JPS6149861B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention performs band compression using video signals.
The purpose of this paper is to simplify the circuit configuration and realize a DPCM device with less quantization error by improving the characteristics of the quantizer that quantizes the difference between the input value and the predicted value in the DPCM device. Conventionally, as a predictive coding method in a DPCM device that performs band compression using a video signal, a one-dimensional, one-point prediction method, which is simple in configuration and economical, has been used. In FIG. 1, it is assumed that the prediction signal for the input signal x _o is three points before, that is, (X _o-3 ). The conventional configuration consists of an encoding block A, a transmission path B, and a restoration block C, as shown in FIG. 2, and each block will be described below. Encoding block A generates a signal x _o (hereinafter referred to as input value x _o ) obtained by A/D converting a video signal, and an encoded value _{R o} that contains a quantization error due to DPCM encoding processing for input value x o. an arithmetic unit 1 which obtains a difference value e _o from the predicted value x _o , and a quantizer which takes the difference value _{e o} as input and obtains a predetermined quantized value Q _o according to the characteristics shown in Table 1, for example. a predictor 3 that obtains a predicted value x _o of an input value x _o ; and a code that receives a quantized value Q _o as input and obtains a predetermined encoded value C _o according to the characteristics shown in Table 2, for example. 4, and a synchronization signal extraction circuit 5 that extracts the synchronization signal S of the video signal by using an AND element using the A/D-converted D ₁ to D ₈ , that is, the input value x _o as input, as shown in FIG. _{As shown in FIG . _ _ _ _} and the synchronization signal S is input, and during the period when the synchronization signal S is signal "H", the encoded value _Co (C ₄ , C ₃ , C ₂ , C ₁ ) is selected and the output terminal (C ₄ , C ₃ , C ₂ , C ₁ ), and during the period when the synchronization signal S is no signal "L", selects the synchronization code P (P ₄ , P ₃ , P ₂ , P ₁ ) and outputs (P ₄ ) to the output terminal. , P ₃ , P ₂ , P ₁ ).

【table】

【table】

[Table] Decoding block C has a coded value C' containing a synchronization code P (P ₄ , P ₃ , P ₂ , P ₁ ) sent from signal switching element 7 of coding block A via transmission line B. A decoder 8 takes _o as an input and obtains a predetermined decoded value _Do according to the decoding characteristics shown in Table 3, for example, and adds the decoded value _Do and the predicted value _yo to encode the encoded block A.
an arithmetic unit 9 that obtains a restored value y _o that is restored to the encoded value R _o ; a predictor 10 that delays the restored value _{y o to} obtain a predicted value y _o ; As shown in Figure 5, the synchronization code P (P ₄ , P ₃ , P ₂ , P ₁ ) contained in the encoded value C′ _o is detected by a matching circuit consisting of an exclusive OR element and an AND element. The synchronizing signal reproducing circuit 11 reproduces the synchronizing signal S'.

