphl can be detected, 525/4
Edges of [cph] or more are detected by the edge detection circuit 25. The edge detection circuit 25 receives a luminance signal Y0.
and the luminance signal Y output from the first field memory l2, and the difference is taken between L2 and L shown in FIG. 3. If the difference exceeds a predetermined value, an edge detection signal is detected. Output. This makes it possible to detect edge components up to 525/2 [eph], resulting in more accurate edge detection. However, the edge detection circuit 2
Since the edge detection signal detected in step 5 is detected by pixels having a time difference of one field, it also includes a motion component. Therefore, the motion correction circuit 26 removes the motion correction from the inter-field edge detection signal output from the edge detection circuit 25.

Next, motion detection of pixels L2, L, which performs inter-field edge detection will be described. Luminance signal Y.

and the luminance signal Y2 output from the second field memory.
is supplied to the motion detection circuit 29, and the difference between frames is taken.
A motion detection signal is generated with the characteristics shown in FIG. 5(d).

The motion detection signal is supplied to a field memory 30, a maximum value selection circuit 31, or a certain control circuit 32 as shown in FIG. is removed and supplied to the motion component removal circuit 2B as a motion detection signal for edge detection correction.

FIG. 8 shows an example of the configuration of the motion component removal circuit 2B. When there is movement in the pixel L2, L, which performs edge detection,
The motion detection signal becomes "1", and the inversion circuit G. The output of
0". Then, the output of AND circuit Gl2 becomes "0", and even if the inter-field edge detection signal is input (level "1"), the output of AND circuit Gl2 is "0", so the motion component is removed from the inter-field edge detection signal. be done. The OR circuit Gl3 performs the logical sum of the inter-field edge detection signal from which the motion component has been removed and the intra-field edge detection signal, and outputs the result as an edge determination signal.

However, in the configuration described above that generates two motion detection signals, one for interpolation signal mixing and one for edge detection correction, a large number of field memories are required to protect against loss of motion detection signals, which increases the size of the circuit and reduces cost. The problem arises that this results in financial disadvantage.

(Problems to be Solved by the Invention) As described above, in the conventional motion adaptive processing device, in order to generate two motion detection signals for interpolation signal mixing and edge detection correction, loss of motion detection signals is protected. It is necessary to provide a large number of field memories for this purpose, which poses problems such as increased circuit size and economical disadvantages.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an economically advantageous motion adaptive processing device that can perform accurate motion detection with a simple configuration.

[Structure of the Invention] (Means for Solving the Problems) The present invention is capable of generating one motion detection signal using one nonlinear characteristic that is a combination of motion detection sensitivity characteristics for interpolation signal mixing and motion detection sensitivity characteristics for edge detection correction. After reducing the number of signals and delaying the field, the synthesized motion detection signal with nonlinear characteristics is converted into a motion detection signal with motion detection sensitivity characteristics for interpolation signal mixing and motion detection sensitivity characteristics for edge detection correction. It is configured to do so.

(Function) The above means reduces the number of motion detection signals, so less memory is needed for field delay, and accurate motion detection can be performed without compromising accuracy, which is also economically advantageous. It can be done.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings. FIG. 1 shows an embodiment of the present invention. However, in FIG. 1, the same parts as in FIG. 4 are denoted by the same reference numerals, and the explanation thereof will be omitted.

FIG. 1 shows a configuration in which the present invention is applied to a motion adaptive non-interlaced conversion circuit, in which a motion detection circuit {4 represents the luminance signal Y; The interframe difference is taken from Y2 and Y2, and a motion detection signal is obtained by one nonlinear motion detection sensitivity characteristic that combines the motion detection sensitivity characteristic for interpolation signal mixing and the motion detection sensitivity characteristic for edge detection correction. This motion detection signal is transmitted to the field memory 20 and the maximum value selection circuit 2.
1, the signal is sent to a control circuit 22 consisting of 1, temporally divided, and sent to a conversion circuit 27.

