FI92540C - High-resolution television's MAC decoder for the chrominance signal - Google Patents

Description

5,92540

High Definition Television Receiver Chrominance Signal MAC Decoder - Hogupplosningstelevisionens MAC Decoder for Chrominance Signal

The invention relates to a block of a high-definition television receiver in which original subsampling patterns are formed from chrominance samples of a received signal.

10 In the European High Definition Television System HDTV

(High Definition Television) transmission is via satellite or cable using a MAC system that is not compatible with any other traditional TV system. For this reason, current receivers need a special decoder to receive the MAC transmission.

The HDTV studio standard has an aspect ratio of 16: 9 and comprises 1250 lines displayed at an interleaving ratio of 2: 1 (or later 1: 1) and a field frequency of 50 Hz. Of the 20 lines, 1152 are active, visible on the image surface, and one line has 1440 luminance pixels and half less, i.e. 720 chrominance pixels. When transmitting, the image is encoded into a MAC-compatible HDMAC signal by compressing the luminance signal to one-fourth and the chrominance signal to one-eighth of a bandwidth-reducing encoder (for the luminance and chrominance, its own encoder can be returned). The line number of the transmission signal is thus half of the original number, i.e. 625 lines, and the interleaving ratio is 2: 1. The format of the input signal has been agreed to be 1250/2: 1/50 Hz at this time, but the interleaving ratio will probably change to 1: 1, in which case the compression will have to be further increased. Keeping an eye on the receiver, the so-called HDMAC signal is also transmitted. A DATV signal that transmits information about the motion-35 content of the image.

Compression is based on controlled subsampling, which is controlled by motion information. This means that the less motion there is in the image, the higher the spatial resolution is used and the greater the temporal resolution within the motion of the image at the expense of the spatial.

5

In the above-mentioned bandwidth reduction encoder BRE (Bandwidth Reduction Encoder), the image to be transmitted is divided into blocks of 16 * 16 pixels, so that the whole image is divided horizontally into 90 blocks and vertically into 72 blocks. The blocks, 10 in total, are treated differently on the basis of their business content. In addition, the processing is different for the chrominance signal components U, V than for the luminance signal Y. Since the image sensation in the human eye is largely based on luminance, the horizontal density of the luminance pixels 15 of the HDTV image is twice that of the chrominance pixels. In this case, the aforementioned 16 * 16 pixel block is reduced to a 8 * 16 pixel block when looking at chrominance pixels, and due to the sensitivity of the eye, as few as 20 subsampling patterns as chrominance cannot be used for luminance samples when lowering the bandwidth.

Thus, bandwidth reduction is more complex for luminance samples than for chrominance samples.

Referring now to Figures 1-3, it is described to reduce the bandwidth of the chrominance signal 25 to match the MAC signal. The • sampling frequency of the chrominance signal is 27 MHz. As stated above, the color resolution in the horizontal direction is lower than the luminance resolution. Since for the human eye the majority of the image resolution effect comes from the luminance component, in the complete HDTV image, the chrominance components in each line are reduced so that only every other pixel contains chrominance information. Since the U and V components are transmitted on the MAC channel in alternating lines, the vertical resolution of the color signal must also be lowered. In compression 35, the procedure is to ignore the second field of the image completely and to reduce the chrominance samples from the second field. Depending on the motion content of the HDTV sub picture (8 * 16 chrominance pixels) under consideration,

II

3,92540, a color signal is applied to one of three image processing branches, in each of which the signal is first filtered with anti-crease filters and then subsampled using a different sampling pattern in each branch. Thus, each of the 5 branches operates in its own mode, which are 80 ms mode, 40 ms mode and 20 ms mode.

If there is no movement, the signal is controlled. 80 ms mode branch where samples are taken from only one HDTV image.

10 The subsampling pattern (decimation) is as shown in Figure 1. In the picture, the squares represent the chrominance pixels of a complete HDTV picture, and for simplicity, the picture shows only 1/4 of the entire 8 * 16 pixel block. In the 80 ms branch, only those Kro miniature pixels with a number inside are included in the 15 subsampling patterns. It can be seen from the image that every other line of the image is ignored (i.e., the second field) and only every other chrominance pixel of every other line is transmitted to the MAC channel. In one MAC field of 20 ms, the samples are not placed in order from the line, but the samples 20 are scattered in different fields. The number inside the four indicates in which field the chrominance information of each pixel is placed. Thus, for example, the chrominance pixels of the first line are placed alternately in the MAC fields 2 and 4. It is essential to note that the chrominance information of one image is placed in 25 consecutive MAC fields. Conversely, this means that * when decoding an image at the receiver, information from four consecutive MAC fields is required to obtain a single original HDTV image.

