JPH0260288A - Reproducing system for transmitted television signal - Google Patents

Abstract

PURPOSE:To reproduce a high definition picture without spoiling a band by discriminating the still picture and the dynamic picture areas of a remarked frame from plural dynamic picture area candidates, and performing inter-field interpolation for the still picture area and intra-field interpolation after inter- frame offset sub-sampling for the dynamic picture area. CONSTITUTION:The dynamic picture area candidate is detected from among the remarked frame and relational frames within two frame before and after by frame correlation, and the logical product of those is taken by an AND circuit 73, and the dynamic picture area of the remarked frame is obtained, and by using switches 55, 56, a necessary omitted signal is interpolated by the inter-field interpolation in the still picture area, and in the dynamic picture area, the inter-frame offset sub-sampling is performed by a circuit 45 and the necessary omitted signal is interpolated by the intra-field interpolation, and further, the signal of the motion-vector-corrected field at a transmitting side is motion-vector-corrected in a contrary direction by a RAM 63, and is returned to a former position.

Description

DETAILED DESCRIPTION OF THE INVENTION "Field of Industrial Application" The present invention relates to a method for reproducing a transmitted television signal whose signal band is compressed and transmitted.

"Prior Art" As a method for compressing the transmission signal band of a television signal, a multiple subsample transmission method using interfield and interframe offset subsampling is known. MUSE-tl (Multiple 5ub-Ny
c + uist Sampling Encoding)
has been proposed, for example, as described in Japanese Unexamined Patent Publication No. 61-264889.

FIG. 14 shows the sampling pattern of the luminance signal of the MUSE-■ transmission system.

The Q mark, low mark, - mark, and ■ mark in the figure are the 4n+11th and 4th subsampling patterns for 2 frames and 1 round, respectively.
The positions that are subsampled in the 0+2nd, 41st+3rd, and 4n+4th fields (n=0.1.2°...) are shown, and the X marks are the positions that are not sampled in any field.

In the figure, d2 indicates sampling l1lXII, and the sampling frequency 1/d2 is, for example, 48°6 MHz.

Further, FIG. 16 shows a processing circuit (encoder) on the transmitting side. In the figure, a human input signal supplied to an input terminal 1 is sampled by an A/D converter 2 at a sampling frequency of 48.6 MHz and converted into a digital signal.

FIG. 16A shows the signal band at this time.

In FIG. 16, the horizontal axis represents the horizontal component H1, and the vertical axis represents the vertical component V.

Further, the output signal of the A/D converter 2 is supplied to an interfield filter 3. In this inter-field blurring filter 3, as a process for the still image area, high frequency components in the diagonal direction of the screen are removed as shown in FIG. 16B.

Further, the output signal of the inter-field bristle filter 3 is supplied to a sub-sampling circuit 4. In this sub-sampling circuit 4, inter-field offset sub-sampling is performed at a sampling frequency of 24.3 MHz. In this case, the signal in the band of 12.15 MHz or higher is folded back, and the signal band becomes as shown in FIG. 16C.

Further, the output signal of the sub-sampling circuit 4 is supplied to a sampling frequency conversion circuit 5, and its sampling frequency is converted from 24.3 MHz to 32.4 MHz. In this case, the signal band remains in the state shown in FIG. 16C.

The output signal of the A/D converter 2 is also supplied to an intra-field prefilter. In this intra-field pre-filter, as shown in Figure 16, the band is limited to 12.15 MHz as processing for the moving image area.

Further, the output signal of the intra-field prefilter is supplied to a sampling frequency conversion circuit 7, and its sampling frequency is converted from 48.6 MHz to 32.4 MHz. In this case, the signal band remains as shown in Figure 16.

Further, the output signals of the sampling frequency conversion circuits 5 and 7 are supplied to a linear mixing circuit 8. Further, the output signal of the A/D converter 2 is supplied to a motion detection circuit 9. In this motion detection circuit 9, non-linear processing is performed on the absolute value of the inter-frame difference to detect the amount of motion. The detection signal of this motion detection circuit 9 is then sent to the linear mixing circuit 8 for control 1.
It is supplied as the Ill signal, and in this linear mixing circuit 8,
The sampling frequency changing circuit 5 at a rate according to the amount of movement.
and 7 output signals are mixed.

Further, the output signal of the linear mixing circuit 8 is supplied to a sub-sampling circuit 10. This sub-sampling time vNl
At o, interframe offset subsampling is performed at a sampling frequency of 16.2 MHz. In this case, the signal band as shown in FIG. 16C for the still image system is as follows:
The band above 8.1 MHz is folded back, as shown in Figure E.

On the other hand, the signal band for video systems as shown in the same diagram is 8.1
The band above MHz is folded back, as shown in F in the figure.