According to the edge determination signal from the motion component removal circuit 2B, this conversion circuit 27 converts an edge portion into a motion detection signal with a motion detection sensitivity characteristic for an edge portion, and a flat portion into a motion detection signal with a motion detection sensitivity characteristic for a flat portion. Even if the motion detection signal for edge detection correction is synthesized, the conversion can be performed without any problem, and the motion detection signal thus converted is supplied to the mixing circuit 17. Further, the MSB bit and soto of the motion detection signal outputted by the control circuit 22 is supplied to a motion component removal circuit 26, and the motion component is removed from the inter-field edge detection signal.

More specifically, first, the nonlinear motion detection sensitivity characteristic of the motion detection circuit 14 is as shown in FIG.
The motion detection sensitivity characteristics for flat areas shown in Figure (b), the motion detection sensitivity characteristics for edge areas shown in Figure (C), and the motion detection sensitivity for edge detection correction shown in Figure (d). It is an additive combination of the characteristics. That is, the motion detection signal is rOJ
From "3" to "3" are for flat areas, from "4" to "7",
This is the sensitivity characteristic for the edge portion, and the MSB bit is the sensitivity characteristic for edge detection correction.

The motion detection sensitivity characteristics for flat areas shown in Figure 5(b) are such that the difference between frames is 3LSB (when the video input signal is 8 bits) so that even motion in a pattern with a small luminance difference from the background can be detected. When it is above, it is determined to be moving, and when it is less than that, it is determined to be stationary. If this r3 L S BJ is increased, the motion determination becomes insufficient and the ratio of the inter-field interpolation signal increases, causing problems such as two-line blurring and jagged edges.

The motion detection sensitivity characteristics for the edge portion shown in FIG. 5(c) are such that the motion detection sensitivity is lowered so that changes in the luminance signal due to fluctuations in the input video signal are not determined as motion, and the difference between frames is 1.
When it is 6LSB or less, it is determined to be stationary, and if rl6LsBJ is reduced, it becomes easier to determine that it is moving, and the ratio of the in-field signal at the edge of the still image increases, causing a problem of flickering at the edge. do.

Here, the characteristic shown in FIG. 5(a) is that when a flat portion is determined by the edge determination as described above, the 3-bit motion detection signal is subjected to overflow processing for values of "4" or more, and as shown in FIG. 5(b). ), and if it is determined to be an edge part, the value of "3" or less of the motion detection signal is subjected to underflow processing, and the MSB bit is further deleted, resulting in the characteristic shown in Figure 5 (C). is converted to Therefore, Fig. 5(e)
The characteristics shown in FIG. 5(b) and FIG. 5(
It can be converted into the characteristics shown in c).

From Figure 5(a) and Figure 5(e), the difference between frames is 6LSB.
Between 16LSB and 16LSB, the motion detection signal is "3" and "4".
It can be seen that either value is acceptable. In other words, M
The SB signal is rOJ, 17LS for OLSB to 5LSB.
If it is B or higher, it is sufficient if it is "1".

FIG. 5(d) shows motion detection sensitivity characteristics for edge detection correction, in which it is determined that there is movement when the inter-frame difference is greater than 9 LSB, and it is determined that the frame is stationary when the difference is greater than 9LSB. This value of 9LSB is an intermediate value between the threshold value of the flat part characteristic and the threshold value of the edge part characteristic, and is also experimentally the best value.

If this threshold value is made smaller than 9LSB, the motion detection sensitivity for edge detection correction increases, so not only motion components but also edge components are removed from the inter-field edge detection signal, resulting in insufficient edge detection and image The motion detection sensitivity is for flat areas even at the edge portions, so flickering occurs at the edge portions. Furthermore, if the threshold value is made larger than 9LSB, the motion detection sensitivity for edge detection correction will be lowered, making it impossible to sufficiently remove motion components from the inter-field edge detection signal. , and two-line blurring occurs.