If there are slow-moving objects in a part of the image, the chrominance signal to be compressed is directed to the 40 ms mode branch, where the field transmission time is increased from 20 ms to 40 ms. This is shown in Figure 2. It compresses two complete HDTV images into four 20 ms MAC fields. The sub-pattern 35 is now such that every other chrominance sample in every other line of one field of one image is included and samples in the second field of the image are not included at all. The numbers refer to the MAC field in which the chromium-4 92540 nance sample is placed. Thus, the samples of one line of the second field of the first HDTV image, denoted by the number 1, are placed in the MAC field 1, jumped over the same line in the same field, and the samples of the next line, denoted by the number 2, are placed in the field 2. is that the chrominance samples of the first HDTV image are in MAC fields 1 and 2 and the samples of the second HDTV image are in MAC fields 3 and 4. Compared to Figure 1, the chrominance information of two HDTV images is placed in a period of 80 ms. van instead. Thus, the 80 ms period has become temporal. Thus, what is essential in this 40 ms mode is that the under-sampled chrominance information of one image is placed in two consecutive MAC fields. Conversely, this means that when decoding an image at the receiver, information from two 15 consecutive MAC fields is required to obtain a single original HDTV image.

If there are fast-moving parts in the image, the compressable chrominance signal is routed to a 20 ms branch, where the chrominance information from the four 20 HDTV images is compressed into four consecutive MAC fields. For each image, samples are taken only from the second field and, from that point on, extremely sparsely, only every 8th sample from every other drink is included. The numbers indicate in which MAC field the 25 mic sample is placed. It is observed that the samples of one MAC field are always from one image. Thus, the chrominance information of the four HDTV images is included in the 80 ms period. Thus, the temporality is even greater than in the case of Figure 2, where two images are compressed into an 80 ms period.

30

According to the information provided by the motion detector, a switch is controlled, which switches only one mode at a time to the output of the BRE. This information in which branch the source signal is generated is encoded into a DATV signal, according to which the encoder directs the signal to the correct branch at the receiving end-35.

The compression of the transmission signal briefly described above will be apparent to one skilled in the art and the subject has been discussed in detail.

in detail e.g. F.W.P. Vreeswijk, M.R. Haghiri and C.M. Carey-Smith, HDMAC Coding for Compatible Broadcasting of High Definition Television Signals, 16th International TV Symposium, Montreaux, Switzerland, June 17-22-1989 and 5 F.W.P. Vreeswijk and M.R. Haghiri, HDMAC Coding for MAC

Compatible Broadcasting of HDTV Signals, Third International Workshop on HDTV, Turin, Italy, August 30 - September 1, 1989.

10 In the receiver's MAC decoder, the compressed picture 625/2: 1/50 Hz must be restored to the original HDTV format, ie 1250/2: 1/50 Hz. The analog television signal received in the decoder is converted into video, data and audio signals. The decoder can be called a Bandwidth Restoration Decoder (BRD) and is analogous to the BRE described above. The BRD encoder will now be described in more detail with reference to Figure 4, and the description relates to chrominance. The received sample frequency is 6.75 MHz and the BRD output frequency is 27 MHz. The encoder comprises a "line 20 deshuffler" block 41, at least 3 field memories 42 connected, e.g. to a cascade, a subsampling pattern converter 43 and three interpolator branches 44, 45 and 46. so that input-25 in slot 1 is temporally "oldest", i.e. first received in the field, and input 4 is temporally youngest, i.e. last received in the field. The SSPC generates the original subsampling patterns shown in Figures 1-3 from the input block, from which the missing chrominance pixels are interpolated in the interpolation branches. Each interpolator branch produces a complete HDTV field with respect to the wrong signal and the output of the selected branch is connected by a switch C to a demultiplexer 47, which separates the time-multiplexed U and V components from the chrominance signal. The components are the output signals of the BRD. The position of the switch 35 is controlled by the motion information contained in the received DATV signal.