Further, the output signal of the sub-sampling circuit 10 is converted into an analog signal by a D/A converter 11, and then led out to an output terminal 13 via a transmission line filter 12 and sent out to a transmission line. This transmission line filter 12 is assumed to have a cosine roll-off characteristic at 8.1 MHz.

"Problem to be Solved by the Invention" According to the method of transmitting data through a processing circuit as shown in FIG.
．． There was a drawback that only a 3 MHz luminance signal could be band-compressed.

In addition, as shown in Figure 116 E and F, 4.05M
Since there is an aliasing component between frames in the signal of Hz or more, when detecting motion on the receiving side, only signal components of 4.05 MHz or less can be used, which has the disadvantage that the movement of small objects cannot be detected.

Therefore, instead of the above-mentioned transmission method, it is conceivable to compress the transmission band using inter-field and inter-frame offset subsampling and time axis expansion techniques to transmit a higher-band signal.

An object of the present invention is to provide a reproduction method that better reproduces an original television signal from a transmitted television signal whose transmission band has been compressed using interfield and interframe offset subsampling and time axis expansion technology. It is something to do.

"Means for Solving the Problem" The television signal reproduction method according to the first invention is such that in the still image area, one frame of the television signal out of two consecutive frames of television No. 1g is subtracted by field offset. In the video domain, two consecutive frames of the television signal are subjected to inter-field and inter-frame offset subsampling to form signals of the first to fourth fields, and then frames are sampled to form signals of one frame. Interpolation is performed to form an l frame signal, and when decoding a transmitted television signal in which this one frame signal is time-axis expanded and transmitted, multiple frames are interpolated between the frame of interest and multiple comparison frames due to interframe correlation. The video area candidates are detected, and the still image and video area of the frame of interest are determined from these multiple video area candidates. In the still image area, inter-field interpolation is performed, and in the video area, inter-frame offset subsampling is performed. and then the field is interpolated and the first to fourth
A field signal is formed, time-axis compressed, and a continuous two-frame television signal is reproduced.

The transmission television signal reproduction method according to the second invention includes:
In the still image area, the average of two consecutive frames of television signals is taken between frames, and the interfield offset subsampling is performed to form one frame signal.In the video area, the two consecutive frames of the television signal are averaged between fields Then, frame leap offset subsampling is performed to form signals of the first to fourth fields, which are interpolated between frames to form one frame signal, and this l frame signal is time-axis expanded and transmitted. When decoding a transmitted television signal, a plurality of video region candidates are detected between the frame of interest and a plurality of comparison frames by frame interrogation, and the still image and video region of the frame of interest are determined from these multiple video region candidates. In the still image area, inter-field interpolation is performed, and in the moving image area, inter-frame offset subsampling is performed and intra-field interpolation is performed to form the signals of the first to fourth fields, which are time-axis compressed. consecutive 2
A frame television signal is reproduced.

Further, in the reproduction method of the transmitted television signal according to the third invention, after the television signal of two consecutive frames is subjected to inter-field and inter-frame offset subsampling to form the signals of the first to fourth fields. ,
One frame signal is formed by interpolation between frames, and when this l frame signal is time-axis expanded and transmitted, when decoding the transmitted television signal, the interframe correlation between the frame of interest and multiple comparison frames is used. Multiple video area candidates are detected, and the still image and video area of the frame of interest are determined from these multiple video area candidates. In the still image area, inter-field interpolation is performed, and in the video M area, inter-frame offset is performed. After subsampling, field interpolation is performed to form signals of the first to fourth fields, which are time-base compressed and two consecutive frames of television signals are reproduced.

"Function" In the configurations of the first, second, and third inventions, a plurality of video regions are created by interframe correlation between the frame of interest and comparison frames before and after m frames (m=2, 4, 6, etc.). Candidates are detected, the still image and video area of the frame of interest are determined from these multiple video area candidates, processing is performed according to each area, and the original television signal is played back after being further time-axis compressed. Therefore, high-definition images can be reproduced without impairing the band of transmitted signals.

"Embodiment" First, an example of a transmission method in which the transmission band is compressed and transmitted using inter-field and inter-frame offset subsampling and time axis expansion technology will be described.

1 and 2 show the sampling pattern of the luminance signal of this transmission system, FIG. 1 is the sampling pattern for the still image area, and FIG. 2 is the sampling pattern for the moving image area. Both are sub-sampling patterns that go around once in two frames, and the ○, 0, ・, and ■ marks in the figure are the 40th+1st, 40th+2nd, and 4nth, respectively.
+3rd field, 40th +4th field (n=0.1
．． 2+9. ) indicates a position that is subsampled, and an X mark indicates a position that is not sampled in any field.

In the figure, dl indicates the sampling period,
The sampling frequency 17d1 is, for example, 64.8MH2.
It is.

In this example, the resolution in the diagonal direction is reduced by inter-field offset subsampling, and information is reduced to 1/2.

In the still image area, we focused on the fact that temporal interpolation can be performed by performing motion vector correction between signals of two consecutive frames. As shown in Figure 1,
Only the signals of the 4n+3th and 4n+4th fields are prepared.

On the other hand, in the video domain, interframe offset subsampling is performed, focusing on the fact that human vision significantly reduces resolution for moving objects.

As shown in FIG. 2, in the 4n+1th field, subsampling is performed at the O mark position, and
In the third field, the 4nth+1st signal is subsampled at the position marked with motion vector correction, and then interframe interpolated at the same position in the 4n+3rd field. In the second field, subsampling is performed at the position of the 0 mark, and in the 4n+4th field, subsampling is performed at the position of the signature mark. After the 40th+2nd signal is subjected to motion vector correction, the 40th+4th signal is The signals of the first field and the second field are interpolated between frames at the same position.

Then, the signal of one frame consisting of the first field and the second field prepared in this way is time-extended by 2 times and transmitted over 2 frames, reducing the information per 2 frames to 172. Ru.

FIG. 3 shows the processing times U<encoder on the transmitting side in this transmission system.

In the figure, a human power signal (luminance signal) supplied to a human power terminal 21 is sampled at a sampling frequency of 64.8 MHz by an A/D converter 22 and converted into a digital signal.

FIG. 4A shows the signal band at this time. In FIG. 4, the horizontal axis represents the horizontal component H1, and the vertical axis represents the vertical component V.

Further, the output signal of the A/D converter 22 is supplied to an interfield filter 24 via a delay circuit 23. The delay circuit 23 is provided for time adjustment. In the inter-field blurring filter 24, as shown in FIG. 4B, high-frequency components in the diagonal direction of the screen are removed as processing for the still image area.

In addition, the output signal of the inter-field bris filter 2.4 is
The signal is supplied to the sub-sampling circuit 25.

In this sub-sampling time #i25, 32.4M
Inter-field offset subsampling is performed at a sampling frequency of Hz to form a still image signal. In this case, signals of 16.2 MHz or higher are folded back, and the signal band becomes as shown in FIG. 4C.

Further, the output signal of the A/D converter 22 has a delay amount gII2
3 to the intra-field filter 26. In this intra-field blurring filter 26, as shown in the fourth diagram, 16.2MHz is used for processing the moving image area.
Bandwidth is limited to

Further, the output signal of the intra-field bristle filter 26 is supplied to a sub-sampling circuit 27.

In this sub-sampling circuit 27, 32.4M
Field interleaved offset subsampling is performed at a sampling frequency of Hz. In this case, since the band is limited to 16.2 MHz by the intra-field filter 26, there is no aliasing and the signal band remains as shown in the fourth diagram.

Further, the output signal of the sub-sampling circuit 27 is supplied to the a side terminal of the changeover switch 29, and is also supplied to the b side of the changeover switch 29 via a frame memory 30 and a RAM (random access memory) 28. is supplied to the terminal. By this changeover switch 29, the current signal on the a side and the one-frame delayed signal on the bs are subsampled with frame offset at 16.2 MHz, and the signals of two consecutive frames are sampled according to the sampling pattern shown in FIG. Interframe interpolation is performed within one frame to form a moving image signal. In this case, the band of the signal subsampled in each field is 8.1
The signal is folded back at MHz, as shown by the broken line in FIG. 4F, but the apparent signal band after interpolation between frames remains as shown by the solid line in the same figure.

Further, the output signal of the A/D converter 22 is supplied to a motion vector detection circuit 31. This motion vector detection circuit 31
In this step, a motion vector is detected, and it is determined whether the area is a still image area or a moving image area on a pixel by pixel basis. Then, the amount of delay of the RAM 2B described above is controlled by the motion vector, and motion vector correction is performed.

Further, the output signal of the sub-sampling circuit 25 is supplied to the all terminal of the changeover switch 32, and the output signal of the changeover switch 29 is supplied to the bm terminal of the changeover switch 32. Switching of the changeover switch 32 is controlled by an area discrimination signal output from the motion vector detection circuit 31, and is switched to the a side and the b side in the still image area and the moving image area, respectively.

Further, the output signal of the changeover switch 32 is supplied to a time axis expansion circuit 33, and the signals of the first field and the second field are expanded in time axis into two frames. As a result, the still image signal band as shown in Figure 4C is
The band is compressed as shown in Figure F, and on the other hand, the moving image signal band as shown in Figure F is compressed as shown in Figure G, resulting in a signal band of 8.1 MH2. However, when focusing only on the signals of each field in the moving image system, as shown by the broken line in G in the figure, the signals in the band of 4.05 MHz or more are folded back between frames.

Further, the output signal of the time axis expansion circuit 33 and the motion vector from the motion vector detection circuit 31 are supplied to the multiplexing circuit 34 and time-division multiplexed. After this multiplexed signal is converted into an analog signal by the D/A converter 35,
is led out to the output terminal 37 via the transmission line filter 36,
sent out to the transmission path. Transmission line filter 36 is 8.1M
It has a cosine roll-off characteristic at Hz.

FIG. 5 shows a specific configuration of the motion vector detection circuit 31.

In the figure, the output signal of the A/D converter 22 is supplied to the current field buffer 301A and written therein, and is also supplied to the previous field buffer 303A via the 2-frame memory 302 and written therein. and,
During the vertical blanking period, buffer 3 for the current field
Of the one-frame image written to 01A, the signal of each pixel in the block of interest consisting of multiple pixels is read out, and the one-frame image written to the previous field buffer 303A matches the block of interest described above. A plurality of comparison blocks are set up based on the position of the comparison block, and the signal of each pixel in this comparison block is read out. The signals of each pixel read out from each pixel are supplied to a subtracter 306 via changeover switches 304 and 305, and subjected to subtraction processing. In this case, the output signal of the address generation circuit 307,
The shift vector from the shift vector generation circuit 309 selected by the switch switch 308 is added by an adder 310, and the output signal of this adder 310 is sent to the switch 31.
1A to the previous field buffer 303A as a read address signal. Here, the selection of the comparison block is performed by the shift vector from the shift vector generation circuit 309, while the selection of each pixel within the comparison block is performed by the output signal of the address generation circuit 307. In addition, in the vertical blanking visit of the next field, the output signal of the adder 310 is switched to the selector switch 3.
11B to the previous field buffer 303B as a read address signal.

Further, the output signal of the subtracter 306 is supplied to the integration 1313 via the absolute value circuit 312. And this cumulative tB
At 313, all pixels in the block of interest are integrated. The output signal of the integrator 313 and the output signal of the memory 314 initialized to an appropriate threshold are sent to the comparator 3.
15.

When the output signal of the integrator 313 is smaller than the output signal of the memory 314, the memory 314 is controlled by the output signal of the comparator 315, and at the same time the output signal of the integrator 313 is written to the memory 314, the comparator 314
A memory 316 is controlled by the output signal of 15, and the output signal of the shift vector generation circuit 309 is written into this memory 316.

The above process is similarly repeated for a plurality of comparison blocks, and the shift vector that provides the most two-frame correlation between the block of interest and the comparison block is finally stored in the memory 316 as a motion vector candidate. This operation is repeated for a plurality of blocks of interest within the frame of interest, and motion vector candidates for all blocks of interest within the frame of interest are stored in the memory 316.

All motion vector candidates stored in this memory 316 are:
The motion vector is supplied to a motion vector generation circuit 317, and a motion vector is determined by calculation processing such as an average value or majority vote.
The motion vector of the two-frame correlation is stored in the memory 31B.

Further, the signal of the next field is written to the current field buffer 301B and the previous field buffer 303B, and the motion vector of the two-frame correlation is stored in the memory 318 using the same procedure.

Furthermore, in the next field, 2 frame memory 302
The signal in the previous field buffer 303A is rewritten via the buffer 303A for the current field, and a signal with a one frame difference is prepared in the current field buffer 301A and the previous field buffer 303A, and the motion vector of one frame correlation is stored in the memory using the same procedure as described above. 318. Then, the changeover switch 308 is switched, and the motion vector generation circuit 317
A motion vector with one frame correlation from
The read address is controlled and the motion vector is corrected. After that, the current field buffer 301
Signals for all pixels are read out from the A and previous field buffers 303A. At the same time, the next field signal is supplied to the previous field buffer 303B via the 2-frame memory 302 and written therein.

The output signal of the subtracter 306 is then supplied to the comparator 319 via the absolute value circuit 312, and the reference level generation circuit 3
20. In this case, if the output signal of the absolute value circuit 312 is larger than the reference level, the comparator 319 determines that it is a moving image area, and, for example, a signal of high level "1 ++" is detected. It is determined that it is an image area, and a signal of low level "0" is output, and this area discrimination signal is supplied as a switching control signal to the changeover switch 32 via the output terminal 322 as shown in FIG. 3 above. .

For the signal of the next field, a motion vector of one frame correlation and a signal for each region 1'l are prepared using the same procedure.

In this way, among the four motion vectors stored in the memory 318, for example, one frame of an odd field,
The motion vector between two frames is read out at an appropriate timing, and after a delay of one frame, the motion vector between one frame and two frames of an even field is read out, and the motion vector between two frames is read out via the output terminal 321. A multiplex circuit 34 is provided as shown.

In such a transmission system, for example, the transmission band 8. Signals up to 32.4 MHz can be band-compressed within I MHz. Further, at least in the still image area, since a signal is transmitted without aliasing between frames, the receiving side can accurately detect the still image area and the moving image area.

Hereinafter, an embodiment of a method for reproducing a television signal transmitted in this manner will be described with reference to the drawings.

In the playback method of this example, the frame of interest and m frames (m
=2. 4-+6...) A plurality of video area candidates are detected by inter-frame correlation between the previous and subsequent comparison frames, and the still image and video area of the frame of interest are determined from these multiple video area candidates. In the area, the transmitted 4n+3. Similar to the ftth and 4n+43rd fields, necessary missing signals are interpolated by interfield interpolation in the 40th+1@th and 40+1@th field systems that are not transmitted, and in the video area, interframe offset sub Sampling is performed, necessary missing signals are interpolated by intra-field interpolation, and motion vector correction is performed on the transmitting side.
The signal of the n+28th field is subjected to motion vector correction in the opposite direction and returned to its original position. Furthermore, two consecutive frames of the television signal are reproduced by time axis compression.

FIG. 6 shows a processing circuit (decoder) on the receiving side according to the reproduction method of this example.

In the figure, the human input signal supplied to the input terminal 42 is
The A/D converter 43 samples the signal at a sampling frequency of 16.2 MHz and converts it into a digital signal. At this time, the signal band for still images becomes as shown in FIG. 7A, and the signal band for moving images becomes as shown by the solid line in FIG. 7B. In video systems, inter-frame offset subsampling is performed as shown in Figure 2, so if we focus only on the signals of each field, we can detect signals with frequencies of 4.05 MHz or higher, as shown by the broken line in Figure 7B. However, the signal band after transmission processing and interframe interpolation is apparently as shown by the solid line in Figure B.

Further, the output signal of the A/D converter 43 is supplied to an interfield interpolation circuit 44 via a 2-frame memory 49. In this interfield interpolation circuit 44, interfield interpolation is performed as still image processing, and as shown in FIG. 7C, aliasing components are restored.

Further, the output signal of the A/D converter 43 is supplied to the sub-sampling circuit 45 via the 2-frame memory 49. In this sub-sampling circuit 45, 8, 1
After subsampling is performed at a sampling frequency of MHz, the signals are supplied to intra-field interpolation circuits 46 and 47.

That is, the 41st+1st and 4n+th subsampled signals in the first field of the signal transmitted to the receiving side
The third field signal is supplied to field interpolation circuits 46 and 47, respectively, and the 40th+2nd and 4nth signals subsampled in the second field.
The signals of the +4th field are supplied to intra-field interpolation circuits 46 and 47, respectively.

These intra-field interpolation circuits 46 and 47 restore the folded components indicated by the broken lines in FIG. 7B, as shown by the solid lines in FIG. 7B.

Further, the output signal of the A/D converter 43 is sent to a subtracter 50 via a 2-frame memory 49 that constitutes a 2-frame delay line.
It is also supplied to a subtracter 50 via a RAM 51 that constitutes a motion vector correction circuit. and,
The output signal of this subtracter 60 is supplied to a comparator 53 via an absolute value circuit 52 and compared with a reference level supplied from a reference level generation circuit 54. In this case, comparator 5
3, if the output signal of the absolute value circuit 52 is higher than the reference level, it is determined to be a moving image area candidate, and for example, a signal with a high level of "1" is determined as a still image area candidate. It is determined that it is a candidate, and a signal of low level "'0pa" is output.

Further, the output signal of the comparator 53 is supplied to an interfield interpolation circuit 48 to perform interfield interpolation. The output signal of this inter-field interpolation circuit on the fourth day is sent to an ant circuit 73 via a 2-frame memory 71 which constitutes a 2-frame delay line and a RAM 72 which constitutes a motion vector correction circuit.
At the same time, the output signal of this interfield interpolation circuit 48 is directly supplied to an AND circuit 73 to perform a logical product. The output signal of this AND circuit 73 is used as a still image/moving image area determination signal.

Further, the output signal of the A/D converter 43 is supplied to a motion vector separation circuit 57.
Then, the motion vector between 1 frames and the motion vector between 2 frames multiplexed on the transmitting side are obtained. Then, J! of the RAM 51 is determined by the motion vector between two frames. ! The amount of extension is controlled.

Also, the motion vector between these two frames is determined by the delay circuit 70.
RAM? The amount of delay in step 2 is controlled.

Further, the output signal of the interfield interpolation circuit 44 is supplied to the a side terminals of the changeover switches 55 and 56. Furthermore, the output signals of the field interpolation circuits 46 and 47 are as follows:
They are supplied to the b-side terminals of changeover switches 55 and 56, respectively.

These changeover switches 56 and 56 have an AND circuit 73
The output signal is supplied as a switching control signal, and these switches 55 and 56 are switched to the a side and the b side 1 in the still image area and the moving image area, respectively. As a result, the changeover switches 55 and 56 mix the still image system and the moving image system, and the changeover switches 55 and 56 mix the still image system and the video system.
The signals of the 4n+1st and 4n+3rd fields are regenerated in the first field of the signal transmitted to the receiving side, and the signals of the 4n+2nd and 4n+4th fields are regenerated in the second field. be obtained.

Further, the signals of the 41st+1st and 4n+2nd fields outputted from the changeover switch 55 are passed through the RAM 63, and then sent through the changeover switch 58 which is alternately switched to the a side and bIII in accordance with the frame change. The time axis is compressed to 172 by being supplied to field memories 59 and 60 forming an axis compression circuit. The output signals of the field memories 59 and 60 are switched to the opposite side of the changeover switch 58, and
In the same frame alternation, a side and b I! The signal is supplied to the a side terminal of the changeover switch 62 via the changeover switch 61 which is alternately switched to IJ.

Further, the signals of the 4n+3rd and 40th+4th fields outputted from the changeover switch 56 are sent to the time axis compression circuits through changeover switches 64 which are alternately switched to the an+1 and b sides in two-frame alternations. The interaxle axis is compressed by half.

The output signals of the frame memories 65 and 66 are switched to the side opposite to the switch 64 described above, and are switched via a switch 67 which is also switched alternately to the a side and the b side every two frames. switch 62
is supplied to the b-side terminal of the

Also, J! with RAM63! ! The amount of extension is controlled according to the motion vector between each frame separated by the motion vector separation circuit 57, and the position of the signal is returned to its original position.

In this case, the signal band for still images will be as shown in Figure 7, and the signal band for moving images will be as shown in Figure E.

In addition, the changeover switch 62 is switched between the a side and the b side by frame alternation.
This changeover switch 62 sequentially outputs two frames of signals consisting of a series of four fields. The output signal of this changeover switch 62 is D
After being converted into an analog signal by the /A converter 68, it is output to an output terminal 69.

In this way, according to this example, video region candidates are detected by frame correlation between the frame of interest and comparison frames two frames before and after the frame of interest, and the area resulting from the logical product of these is set as the video region of the frame of interest. Therefore, moving image areas can be detected more faithfully.

That is, Fig. 8A is a frame of N-2 frame, and Fig. 8B is a frame of N-2 frame.
C! Assuming that C in the same figure shows a frame of N frames, and C in the same figure shows a frame in N+2 frames, the same figure shows a still image between N-2 frames and N frames, and a moving image area vc complement, and E!
! Still image and video area candidates between frame N and frame N+2 are shown. Here a and b are each N
-2 frame still image area and video area, C and d
are a still image area and a video area of N frames, e
and f indicate a still image area and a moving image area of N+2 frames, respectively, and dashes indicate areas subjected to motion vector correction.

In the same figure, in the area b' + d, the subtraction circuit 50 detects a difference that is at least the reference level and becomes a moving image area candidate, and in the area a' and C, the subtraction circuit 50 detects a difference that is at least the reference level. It is not detected and becomes a still image area candidate. In addition, in the same figure E, in the area d+f', the subtraction circuit 50 detects a difference equal to or higher than the reference level and becomes a moving image area candidate, and in the region c'e', the subtraction circuit 50 detects a difference equal to or higher than the reference level. It becomes a still image area candidate. Therefore, although the regions b', d and df' in the eighth diagram and E are still image regions in the frame of interest (N frames), they are detected as moving image region candidates.

However, when the AND circuit 73 performs a logical product of these moving image area candidates, only the area d is detected as a moving image area, as shown in FIG. 8F. Therefore, according to this example, a moving image area can be detected more faithfully.

According to the reproduction method of this example, a moving image area is faithfully detected as described above, and a still image and a moving image area are discriminated. Then, processing is performed according to each area, and the transmitted field signal is subjected to motion vector correction on the transmitting side, and the motion vector is corrected in the opposite direction, and further time axis compression is performed to reproduce the original television signal. Therefore, high-definition images can be reproduced without impairing the transmitted signal band.

Note that there are other examples of transmission methods that compress the transmission band and transmit using inter-field and inter-frame offset subsampling and time axis expansion techniques.

Figure 9 shows the first processing circuit on the transmitting side in another transmission method.
This is an example. The output signal of the sub-sampling circuit 25 is supplied to the average value circuit 41 via the frame memory 40 and the RAM 38, and also directly to the average value circuit 41, and the output signal of this average value circuit 41 is switched as a still image signal. The signal is supplied to the a-side terminal of the switch 32.

The example shown in FIG. 9 is constructed as described above, and the rest is constructed similarly to the example shown in FIG. 3.

The reproduction of television signals transmitted in this way also
This can be carried out using the reproduction method of the above-described embodiment, and the processing circuit on the receiving side is configured similarly to that shown in FIG.

FIG. 10 shows a sampling pattern of a luminance signal in a second example of another transmission method. This FIG. 10 is similar to FIG. 2, so the explanation will be omitted.

This transmission method uses a combination of inter-field and inter-frame offset subsampling and time-base expansion techniques.

The resolution in the diagonal direction is reduced by inter-field offset subsampling, and information is reduced by 1/2.

Also, interframe offset subsampling is performed and interframe interpolation is performed. That is, as shown in FIG. 1O, in the 40th+1st field, subsampling is performed at the position marked O, and in the 40th+3rd field, subsampling is performed at the position marked . After the motion vector is corrected, interframe interpolation is performed to the same value of the 4n+3th field, and subsampling is performed at the 0 mark position in the 4n+28th field, and in the 40th+4th field, After being sub-sampled at the marked position and motion vector corrected for the 4n+28th signal, interframe interpolation is performed at the same position at the 40th+4th, thereby preparing the signals of the first field and the second field. Then, the signal of one frame consisting of the first field and the second field prepared in this way is time-expanded twice, and is transmitted over two frames, reducing the information per two frames to 172. Ru.

Figure 11 shows the processing circuit (
encoder). In FIG. 11, parts corresponding to those in FIG. 3 are designated by the same reference numerals, and detailed explanation thereof will be omitted.

In the figure, the output signal of the sub-sampling circuit 25 is supplied to the a-side terminal of the changeover switch 39, and
The signal is supplied to the b-side terminal of the changeover switch 39 via the frame memory 40 and the RAM 38. By this changeover switch 39, the current signal on the a side and the one-frame delayed signal on the b side are subsampled with an inter-frame offset at 16.2 MHz, and the signals of two consecutive frames are converted into the sampling pattern shown in FIG. According to 1
Interframe interpolation is performed within a frame to form a still image signal. In this case, the band of the signal subsampled in each field is folded back at 8.1 MH2, and the 12th
As shown by the broken line in Figure E, the substantial signal band after interpolation between frames remains as shown by the solid line in the figure. Furthermore, the amount of delay in the RAM 38 is controlled by the motion vector detected by the motion vector detection circuit 31, and motion vector correction is performed.

In addition, the output 1g of the changeover switch 39 is the changeover switch 3
It is supplied to the a side terminal of 2.

Switching of the changeover switch 32 is controlled by an area discrimination signal output from the motion vector detection circuit 31, and is switched to the a side and the b side in the still image area and the moving image area, respectively. The output signal of this changeover switch 32 is supplied to a time axis expansion circuit 33, and the signals of the first field and the second field are time axis expanded into two frames. As a result, the still image signal band as shown in FIG. 12E is compressed as shown in FIG. 12G, while the moving image signal band as shown in FIG. The bandwidth is compressed as shown in , and both are 8.1 M
The signal band is Hz. In this case, as shown by the solid line in FIG. 12G, in the still image system, the transmission band is substantially 8
, 1 MHz, there is no aliasing between frames.

Here, FIGS. 12A-D and F are the same as FIGS. 4A-D and F.

The example shown in FIG. 11 is constructed as described above, and the rest is constructed similarly to the example shown in FIG. 3.

In this way, the transmission method of this example also has a maximum of 32.4 MHz within the transmission band 8, 1 MHz of satellite broadcasting.
It is possible to band-compress a Hz signal. In addition, since a signal is transmitted with virtually no aliasing between frames, still image areas and moving image areas can be detected with high accuracy by receiving 1N ([1).

The reproduction of the television signal transmitted in this manner can also be performed by the reproduction method of the above-described embodiment, and the processing circuit on the receiving side is constructed in the same manner as that shown in FIG.

In the above-mentioned embodiment, video region candidates are detected by inter-frame correlation between the frame of interest and comparison frames two frames before and after the frame of interest, and the area resulting from the logical product of these is set as the video region of the frame of interest. However, an additional m frames (m
=4.6.8...) It is also possible to perform a logical product with the video M area candidate between the previous and subsequent frames.

In addition to the logical product, as shown in Figure 13, three of the four video area candidates detected by the inter-frame correlation between the frame of interest and comparison frames that are ±2 frames or ±4 frames away from the frame of interest. The area common to the above may be used as the video area of the frame of interest. As a result, in detecting a moving image area based on correlation between two frames, the possibility of erroneously determining a moving image area as a still image area can be extremely reduced.

Further, although the time axis compression process has been brought to a later stage in order to lower the clock frequency of signal processing, the time axis compression process may be performed immediately after A/D conversion to perform signal processing.

Further, in the above example, the motion vectors are multiplexed and transmitted on the transmitting side, but the motion vectors may be detected on the receiving side without being multiplexed on the transmitting side. In this case, only the motion vector between two frames can be detected, but the motion vector between l frames can be estimated from the motion vector between two frames.

In addition, there is a switch 55°56 to switch between still image system and video system.
However, linear mixing may be performed depending on the amount of movement.

