JPH0681301B2 - Television signal reproduction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to the reproduction of a high-definition television video signal,
The present invention relates to a television signal reproducing method for detecting a moving portion of a video and interpolating the television signal.

[Prior Art] One of the methods for band-compressing a television signal is a multi-subsample transmission method using inter-field and inter-frame offset sub-sampling. This system includes, for example, MUSE (Multiple Sub-Nyquist Sa), which is the current high-definition television signal multiple sub-sample transmission system.
mpling Encoding) method, and band compression can be effectively realized.

In this multiple sub-sample transmission system,
The amount of movement of the target frame with respect to the comparison frame preceding in time is detected, and the still image and the moving image are mixed by the mixing ratio which is continuously changed according to the detected amount, and the reproduced image is reproduced. Is getting

Hereinafter, a reproducing method of the conventional multiplex sub-sample transmission system will be described with reference to FIGS. 4 and 5. To simplify the explanation, a still image is selected when the amount of motion is less than or equal to an appropriate threshold value, and a moving image is selected in other cases, and a reproduced image is created by switching and mixing these still images and moving images. Shall be obtained.

In FIG. 4, first, the analog / digital (A / D) converter 10 converts the analog television input signal to 16.2.
Sampling at MHz and converting to digital signal.

Blocks 12 to 20 are still image circuits. flame·
The memory 12 stores the digital television signal from the A / D converter 10 and generates a signal delayed by one frame.

The motion compensation circuit 14 compensates the signal from the frame memory 12 according to the motion vector. Switch 16
The output signals of the A / D converter 10 and the motion compensation circuit 14 are alternately selected at 32.4 MHz alternately to perform interframe interpolation.

The sampling frequency converter 18 sets the sampling frequency of the signal from the switch 16 interpolated between frames to 32.4 MHz.
To 48.6MHz. The inter-field interpolation circuit 20 is
The output signal of the sampling frequency converter 18 is interpolated between fields.

On the other hand, blocks 22 to 24 are moving picture circuits. The field interpolation circuit 22 interpolates the output signal of the A / D converter 10 in the field. The sampling frequency converter 24, like the sampling frequency converter 18, changes the sampling frequency of the output signal of the field interpolation circuit 22 from 32.4 MHz to 48.
Convert to 6MHz.

Blocks 26 to 34 are circuits of a motion detection system. The differential low-pass filter (LPF) 26 is a signal which is corrected by the motion correction circuit 14 according to the motion vector and which is one frame before.
Receives the current frame signal from the D converter 10, limits the band to 4MHz that does not include aliasing components between frames, and
The frame difference signal is output. The absolute value circuit 28 has a differential LP
Calculate the absolute value of the F26 output signal.

The comparison circuit 30 compares the output signal A of the absolute value circuit 28 with the threshold circuit.
Compare with threshold B from 32. For example, when absolute value signal A> threshold value B, the comparison circuit 30 determines that the moving image area is
Outputs "1". In other cases, the comparison circuit 30 outputs "0". Sampling frequency converter 34
Sets the sampling frequency of the output signal of the comparison circuit 30 to 32.4.
Converts from MHz to 48.6MHz and outputs the area discrimination signal.

The switch 36 is a motion detection system (sampling frequency converter
The output signal of the still image system (inter-field interpolation circuit 20) and the output signal of the moving image system (sampling frequency converter 24) are selected according to the area discrimination signal from (34) and the still image system signal and the moving image system are selected. Mix with the signal. The mixed signal is converted back into an analog signal by the digital / analog (D / A) converter 38.

As described above, the moving image area when the television signal is decoded by the conventional technique shown in FIG. 4 will be described with reference to FIG.

In FIG. 5, (a) shows an image of a comparison frame that is one frame before the target frame, that is, the N-1th frame, and
(B) shows the image of the target frame, that is, the Nth frame.

Areas A and C in the comparison piece N-1 and the attention piece N are still image areas, and areas B and D are moving image areas. That is, in this example, the moving image area is changed from the diamond-shaped area B to the circular area D during the passage of time from the comparison piece N-1 to the attention piece N. Further, the still image areas A and C obtained by removing the rhombus or the circle from the rectangle are, for example, the background and are the portions that do not change.

In the prior art of FIG. 4, as described above, the correlation calculation of the image signals corresponding to the images of the comparison frame N-1 and the noticed frame N is performed (blocks 26 to 34). The result of this correlation calculation is shown in FIG.

In FIG. 5 (c), since the comparison frame N-1 has been vector-corrected by the motion correction circuit 14, the corrected areas corresponding to the areas A and B before correction are A'and B '. And

An area A'.C where the still image area A'of the comparison piece N-1 (which has been vector-corrected) and the still image area C of the attention piece N overlap.
Indicates that a significant correlation is detected (comparative frame N-1 and target frame N
The difference between the video signals of and is set to a threshold value or less), and it is determined to be a still image area.

On the other hand, in the area B '+ D occupied by at least one of the moving image area B of the comparison piece N-1 and the moving image area D of the attention piece N, no significant correlation is detected (the comparison piece N-1 and the attention piece N are If the difference between the video signals is larger than the threshold value), it is determined to be the moving image area.

In FIGS. 4 and 5, only the motion detection based on the 1-frame correlation has been described, but the same applies to the 2-frame correlation.

[Problems to be Solved by the Invention] The conventional techniques shown in FIGS. 4 and 5 are only for detecting one frame difference between two frames. Therefore, the correlation calculation result is the fifth
The shaded area B ′ in FIG. 6C is also determined to be the moving image area of the target frame. However, as shown in FIG. 5 (b),
The shaded area B ′ · is not actually a moving picture area. That is, the conventional device has a drawback that it is determined to be the moving image area of the target frame, although it is actually a still image area.

In addition, IDTV (Improved Def
In inversion TV), the same drawback occurs in interpolation when converting from interlaced scanning to progressive scanning.

Therefore, an object of the present invention is to solve the above-mentioned drawbacks of the prior art and to provide a television signal reproducing method for interpolating a television signal by detecting a moving image area more faithfully.

[Means for Solving the Problems] The present invention is a television signal reproducing method for decoding a television signal, which uses a correlation calculation between a target frame and a plurality of comparison frames temporally before and after the target frame. A plurality of moving image area candidates are detected, a moving image area of a target frame is obtained from these plurality of moving image area candidates, and signals of a plurality of comparison frames are interpolated in the still image area according to the obtained moving image area.

[Operation] In the present invention, unlike the prior art, the comparison piece is not only the piece immediately before the attention piece, but the attention piece and the m frame (m = 1,
2, 3, ...) Multiple pieces before and after. Therefore, a plurality of moving image area candidates are obtained. By obtaining the moving image area from the plurality of moving image area candidates, the moving image area in the target frame can be detected more faithfully, so that the area determination signal becomes more practical. Therefore, the interpolation of the television signal can be performed more faithfully.

[Embodiment] A preferred embodiment of the television signal reproducing method of the present invention will be described with reference to FIGS. Note that blocks similar to those in FIG. 4 are designated by the same reference numerals, and different points will be described in detail.

In FIG. 1, an A / D converter 10 samples an analog television input signal and converts it into a digital signal.

In the still picture system, the combination of the frame memory 12 and the motion compensation circuit 14 is the same as in the case of FIG. But,
In the present invention, the output signal of the A / D converter 10 is stored in the frame memory 40, the signal is supplied to the frame memory 12 with a delay of one frame (one frame). Therefore, if the output signal of the frame memory 40 is the target frame N, the output signal of the frame memory 12 becomes the comparison frame N-1 one frame before the target frame N in time, and the A / D converter 10 Of the comparison frame N + 1, which is one frame after the frame N of interest.
Becomes

On the other hand, the motion correction circuit 14 compares the target frame N with the comparison frame N-.
The signal of No. 1 is motion vector corrected.

Further, the motion correction circuit 42 compares the target piece N with the comparison piece N +.
The signal of No. 1 is motion vector corrected.

As described above, the output signal of the frame memory 40 is the target frame N.
In the video system, the field interpolation circuit 22
And sampling frequency converter 24 is a frame memory
The 40 output signals are processed as in FIG.

The blocks 26 to 32 of the motion detection system are the same as in the case of FIG. 4 and operate similarly. However, the output signal M of the comparison circuit 30 is stored in the frame memory 44, delayed by one frame, and supplied to the motion correction circuit 46.

The motion correction circuit 46 performs motion vector correction so that the reference position of the moving image area candidate M ′ is the comparison frame N. Therefore, the output signal M ′ of the motion compensation circuit 46 is the target frame N and the comparison frame N−.
The moving image area candidate M ′ is obtained by the correlation calculation with 1.

The output signal M of the comparison circuit 30 is the same as the target frame N and the comparison frame N.
It becomes the moving image area candidate M obtained by the correlation calculation with +1.