6 92540

The chrominance components U and V are transmitted on the alternating line and the sampling frequency of the chrominance signal is 6.75 MHz.

In the deshuffler 41, the lines of the received MAC fields are arranged so that each line contains samples of only 5 of one original HDTV line. In this case, the SSPC 43 can be implemented without the use of line memories. The SSPC only needs 4 field inputs, so 3 field memories 42 are needed. to the first input of (input 1).

As stated above, it is the task of the SSPC 43 to return the subsampling patterns of the 15 different modes shown in Figures 1-3. It combines the successive MAC fields it receives and compiles a subsampling image of the original HDTV image from them. In addition, it interpolates any missing samples. In the received DATV signal, the SSPC receives information about the min-20 sub-sampling pattern it must form at any given moment. It can be seen from Figures 1-3 that a 20 ms mode subsampling pattern can be restored from a single MAC field, a 40 ms mode subsampling pattern can be restored when two MAC fields are available, but all four MAC samples are required to restore an 80 ms mode subsampling pattern. kenttåå. The SSPC always generates a subsampling pattern field during the time it takes to receive one MAC field, i.e., for one 80 ms base period (i.e., one HDTV picture), four subsampling fields are formed, numbered 30 to 1-4 and referred to below as 1TV HD , HDTV field 2, HDTV field 3, and HDTV field 4. It is conceivable that HDTV field 1 corresponds to MAC field 1, HDTV field 2 corresponds to MAC field 2, etc. For example, when the SSPC inputs have the MAC fields 1-4 of Fig. 1, the SSPC processes the HDTV field 2 from them. 35 If the next 40-ms mode block of the next picture sequence, Fig. 2, is received, after the processing of the HDTV field 2 of the first sequence, the situation is that the SSPC inputs have MAC fields 2, 3 and 4 in Figure 1 and Figure 2 in Figure 2

II

7 92540 MAC field 1, which already belongs to the next image sequence.

The SSPC should further process the 80 ms mode block into the HDTV field 3, although not all of the 80 ms mode MAC fields of the same image sequence block are available. Myohem-5 min, after 20 ms, fields 3 and 4 of Figure 1 and fields 1 and 2 of Figure 2 are input, but the SSPC should still process the 80 ms mode block to the HDTV field 4, although not all of the 80 ms mode blocks MAC fields of the same image sequence are no longer available.

10

The problem is how to restore the 80 ms mode subsampling pattern when the SSPC has to switch from the 40 ms or 20 ms subsampling pattern to the 80 ms subsampling pattern, or when the SSPC has to move from the 80 ms subsampling pattern. ms pattern return to form a 20 ms or 40 ms subsampling pattern. In other words, the problem arises when viewing the same area with respect to time on the image surface before or following the "stationary image area" in the motion motion decision, where 20 created all four fields necessary to reproduce the under-sampling pattern of the same block in the 80 ms mode. at hand.

The prior art subsampling pattern restorer 25 The SSPC solves this problem simply by horizontally copying the missing samples from the nearest adjacent known samples. Copying is performed when one of the consecutive 80 ms MAC fields of the same block is missing. If two consecutive fields are missing, the interpolation of the missing sample-30 paths is given to a 20 ms interpolator, which improves the result. However, copying results in the errors shown in the image, because when the image or, rather, the block changes from stationary motion, i.e. being within the boundary of these blocks and copying, the pixels are missing from stationary (80 ms) pixels, horizontal lines and horizontal lines appear in the image, The vertical and diagonal edges and lines will have horizontal lines and the vertical and diagonal edges may become toothed.

92540 θ

The same applies when the block changes from stationary to stationary. The image error is visible in the entire block area.