Furthermore, although only the luminance signal is explained,
Of course, color signals can also be similarly processed and reproduced.

"Effects of the Invention" As is clear from the embodiments described above, according to the first, second and third inventions, the frame of interest and m frames (m=2
．． 4. 6...) A plurality of video region vc complements are detected by inter-frame correlation between the previous and subsequent comparison frames, and the still image and video region of the frame of interest are well distinguished from these multiple video region candidates, and each Since the original television signal is reproduced after processing is performed according to the area and the original television signal is roughly compressed on the time axis, it is possible to reproduce high-definition images without impairing the band of the transmitted signal.

[Brief explanation of the drawing]

Figures 1 and 2 are diagrams showing sampling patterns of an example of a transmission system, Figure 3 is a diagram showing a processing circuit on the transmitting side in that example, and Figure 4 is a diagram showing signals of each part of the processing circuit on the transmitting side. 5 is a diagram showing a motion vector detection circuit; FIG. 6 is a diagram showing a receiving side processing circuit in an embodiment of the reproduction method according to the first, second, and third inventions; FIG. Figure 7 is a diagram showing the signal bands of each part of the processing circuit on the receiving side, Figure 8 is a diagram for explaining video area detection, and Figure 9 is a diagram showing the signal bands of each part of the processing circuit on the receiving side.
The figure shows the processing circuit on the transmitting side, Figure 10 shows the sampling pattern of other transmission methods, Figure 11 shows the processing circuit on the transmitting side, and Figure 12 shows the processing circuit on the transmitting side. A diagram showing signal bands of each part, FIG. 13 is a diagram for explaining video area detection, I! FIG. 14 is a diagram showing a sampling pattern in a conventional transmission system, FIG. 15 is a diagram showing a processing circuit on the transmission side, and FIG. 16 is a diagram showing signal bands of each part of the processing circuit on the transmission side. 65.66--Frame memory 68...D/A converter 69...Output terminal 42 ・ 43 @ 44.48 ・ No. 45 46.47 ・ 49.71 ・ 50 ・ 51.63゜55゜53・ 56゜57 ・ 59゜60 ・・Input terminal・・A/D converter・・Interfield interpolation circuit・・Subsampling circuit・・Intrafield interpolation circuit・・2 frame memory・・Subtraction circuit・・RAM... Absolute value circuit... Comparator 58, 61. 62. 64. 67...Choice switch...Motion vector separation circuit...Field memory

Claims (3)

[Claims] (1) In the still image area, one frame of the television signal out of two consecutive frames is subjected to inter-field offset subsampling to form one frame signal, and in the moving image area, one frame of the television signal of the two consecutive frames is subsampled. The television signal is subjected to interfield and interframe offset subsampling to form signals of the first to fourth fields, and then interpolated between frames to form one frame signal, and the one frame signal is time-sampled. When decoding a transmitted television signal that is axially expanded and transmitted, multiple video region candidates are detected by interframe correlation between the frame of interest and multiple comparison frames, and a still image of the frame of interest is extracted from these multiple video region candidates. , a video region is determined, and in the still image region, inter-field interpolation is performed, and in the video region, inter-frame offset subsampling is performed, and then inter-field interpolation is performed.
A method for reproducing a transmitted television signal, characterized in that a fourth field signal is formed, time axis compressed, and the television signal of two consecutive frames is reproduced. (2) In the still image area, one frame signal is formed by taking the average between the frames of two consecutive frames of the television signal and performing inter-field offset subsampling; The 1st to 4th field signals are formed by inter-field and inter-frame offset subsampling of the John signal, and then interpolated between frames to form a 1-frame signal. When decoding a transmitted television signal that is expanded and transmitted, multiple video area candidates are detected by interframe correlation between the frame of interest and multiple comparison frames, and from these multiple video area candidates, a still image of the frame of interest, The video region is determined, and in the still image region, inter-field interpolation is performed, and in the video region, inter-frame offset subsampling is performed, and then inter-field interpolation is performed.
A method for reproducing a transmitted television signal, characterized in that a fourth field signal is formed, time axis compressed, and the television signal of two consecutive frames is reproduced. (3) two consecutive frames of the television signal are subjected to inter-field and inter-frame offset subsampling to form first to fourth field signals, and then inter-frame interpolation is performed to form one frame signal; When decoding the transmitted television signal in which the above-mentioned one-frame signal is time-extended and transmitted, multiple video region candidates are detected by interframe correlation between the frame of interest and multiple comparison frames, and these multiple video regions The still image and video region of the frame of interest are determined from the candidates, and inter-field interpolation is performed in the still image region, and inter-field offset subsampling is performed in the video region, followed by intra-field interpolation. ~
A method for reproducing a transmitted television signal, characterized in that a fourth field signal is formed, time axis compressed, and the television signal of two consecutive frames is reproduced.