The AND gate 48 includes a moving image area candidate M and a motion correction circuit 46.
And the output signal M'of The sampling frequency converter 34 converts the sampling frequency of the output signal of the AND gate 48 from 32.4 MHz to 48.6 MHz, and the target frame N
The area discrimination signal V is output.

The process of obtaining the area discrimination signal V of the target frame N will be further described with reference to FIG.

In FIG. 2, (a) shows an image of a comparison frame one frame (one frame) before the target frame, that is, the N-1th frame, and (b) shows the target frame, that is, the target frame. , The image of the Nth frame, and (c) shows the image of the comparison frame one frame (one frame) after the target frame, that is, the image of the (N + 1) th frame.

Areas A, C and E in the comparison frames N-1 and N + 1 and the target frame N are still image areas, and areas B,
D and F are moving image areas.

That is, in this embodiment, the comparison piece N-1 to the attention piece N
Then, during the passage of time to the comparison frame N + 1, the moving image area changes from the diamond-shaped area B to the circular area D and to the triangular area F.

Further, the still image areas A, C, and E obtained by removing the rhombus, the circle, or the triangle from the rectangle are, for example, the background and are the portions that do not change.

In the present invention, first, the correlation calculation of the image signals corresponding to the images of the comparison frame N-1 and the attention frame N is performed to obtain the moving image area candidate M '. The result of this correlation calculation is shown in FIG.
Shown in. It should be noted that A'and B'represent that the areas A and B are motion-corrected according to the motion vector.

In FIG. 2 (d), an area A'.C where the still image area A'of the comparison piece N-1 and the still image area C of the attention piece N overlap.
In (the logical product of the areas A ′ and C), a significant correlation is detected (the difference between the signals of the comparison frame N−1 and the frame of interest N is less than or equal to a threshold), and it is determined to be a still image area.

On the other hand, no significant correlation is detected in the area B '+ D (logical sum of areas B'and D) occupied by at least one of the moving picture area B'of the comparison piece N-1 and the moving picture area D of the attention piece N ( The difference between the signals of the comparison piece N-1 and the attention piece N is larger than the threshold value), it is determined as the moving image area candidate M '. Therefore, the content of FIG. 2D becomes the output signal of the motion correction circuit 46.

Next, the correlation calculation of the image signals corresponding to the images of the comparison frame N + 1 and the attention frame N is performed. The result of this correlation calculation is shown in FIG. It should be noted that E'and F'represent that the regions E and F were motion-corrected according to the motion vector.

In FIG. 2 (e), a region C · E ′ where the still image region E ′ of the comparison frame N + 1 and the still image region C of the attention frame N overlap each other.
In (the logical product of areas C and E ′), a significant correlation is detected, and it is determined to be a still image area. On the other hand, a region D + F ′ (logical sum of regions D and F ′) occupied by at least one of the moving image region F ′ of the comparison frame N + 1 and the moving image region D of the attention frame N is
No significant correlation is detected and it is determined to be the moving image area candidate M.
The content of FIG. 2 (e) corresponds to the output signal of the comparison circuit 30.

As shown in FIGS. 2 (d) and (e), in the correlation calculation between the target piece and one comparison piece, as in the prior art, the shaded area B '(area B' , And F ′ (logical product of areas other than area D and area F ′) are detected as moving image area candidates in the target frame N regardless of the still image area.

However, in the present invention, due to the AND gate 48, (d)
Since the logical product of the moving image area candidates of (e) and (e) is obtained, the result is as shown in FIG. 2 (f). That is, only the moving image area D of the attention piece N is detected as the moving image area. Therefore, it can be understood that the moving image area can be detected more faithfully.

Referring again to FIG.

In the still image system, the switch 50, depending on the moving image area candidate M ′ delayed by the frame memory 44, is a signal N−1, which is one frame before the frame (current signal) N of interest, or one frame. Either one of the subsequent signals N + 1 is selected. That is, when the moving image area candidate M ′ is “1”, the signal N + 1 is selected, and when the moving image area candidate M ′ is “0”,
Select signal N-1.

The switch 16 alternately selects the signal selected by the switch 50 and the current signal N from the frame memory 40 at a 32.4 MHz alternation to perform interframe interpolation. The output signal of the switch 16 is processed by the sampling frequency converter 18 and the inter-field interpolation circuit 20 in the same manner as in FIG.