5 These problem areas are illustrated in Figure 5. It looks at the same block (part of the image) over time. The block is initially stationary, with the information being placed in four consecutive 80 ms mode MAC fields, represented by numbers within the block. The chronological order of the fields 10 is from left to right. Each MAC field has its own relative field number of 1, 2, 3, or 4. In the next step, the block becomes motion-containing, so the information is placed in four consecutive 40 ms MAC fields 1, 2, 3, and 4. Then again the block becomes stationary-15, so the information is in the 80 ms mode MAC fields 1, 2, 3, and 4. Next, the motion content is large, so the block information is in the fourth consecutive 20 ms mode MAC field 1, 2, 3, and 4. The number below each field indicates which mode the field belongs to. Thus, successive field-20s 1-4 always form one block, as shown in Figures 1-3. Each block 1-4 of the four fields is associated with its own motion information, which is received on a separate DATV channel, decoded and transmitted to the SSPC, whereby it is able to return the correct subsampling pattern based on the DATV information.

When the input to the SSPC has fields marked with an arc a, the SSPC should first process the HDTV field 2 to form an 80 ms mode subsampling pattern, which it does because 30 consecutive fields of the same mode are available.

But when the entry has fields marked with an arc b,; only three consecutive 80 ms mode fields are available.

The known SSPC then generates the missing samples required for processing the HDTV field 3 by horizontal copying, which results in image errors. When the input has fields marked with an arc line c, i.e. two 80 ms mode fields of the block to be processed are missing, the known SSPC allows the interpolation of the missing samples to a 20 ms interpolator instead of an 80 ms interpolator (i 9 92540). field interpolation is given to a 20 ms interpolator only when switching from stationary mode to motion mode, and even then only for the last 5 fields.These cases are denoted by circled mode numbers in Figure 5. However, image errors occur when moving from mode to mode. These image errors are not corrected by the prior art subsampling pattern restorer SSPC 10.

The present invention provides a way in which image errors caused by the use of the prior art can be eliminated. The invention is characterized by what is stated in claim 1.

15

The inputs of the sampler of the subsampling pattern can be marked as the first, second, third and fourth inputs. The fourth input comes in the MAC field just to be received, the third comes the previous Kent-20 delayed with a field delay, the second a field delayed with two field delays, and the first comes a field delayed with three field delays. In other words, the input is affected by four consecutive MAC-Fields such that input 1 has the oldest field. According to the invention, the sub-sampling pattern restorer SSPC is forced to process the field in a 20 ms field whenever a) when the situation where the MAC field of the 80 ms mode affected by all inputs has changed to a situation where the fourth or fourth and third inputs are affected by either 40 ms or 20 ms field, and 30 b) when the last three inputs are affected by the MAC fields of the 80 ms mode and the first input is affected by: the field of another mode.

The operation of (a) means that the stationary block is changing to a motion-containing block, while the operation of (b) means that the motion-containing block is changing to a stationary block.

10 92540

Since the DATV information associated with each received MAC field indicates the block mode, a function can be constructed by simple logic to monitor the status of the inputs and force the stationary image of the 20 SSPC 20 ms interpolator while the conditions in a or b are in effect. when interpolating HDTV fields in the area.

The invention is illustrated by the accompanying figures, in which Fig. 1 shows a subsampling pattern of 10 chrominance samples of a stationary block, Fig. Fig. 6A is a table of SSPC input combinations when processing different HDTV fields, Fig. 6B is a table corresponding to Table 6A, and the cases in which the processing of the invention is used are shown, Fig. 7 is a view corresponding to Fig. 5. , where points are used to process the invention, Fig. 8 illustrates the use of a DATV signal in decision making, Fig. 9 shows a block diagram of one possible constraint arrangement, and Fig. 10 is a block diagram of another possible constraint-3. 0 tusjårjestelystå.

Figures 1-5 have already been described above and will be referred to below as necessary. Table 6A shows the problem areas involved in using the prior art SSPC. In the left column of the table, "Field" refers to the HDTV subsampling pattern field to be processed, "RFN" refers to the number of the MAC field at each input of the SSPC (see Figures 1-3), and "Case" refers to a combination of input fields during processing. that is, in the HDTV field 1, 2, 3, or 4. Cases in which the prior art causes problems are shown in bold. Let's start with an example from case 6, i.e. Case 6. In this case, the input fields of the SSPC are the MAC fields 1, 2, 3 and 4 of the 80 ms mode block. The SSPC has all the fields of this mode in the four fields of this block, so it can easily form i.e. HDTV field 2. This corresponds to the situation in Figure 5, where the 10 inputs are the fields in the first four fields, i.e. the fields marked with an arc a, from the top left.

After a period of 20 ms, the MAC fields have "shifted" one input forward and the fourth input already has the next MAC block, in this case a 40 ms mode block, in the first MAC field, the number (RFN) of which is thus 1. Referring to Figure 5, the input is now the fields marked with an arc b, i.e. 80 ms mode fields 2, 3, 4 and 40 ms mode field 1. Table 6A now moves to Case 10 20 and processes 80 ms mode HDTV field 3. It is observed that all four fields of the 80 ms mode of the block are no longer available, in which case the 80 ms interpolator of the SSPC does not have all the necessary samples available. According to the state of the art, as mentioned in connection with Figure 5, the missing samples would be copied by horizontally copying from existing adjacent samples, so that the final HDTV image would be image defective. Accordingly, according to the invention, the SSPC is forced to form a 20 ms mode subsampling pattern and a 20 ms interpolator branch is connected to the output of the 30 BRD (switch C Fig. 4). This means that the 80 ms mode HDTV field 3 is processed by a 20 ms mode interpolator. The samples it needs are all available, so there is no need to undertake horizontal copying. Tål-Ιδϊη image errors are decreasing.

35 After another 20 ms period, the fields have "shifted" by one. input forward, with the first and second inputs having the first block, a 40 ms mode block, 12 92540 in the first two fields, the numbers (RFN) of which are thus 1 and 2. Referring to Figure 5, the input is now the fields marked with an arc c, i.e. 80 ms: n mode block fields 3, 4, and 40 ms mode block fields 1 and 2. In Table 6A, 5 has now been moved to Case 15 and further processing the same 80 ms mode block field, HDTV field 4. It is observed that no more than two of these 80 The ms mode field is available.

Accordingly, in accordance with the invention, the SSPC interpolator is now forced to further interpolate in a 20 ms mode and processed into 10 80 ms mode HDTV fields with a 20 ms mode interpolator for which the required samples are all available.

After another 20 ms, the input combination is according to Table 6A Case 4, so it is processed in a 40 ms field. After the four 15 received MAC fields, the input fields are the fields marked with arc d in Figure 5, i.e. 40 ms mode field 4 and 80 ms mode fields 1, 2 and 3. In Table 6A, the situation corresponds to Case 2 and the interpolator should process 80 ms mode HDTV field 1. Now, horizontal-20 sampling of the samples is not used, but the interpolator is forced into the 20 ms mode, whereby the necessary samples are available. Thus, the 80 ms mode HDTV field 1 is processed by a 20 ms mode interpolator. In the next step, after one 20 ms period, all 80 ms mode fields in this block are available, 25 so the situation is according to Case 6 and HDTV field 2 is normally processed by an 80 ms mode interpolator.

In the next step, the inputs are the fields marked with the arc line e, i.e. the 80 ms block HDTV-30 should be processed in field 3, but only the block three in this mode field are available. The fourth input field is the MAC field 1 of the block 20 ms mode. In Table 6A, the situation corresponds to Case 11. Instead of horizontal copying of samples according to the prior art, the interpolator is forced into 20 ms mode, whereby all 35 samples it needs are available.

The interpolator is likewise forced from the 80 ms mode to the 20 ms mode also in the next step, where the input fields

II

13 92540 is marked with an arc line f in Fig. 5. The HDTV field 4 of the 80 ms mode is to be processed. The situation is therefore in accordance with Table 6A Case 16. Again, the interpolator is forced into a 20 ms mode so that all the samples it needs are available.

5

For any sequence of successive blocks, a suitable Case can be found in Table 6A, and it can be stated that the maximum number of possible cases is 18. In the table, a small horizontal line indicates that there can be any mode field at that point.

In Fig. 7, which corresponds to Fig. 5, the mode markings of the fields of the 80 ms blocks, which the corresponding field of the sub-sampling pattern is processed in the forced 20 ms: η 15 mode, are marked in a circle. In Fig. 5, the fields in which the 20 ms mode processing is used in the prior art solution are marked in a corresponding manner. A comparison of the figures clearly shows the difference between the prior art and the invention.

Correspondingly, in the table of Fig. 6B, which corresponds to the tau lock of Fig. 6A, the cases in which a 20 ms interpolator is forcibly used in the processing of the stationary field are shaded. Thus, it can be seen that in processing field 1, the cases Case 2 and Case 3 in Figure 6A have been replaced by Case 5, when processing in field 3 the cases Case 10 and Case 11 replace Case 13 and when processing in field 4 there are cases Case 15 and Case 16 has been replaced by Case 18. The maximum number of different cases has therefore been reduced from six to 12.

30

Figure 8 shows the principle of forced processing. As is known, the received DATV signal is decoded, whereby the motion information of each block is obtained. The DATV information is fed to three cascaded field memories 35 84, 85 and 86. Since the HDTV image is divided into 90 * 72 blocks, and each block is associated with its motion information, a DATV memory size of 90 * 72 items is sufficient. The motion information associated with each image is thus stored in the DATV field memory. As before 14 92540

said, the received MAC image signal is stored in three field memories, memories 81, 82 and 83 in Figure 8, of which the SSPC reads the received MAC fields with the current received field being one SSPCm input No. 5. The output of the field memory 82 corresponds to the field that the SSPC

processes, i.e. if there is a MAC field 2 in the field memory, the HDTV field 2 must be processed in the field 2. The contents of the DATV field memories and the MAC field memories are completely synchronous, so e.g. at the same time, the motion data of the current block of the field to be transferred can be read from the DATV field memory 84. To this end, a control logic is arranged which monitors the motion information 15 obtained from the outputs datv (1), datv (2) of the DATV field memories 84, 85, as well as the motion information information datv (4) of the field just received. When the logic detects that one of the shaded areas shown in Table 6A is valid, it provides a first control signal that forces the SSPC to generate a 20 ms subsampling pattern and a second control signal that forces a 20 ms branch to output the BRD. , i.e., turn on switch C, Figure 4, 20 ms decoder output.

Forcing is performed when the following logical condition is in effect: 25 Stationary image processing is performed in 20 ms mode when

datv (2) = "80 ms block" and datv (l) is different from "80 ms block" OR

30 datv (2) = "80 ms block" and datv (4) is different from "80 ms block"

Forcing can be done in several different ways. In the first embodiment shown in Figure 9, this inference may be made in the known DATV decoder 35 after the DATV decoder 91 as a subroutine thereof.

The modifier block makes the endpoint of the interpolator to be used. The motion data decoded by the DATV decoder is fed to the field memories A, B, C as well as the data just received

II

15 92540 also for the DATV editing block. The memory arrangement is the same as in Fig. 8. The output data "branch decision", which is the same as the output of the DATV field memory B and which corresponds to the output of the field memory 82, Fig. 8, is applied to the SSPC 942 and the demultiplexer 5 947. The output data is set to SSPC. In addition, the datv (l) and datv (4) data and the field number RFN to be processed are exported to the SSPC. The RFN information and the input signal of the DATV field memory A are additionally applied to the line deshuffler 101. The DATV modification block 92 examines whether the aforementioned logical condition can have a mass of 10 and, if valid, forces the end signal to "fail" 20 ms. mode.

According to another embodiment shown in Figure 10, the SSPC makes a forcing decision. It needs the RFN number of the field 15 to be processed, as well as the datv (i), datv (3) and datv (4) information. Here it receives from the DATV decoder 108 both the DATV field memories B and C. The output of the field memory B is normally a branch end signal, but if said logical condition is met, the monitoring logic in the SSPC raises a flag indicating "forced 20 20 m". The demultiplexer 107 monitors the flag and when it detects that the flag is up, it does not pass the branch end signal received from the DATV memory B but switches to the 204m branch 104. The flag is raised when there is a temporal limit in the received MAC fields. Thus, the flag is raised 25 in the case of Case 2, 3, 10, 11, 15, or 16 of Table 6A. The flag forces the SSPC to 20 ms mode regardless of the branch information.

The present invention does not cause major changes to the prior art under-sampling pattern restorer SSPC. A variable must be added to the control logic to indicate that the 20 ms interpolator is forced, or the control logic changes the received problem point conclusions to 20 ms mode. The invention can prevent the sawing of the vertical and diagonal edges of saturated colors, the so-called "postage stamp effect", as well as horizontal lines of lines at the vertical and diagonal edges. In addition, the various processing options are reduced from eighteen to twelve.