The characteristic of FIG. 5(d) is the same as the above-mentioned
In this case, rOJ satisfies the condition of "1" for 17LSB or more, and can be replaced with the MSB bit of the motion detection signal for interpolation signal mixing in Figure 5(a) or Figure 5(e), and the MSB bit in Figure 2 Becomes a characteristic.

The motion detection signal obtained by this sensitivity characteristic is supplied to the control circuit 22 and temporally expanded. This protects against loss of motion detection. The motion detection signal output from this control circuit 22 has all three bits as interpolation signals. The signal is supplied to the conversion circuit 27 as a motion detection signal for mixing, and the MSB is supplied to the motion component removal circuit 2B as a motion detection signal for edge detection correction.

Therefore, according to the configuration of the above embodiment, the motion detection signal for edge detection correction is combined with the motion detection signal for interpolation signal mixing, which has characteristics for edge portions and flat portions at the same time. It is not necessary to provide two systems of field memories for delaying each motion detection signal, resulting in a smaller size and accurate motion detection, which is also economically advantageous.

In the above embodiment, a non-interlaced conversion method has been described, but a Y/C separation method can be similarly implemented, and various modifications can be made without departing from the gist of the invention.

[Effects of the Invention] As described above, according to the present invention, it is possible to obtain an economically advantageous motion adaptive processing device that can perform accurate motion detection with a simple configuration.

[Brief explanation of drawings]

FIG. 1 is a block configuration diagram showing an embodiment of a motion adaptive non-interlace conversion device as a motion adaptive processing device according to the present invention, and FIG. 2 is a characteristic diagram showing motion detection sensitivity characteristics by the motion detection means of the same embodiment. , FIG. 3 is a diagram showing the principle of converting an interlaced television signal to non-interlace, FIG. 4 is a block diagram showing a conventional motion adaptive non-interlace conversion device, and FIGS. 5 and 6 are respectively conventional FIG. 7 is a circuit diagram showing a specific configuration of a conversion circuit for switching the detection sensitivity characteristics of a conventional motion detection signal, and FIG. 8 is a circuit diagram of a conventional motion detection signal. FIG. 2 is a circuit diagram showing a specific configuration of a motion component removal circuit. l2, l3...Field memory, l4...Motion detection circuit, l5...Interfield interpolation signal generation circuit, 1
6... Line interpolation signal generation circuit, 17... Mixing circuit, 18... Non-interlace conversion circuit, 20...
・Field memory, 2l... Maximum value selection circuit, 22
...Control circuit, 24, 25...Edge detection circuit, 2
6...Motion component removal circuit, 27...Conversion circuit.

Claims (1)

[Scope of Claims] A first storage means that stores a television signal for n frames (n is an integer) and outputs it with a delay; a television signal input to the first storage means; and the first storage means. means for extracting a difference signal from a television signal outputted from the means; a first nonlinear characteristic in which the difference signal is input and the output changes within a range where the difference signal is smaller than a first predetermined value; a second nonlinear characteristic in which the output changes in a range larger than a second predetermined value that is larger than the first predetermined value;
The differential signal has a nonlinear characteristic that is a combination of a third nonlinear characteristic in which the output changes within a range greater than the first predetermined value and smaller than the second predetermined value, and the differential signal is converted into a third nonlinear characteristic based on the nonlinear characteristic. a motion detection means for converting the first motion detection signal into a motion detection signal; a means for obtaining a change point detection signal for detecting a change point in the picture pattern of the television signal; a control circuit whose main component is a second storage means that stores and outputs a delayed signal; Said second
A third motion detection signal based on the non-linear characteristic of a conversion circuit that selectively outputs the motion detection signal of No. 3 and the fourth motion detection signal; and a signal processing section that controls processing of the television signal based on the motion information output from the conversion circuit. A motion adaptive processing device characterized by comprising: