JPH0752946B2 - Band compression method - Google Patents

Description

Description: TECHNICAL FIELD OF THE INVENTION The present invention relates to a band compression processing method performed when transmitting a screen signal.

2. Related Art and Problems Although band compression is often performed in data transmission and the like, in image information transmission in which information is sent in field or frame units, frame drop coding is one means of this band compression. This is done by sending screen signals A, B, C ... as shown in Fig. 1 and when the buffer memory is full at D, signal D
Discards, and next sends signals E, F .... Since the screen signal D is missing on the receiving side, the previous signal B is used instead. Although this method is simple, it is unavoidable that the same screen as before appears, so it feels uncomfortable, especially on a fast-moving screen. Therefore, there is also a method of compensating for the motion, that is, a method of using a proper method for band compression of a fast-moving screen, but the control becomes complicated.

It is an object of the present invention to provide a band compression method that does not cause unnaturalness even in a fast-moving screen, sufficiently reduces the transmission amount, and is relatively simple to control.

According to the band compression processing method of the present invention, the transmission side takes the difference between the encoding target screen signal of the input screen signal sequence and the prediction screen signal, encodes the resulting difference double-sided signal, and the difference code is passed through the buffer memory. In the band compression processing method for transmitting the encoded signal to the receiving side, the frame drop rate from the input screen signal sequence is determined according to the fullness or empty state of the buffer memory, and the differential encoding according to the determined frame drop rate. A control circuit that performs control, an encoder that performs either high-precision encoding or low-precision encoding of the difference screen signal based on an instruction from the control circuit, and an input screen signal string in multiple stages in screen signal units. An input screen signal selecting means for delaying and selecting an encoding target screen signal from a predetermined stage based on an instruction from the control circuit; and a difference based on a differential encoded signal output from the encoder. A prediction screen signal obtained by adding the prediction screen signal used for the present encoding is created and delayed in a plurality of stages in screen signal units, and a predetermined stage based on an instruction from the control circuit
Prediction screen signal selection means for selecting one or a plurality of prediction screen signals, and next encoding based on one or a plurality of prediction screen signals output from the prediction screen signal selection means and an instruction from the control circuit. And a filter means for interpolating and generating one predicted screen signal used for the above, and for a discrete screen signal selected by applying a plurality of types of frame dropping to the input screen signal sequence in accordance with the empty condition of the buffer memory. High-precision differential encoding is performed, and for the remaining intermediate screen signal, low-precision differential encoding is performed for the difference with the interpolated value obtained from the previous and subsequent screen signals. This will be explained in detail with reference to an example.

Embodiment of the Invention FIG. 2 is a diagram for explaining the principle of encoding, in which the left side of the chain line L shows the encoded area and the right side shows the area to be encoded. X indicates the final coded screen (field or frame) information, and 1,2 ... 3 indicate the 1,2,3 ... (1) is the continuous mode, that is, when the frame drop is not performed, the screen signals 1, 2, 3
... is encoded and transmitted without omission.
(2) is a 1/2 mode, in which the screen signal is dropped out every two, so that if one field and two frames are configured, one is dropped to form one field and one frame configuration. (3),
(4) has 1/3 mode, 1/4 mode, and 2 screen signals
One in three, three in four. In the past, the dropped pieces are simply discarded and not sent to the receiving side. That is, FIG. 3 shows a differential encoding (DPCM) circuit. The input signal S ₁ is compared with the prediction signal S ₃ at the subtraction point 12, and the difference is added to the encoder 10 as the output difference signal S ₂ . . This is sent to the receiving side via a buffer memory or the like not shown. The signals S ₂ and S ₃ are added at the addition point 14, and the predicted signal S ₃ is produced by adding the 1-frame delay elements 16, 18 ... When the buffer memory is full, the switch SW ₁ is switched from the contact a to the contact B side. Since the contact b is dropped to the ground, the difference is 0 and the signal S ₂ is 0. When one frame is dropped in this way, the switch SW ₂ is switched from the contact a to the contact b, and the output (S ₂ + S ₃ ) of the addition point 14 two frames before is used as the prediction signal S ₃ . When two frames are to be dropped in succession, the switch SW ₂ is switched to the contact point c, and the output of the addition point 14 before frame drop is used as the prediction signal.

In the present invention, the frame is not dropped in this way and is not sent to the receiving side at all, but the signal S ₃ that is predicted in consideration of the movement is constructed, and the difference in this case is very rough (the number of levels is Quantization).

This will be described with reference to FIG. 4, where, is a screen signal that does not drop frames in the conventional method, and ... Is an intermediate screen signal that drops frames in the conventional method. Therefore, in the conventional method, the difference with respect to the previous signal and the difference with respect to are sent,
, Is not sent, interpolation is performed on the receiving side, and as this interpolation method, the method of simply repeating the previous one and the method of taking the average (proportional) are often adopted.
In the case of 3 ', 4'and the latter, only points 2 ", 3", 4 "are made. In the present invention, all signals ,,,, are sent, except that the difference is sent with the same precision as the conventional one. ,,,, are processed as follows: That is, 3 ″ is calculated as the average value of and and the difference between , And coarsely quantize the difference and send it. Similarly, the average value of and is obtained, and the average value of
Difference from Is obtained, coarsely quantized with respect to the difference, and transmitted. In this way, the signal is effectively ...
The same result can be obtained by sending Since the difference between each of the signals and and is a coarse quantization with respect to the former, the difference is greatly reduced as compared with the case where the difference between the signals ~ is simply sent with the same precision of quantization.

Further, in the method of the present invention, the rate of dropped frames (as described above, it is not actually dropped) changes from continuous mode to 1/2, 1/3 ... mode at any time according to the fullness of the buffer memory. To do. Next, this will be described with reference to FIGS. 6 and 7 show a single drawing in a divided manner, and FIG. 6 is followed by FIG. 7, and these right ends are aligned. In FIG. 5, the same parts as those in FIG. 3 are designated by the same reference numerals. 22 is a switch that selects one of 6 inputs,
24 is a switch for selecting 3 out of 6 inputs, 26 is a control circuit, 28 is a filter section, and comprises multipliers 30 and 32 for coefficients α and β and an adder 34. The input signal S ₁ is input to the switch 22 through the 1-frame delay circuits FM _{5 to} FM ₁ and the inputs A _{6 to} A ₁ of the switch are sequentially set to 1 as shown in A _{6 to} A _{1 of} FIG. It will be delayed by each frame. This is X, 1,2, ……
Is the signal described in FIG. Control circuit for switch 22
When the control signal CONT1 is input from 26 and it commands 4, 3, 2, ... as shown in Fig. 7 (these are 1/4 mode, 1/3 mode, 1/2 mode ... Switch 22 makes the input selection shown in FIG. 6 S ₇ and produces output S ₆ . For example, the beginning of S ₇ is A ₃ , and at this time A ₃ is X, so S ₆ = X. The same shall apply hereinafter.

The switch 24 also includes 1-frame delay elements FM _{10 to} FM ₆ on the input side, and when the input C ₆ is input, signals C _{6 to} C ₁ obtained by sequentially delaying the input C ₆ by 1 frame time are input. Since the input C ₆ is the predicted signal of the signal S ₆ , the same numeral or letter is added with ′.
The control signal CONT2 is also input to the switch 24 from the control circuit 26, which is delayed by one frame time and the content is CO
Same as NT1. Switch 24 when receiving this signal CONT2
Produces outputs X ₁ , X ₂ , X ₃ having the contents shown in FIG. Ie CONT
2 is the output X ₁ contents C _5, C ₅ in Fig. 7 S _8, gives a selection instruction of ..., so that the output X ₁ is as illustrated X ', 4', 2 ',
...... becomes. Also given such selection instruction S _9, S ₁₀ is the output X _2, X _3, output X _2, X ₃ is as shown. Output X ₃
Constitutes the locally decoded output. The control signal CONT3 given to the filter 28 specifies the multiplication coefficients α and β of the multipliers 30 and 32 as shown. The output S ₃ of the filter unit 28 is S ₃ = αX ₁ + β
X ₂ , which is as shown in column S _{3 of} FIG.