Claims (10)

16 92540 A method for controlling an HDMAC decoder for generating a stationary high-definition picture area from a MAC video signal comprising chrominance samples divided into picture areas from a sub-sampled MAC picture divided into picture areas and sub-sampled at the same time, in which encoder is derived: (SSPC) such that its first input has a first field in time, the second input has a second field in time, the third input has a third field in time and the fourth input in time has a last field, and the lowest input in time has a last field, and the lowest one of the three convex subsampling patterns A1-15, - the formed subsampling pattern is passed to three signal link processing branches, the first branch of which forms a high-definition image of the stationary image area mu the second branch forms a low-motion picture of the sub-sample pattern formed from the slow motion picture, and the third branch forms a high-speed motion picture of the low-motion picture. high-definition television image • one at a time to the output of the decoder, characterized in that when the samples of the second input belong to the stationary image area, the sub-sampling pattern restorer (SSPC) is forced to form a fast motion when the samples of the first or last input field belong to the image area containing the movement. 35 A method according to claim 1, characterized in that the forcing is performed when the second, third and fourth inputs contain samples representing the stationary image area II 17 92540 but the first input has samples representing the image area containing motion. A method according to claim 1, characterized in that the forcing is performed when the first and second inputs have samples representing a stationary image area but the third and fourth inputs have samples representing a motion image area. A method according to claim 1, characterized in that the forcing is performed when the first, second and third inputs have samples representing a stationary image area, but the fourth input has samples representing an image area containing motion. 15 A method according to claim 1, characterized in that each MAC field contains the motion information of each image area (DATVj) stored in a data field memory from which it can be read when forming the subsampling pattern of the area. A method according to claim 5, characterized in that the image area of the field entering the first input of the subsampling pattern restorer corresponds to the first motion data datv (1), the image area of the field entering the second input corresponds to the second motion data datv (2), and the image area of the fourth input field corresponds to the fourth motion data datv (4), in which case the forcing is performed when datv (2) = "stationary area" and datv (l) = "area containing motion", or datv (2) = "stationary area" and datv (4) = "area containing movement 30". 7. An HDMAC decoder for generating a high-definition picture from a MAC video signal comprising chrominance samples from a high-definition picture divided into picture areas and sub-sampled according to the motion content, the encoder comprising: field so that its first input 18 92540 (1) has a temporally first field, the second input (2) has a temporally second field, the third input (3) has a temporally third field and a fourth input is the last time in the field where the 5 subsampling pattern restorer (SSPC) forms one of the three original subsampling patterns from the samples derived from its input, and - three signal processing branches (104, 105, 106; 944, 945, 946) with the connected (10) 946) form a high-resolution image of a subsampling pattern formed of a stationary image area, and a second branch (105; 945) generates a high-definition image from a subsampling pattern formed of a slow-motion image area, and a third branch (104; 944) 15 generates a high-definition image from a sub-sampling pattern formed from a fast-moving image coden. 20. first means (A, B, C) for recording the motion content (datv (l), datv (2), datv (4)) of the image areas entering the first (1), second (2) and fourth inputs, - second means (92; 103) coupled to the first means and provided with a forced branch end signal 25 forcing a stationary high definition image area to be formed. allowing the sub-sampling pattern restorer (103, 943) to form a fast-moving image area with the subsampling pattern to force the third signal processing branch to output to the encoder whenever the motion of the first input field (s) ) indicates the movement in the field. HDMAC decoder according to claim 7, characterized in that the first means comprise successive sources of the first (A), second (B) and third (C) DATV field memory and the third memory (C) indicating the first incoming input. The motion content of the MAC field (datv (l)), the output of the second li 19 92540 memory (B) indicates the motion content of the second incoming MAC field (datv (2)) and the input of the first memory (A) indicates the motion content of the last incoming field (datv (4)). ); wherein the output of the second memory (B) (datv (2)) is at the same time a branch end signal. An HDMAC decoder according to claim 7, characterized in that the second means (92) comprise a DATV modification block which provides both a branch end signal and a forced branch end signal. HDMAC decoder according to claim 8, characterized in that the second means are included in the subsampling pattern restorer (103), whereby it provides a forced branch end signal. 20 92540