The switch 36 is a motion detection system (sampling frequency converter
The output signal of the still image system (inter-field interpolation circuit 20) and the output signal of the moving image system (sampling frequency converter 24) are selected according to the area discrimination signal V from Mix with the system signal. The mixed signal is converted back into an analog signal by the digital / analog (D / A) converter 38.

In the above embodiment, the switch 50 is switched according to the moving picture area candidate M ′ delayed by the frame memory 44, but it may be switched according to the moving picture area candidate M which is the output signal of the comparison circuit 30. That is, when the moving image area candidate M is “1”, the signal N−1 is selected and the moving image area candidate M is selected.
Is 0, the signal N + 1 is selected.

Further, the region discrimination signal of the target frame is obtained by using the target frame and the comparison frames of one frame (one frame) before and after the target frame, but m frames (m = 1, 2, 3, ... Even if the area discrimination signal of the target frame is obtained from the comparison frames before and after, the signal can be faithfully interpolated.

Next, a case where a still image and a moving image are linearly mixed according to the amount of movement will be described with reference to FIG. Note that blocks similar to those in FIG. 1 are designated by the same reference numerals, and different points will be described in detail.

The output signal of the absolute value circuit 28 and the threshold value from the threshold value circuit 32 are supplied to the difference truncation circuit 52. That is, when the output signal of the absolute value circuit 28 is larger than the threshold value, a value corresponding to the result of the subtraction is output, and when the output signal of the absolute value circuit 28 is smaller than the threshold value, "0" is output.

The moving picture area candidate M and the moving picture area candidate M ′ through the frame memory 44 are supplied to the minimum value circuit 54, and the output signal thereof is sent through the sampling frequency converter 34 to the area discrimination signal V.
Is output.

Further, the moving image area candidates M and M ′ are supplied to the comparison circuit 56,
The switch 50 is switched according to the output signal. That is, when the moving image area candidate M is larger than M ′, the signal N
-1 is selected, and if the moving image area candidate M is smaller than M ', the signal N + 1 is selected.

Further, according to the area discrimination signal V, the output signal of the still image system (inter-field interpolation circuit 20) and the output signal of the moving image system (sampling frequency converter 24) are linearly mixed in the linear mixing circuit 58. In this way, a signal in which a still image and a moving image are linearly mixed is obtained according to the area discrimination signal V of the target frame N.

[Effects of the Invention] As described above, according to the present invention, a plurality of moving image area candidates are detected by receiving the multiple sub-sample transmission signals and calculating the correlation between the target frame and the comparison frames before and after the target frame. The moving picture area of the target frame is obtained from the moving picture area candidate of No. 3, and the area discrimination signal is generated.

Therefore, the possibility that the still image region is erroneously detected as the moving image region is reduced, and the moving image region can be detected more accurately, and the region determination signal is also more accurate. Then, by using this area discrimination signal to prepare a signal for inter-frame interpolation in the still image area of the target frame from a plurality of comparison frames, the still image area can be faithfully reproduced. Therefore, the image quality can be improved as a whole.

Further, the present invention can be effectively used for the IDTV in the current NTSC system.

[Brief description of drawings]

FIG. 1 is a block diagram of a first embodiment of an apparatus suitable for implementing the present invention, FIG. 2 is a diagram showing an image for explaining the principle of the present invention, and FIG. 3 is implementing the present invention. FIG. 4 is a block diagram of a second embodiment of a device suitable for the above, FIG. 4 is a block diagram of a conventional device, and FIG. 5 is a diagram showing an image for explaining the operation of FIG. N …… Attention frame N + 1, N−1 …… Comparison frame 10 …… A / D converter 12,40,44 …… Frame memory 16,36,50 …… Switch 14,42,46 …… Motion compensation circuit 18, 24, 34 …… Sampling frequency converter 20 …… Inter-field interpolation circuit 22 …… Field interpolation circuit 26 …… Difference LPF 28 …… Absolute value circuit 30 …… Comparison circuit 32 …… Threshold circuit 38… … D / A converter 48 …… and gate

Claims (1)

[Claims] 1. A television signal reproducing method for decoding a television signal, wherein a plurality of moving image area candidates are detected by a correlation calculation between a target frame and a plurality of comparison frames temporally before and after the target frame. A television signal reproducing method characterized in that a moving image area of the target frame is obtained from the plurality of moving image area candidates, and signals of the plurality of comparison frames are interpolated into a still image area in accordance with the obtained moving image area.