As is clear from the S ₃ column in FIG. 6, S ₃ = X ′ is initially, and this is the prediction signal for S ₆ = 4. In this
Since the first four 1234 of the input signal S _{1 such} as X12345 ... are encoded in the 1/4 mode, as described above (see FIG. 2), the first one is the normal precision for the last signal 4. It is DPCM. Therefore, the predicted signal may be X '. In the next encoding, the second signal 2 is the object, and the prediction signal of this is made as (X '+ 4') / 2. Also in FIG. ₆ , S _{6 in} this case is 2, and S ₃ is 4 ′ × 1/2 + X′X1 / 2. The next is the first signal 1, the predicted signal is (2 '+ X') / 2, the next is the third signal 3, and the predicted signal is (2 '+ 4') / 2. This is also the case in FIG. The next encoding is performed in the 1/3 mode and the next encoding is performed in the 1/2 mode, and the encoding procedure is as shown in FIG. 2 and is the same in FIG. .

EFFECTS OF THE INVENTION As described above, in the present invention, precision differential encoding is performed on discrete screen signals selected by applying a plurality of types of frame dropping to the screen signal sequence on the transmission side, and the remaining intermediate Coarse precision differential encoding is performed on the difference between the screen signal and the interpolated value obtained from the preceding and following screen signals, and these are transmitted, so the transmission amount can be reduced while sending the full screen signal, resulting in image quality deterioration. The advantage is that this is not the case.

[Brief description of drawings]

FIG. 1 and FIG. 3 are schematic explanatory views of the piece drop, FIGS. 2 and 4 are explanatory views of the principle of the present invention, and FIG. 5 is a block diagram showing an embodiment of the present invention, FIG. 6 and FIG. The figure is an explanatory view of the operation of FIG. In the drawing, X, 1, 2, 3, ... Are screen signals, 10 is an encoder, S ₁ and S ₆ are input signals, S ₃ is a prediction signal, and S is a differential encoded signal.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Ken Okazaki, 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Toshihiro Honma, 1015, Kamedotachu, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited ( 72) Inventor Shinichi Maki 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Yu Fukuda 1015, Kamikodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (56) References Academic journal "Vol. 53, No. 3 1970, p. 336 ~ 337

Claims (1)

[Claims] 1. A transmission side obtains a difference between an encoding target screen signal of an input screen signal sequence and a prediction screen signal, encodes the resulting difference screen signal, and transmits the difference encoded signal to a receiving side via a buffer memory. In the band compression processing method for sending out, a control circuit for determining a frame drop rate from the input screen signal sequence according to the fullness or empty state of the buffer memory, and performing differential encoding control according to the determined frame drop rate. , An encoder for performing either high-precision encoding or low-precision encoding of a differential screen signal based on an instruction from the control circuit, and an input screen signal string delayed by a plurality of stages in screen signal units, and the control circuit Input screen signal selection means for selecting a coding target screen signal from a predetermined stage based on an instruction from the above, a difference based on a differential coded signal output from the above encoder, and a prediction used for the current coding. A prediction screen signal that is the sum of the measurement screen signal and the prediction screen signal is created and delayed in multiple stages in screen signal units.
Prediction screen signal selection means for selecting one or a plurality of prediction screen signals from a predetermined stage based on an instruction from the control circuit; one or a plurality of prediction screen signals output from the prediction screen signal selection means; Filter means for interpolating and generating one predicted screen signal to be used for the next encoding based on an instruction from the control circuit, and a plurality of types of frames for the input screen signal string depending on the empty condition of the buffer memory. High-precision differential coding is performed on the discrete screen signals selected by dropping, and for the remaining intermediate screen signals, a low-precision differential code is calculated for the difference from the interpolated value obtained from the preceding and following screen signals. A band compression processing method characterized by performing conversion.