[Table] Also, d ₁ , d ₂ ...d ₇ are delay elements that each delay the input signal by one point. Also, the synchronization signal S
and S' are used to reset the circuit. In Fig. 2, the signal sequence of each part is e _o = x _o - x _o ... (1) Q _o = Q・e _o ... (2) (output of the difference value e _o quantized by quantizer 2) ) R _o = Q _o + x′ _o ……(3) (However, x′ _o = R _u-3 ) C _o = C・Q _o ……(4) (The quantizer Q _o is encoded by the encoder 4 (output obtained by decoding _{the encoded value C o by} the decoder 8, which is equal to the quantized value Q _o . _{D o =} Q _o ) y _o = D _o + y _o ... (6) (y _o = R _o , where y _o = y _o-3 ) are respectively expressed. Based on the above formulas (1) to (6), the initial value x ₀ of the input value x _o is set to 0, and then Q _o =
0, R _o = 0, x _o = 0, yo ₌ 0, _yo = 0 as the initial values, and the input values are changed to x _o = 0, x ₁ , x ₂ ... x _i The time chart is shown in Figure 6. In encoding block A, the difference value e _o between the input value x _o and the predicted value x _o is quantized into a quantized value Q _o by the quantizer 2, and then delayed by one point by the delay element d ₁ . . In the predictor 3, after adding the quantized value Q _o and the predicted value x _o , the result is delayed by three points by delay elements d ₂ , d ₃ , and d ₄ . Therefore, the delay amount T ₁ for the input value x _o becomes T ₁ =d ₁ +d ₂ +d ₃ and the delay amount T ₂ for the quantized value Q _o becomes T ₂ =d ₂ +d ₃ +d ₄ . In encoding block A, the quantized value P o- ₃ of X _{o-3 three} points before is used as the predicted value x _o of the input value x o, and the difference value e _o between the input value x _o and the predicted value x _o is calculated. Similarly, in decoding block C, the restored value _{y o-3} three points before is used as the predicted value y _o of the restored value y o, and the decoded value _Do is added to the encoded value R. It can be restored to the restoration value y _o which is equal to _o .
The delay amount T ₃ for the decoded value D _o is T ₃ = d ₅ + d ₆ + d ₇
becomes. In Fig. 6, the input value x _o is 0, x ₁ , x ₂ ,
When changing x ₃ , . . . x _i , the predicted value x _o becomes 0 until n=3. Therefore, the difference value e _o is e ₁ = x ₁ when the input value x _o is x 1, and e ₂ = x ₁ when the input value x _o is x ₂ .
x ₂ , and when the input value x _o is x ₃ , e ₃ =x ₃ . Also, when the input value is x ₄ , the predicted value x ₄ for the input value x ₄ is Q x ₁
becomes. Therefore, for the input value x _o =x ₅ , the predicted value x ₅ =Q·x ₂ and for the input value x _o =x ₆ , the predicted value x ₆ =Q·x ₃ . The quantized value Q _o is sent to the encoder 4
The decoder 8 decodes the encoded value C _o encoded by the decoder 8 to obtain a decoded value D _o . Since the decoded value D _o is equal to the quantized value Q _o , the decoded value D _o is 0, Q ₁ ,
Q ₂ ...Q _i . The predicted value y _o becomes D _o =0 until the decoded value D ₃ =Q ₃ , and when the decoded value D ₄ =Q ₄ , y ₄ =
When Q ₁ , decoded value D ₅ = Q ₅ , y ₅ = Q ₂ , decoded value D ₆ =
When Q ₆ , x ₆ =Q ₃ is obtained, and y _o =Q ₁ , Q ₂ , and Q ₃ are predicted values for the restored values y _o =y ₅ , y ₆ , and y _{7 ,} respectively. When performing DPCM processing using video signals,
After the video signal is sampled by an A/D converter, it is quantized to, for example, 256 levels (8 bits) to provide an input value x _o . In FIG. 7, when the staircase wave video signal a is quantized, x _o =0, x _o =
59, an example of x _o =73...x _o =200 is shown. If typical examples of the quantization operation are taken at points b and c, input values x _o shown in FIG. 7b and c are obtained. When a time chart is created using the quantization characteristics shown in Table 1, the encoding characteristics shown in Table 2, and the decoding characteristics shown in Table 3, the time chart in FIG. 8 near point b in FIG. Also, FIG. 9 is obtained as a time chart near point c in FIG. 7. In Figures 8 and 9, when the input value x _o is constant and continuous, the input value x _o and the predicted value x _o are equal, so the difference value e _o = 0, and when the input value x _o changes, the input value x A difference occurs between _o and the predicted value x _o . As shown in Table 1, when the difference value e _o =0, the quantized value Q _o =0 and as shown in Table 2, the encoded value C _o =15 is transmitted. Furthermore, when a change occurs in the input value x _o , the difference value e _o =x _o -x _o between the input value x _o and the predicted value x _o is quantized, then encoded, and transmitted. An arithmetic unit 3-1 that obtains the predicted value R _o =Q _o +x′ _o of the encoded block A and the added value y of the decoded block C.
The arithmetic unit 9 that obtains _o = D _o +y _o has a calculation result of “+
255", that is, in the case of an overflow, the result is set to "+255", and if the operation result is a "negative value", the result is set to "0". If a "negative value" occurs, as shown in FIG. 9 with the predicted value _Ro -3 and the restored value _yo = -5, the predicted value _Ro = 0 and the restored value _yo = 0 are set. As shown in Figure 7, during the synchronization signal period t,
A case where a transmission error occurs in the encoded value _Co will be described. For example, encoded values C _o (C ₄ , C ₃ , C ₂ ,
The case where an error occurs in any bit of C ₁ ) and its contents are reversed will be explained using FIG. For example, the encoded value C _o = 8 = binary number (1000) means the decoded value D _o = -64, but an error occurs in the C ₃ bits and the content is reversed, resulting in the encoded value C _o = 12 = binary number (1100), the encoded value C _o = 12 means the decoded value D _o = -9, so the restored value y _o = 50, and the normal synchronization signal period t There is a difference of “+50” between the restored value y _o = 0 at , and the restored value y _o = “+
50" is repeatedly restored. For example, if an error occurs three times in a row, the normal encoded value _Co = 8 becomes _Co = 12.
The synchronization signal period t is the restoration value y _o = “+
50", and all subsequent restored values y _o will be biased to "+50". Therefore, the waveform will be as shown by the solid line in Figure 7d. As a result,
When the restored value y _o is converted into a video signal by a D/A converter and displayed on a television receiver, a normal image cannot be obtained because the video signal is biased. In order to prevent the synchronization signal from being biased by transmission errors, a code that does not use encoding in encoding block A, e.g. binary "0000"
is generated by the synchronization code generator 6, switched by the synchronization signal S using the signal switching element 7, inserted as a synchronization code in the synchronization signal period t, and synchronized with the synchronization code binary number "0000" in the decoding block C. It is detected as a synchronizing signal S' by the signal reproducing circuit 11, and the restored value y _o and predicted value y _o are forcibly reset by the synchronizing signal S', and the restored value y _o is set to "0". Although the one-dimensional prediction method has been described above, the same result can be obtained for the higher-order prediction method. However, according to the conventional method, it is necessary to insert a specific code as a synchronization code in order to prevent the synchronization signal period t from being biased due to a transmission error. 6 and the signal switching element 7, and the synchronizing signal reproducing circuit 11 is required in the decoding block C, which increases the number of parts and causes an increase in cost. In addition, in order to provide a special synchronization code, the encoded value C _o is 15
Since the quantization value Q _o is also limited to 15 levels, the quantization characteristics have become coarse. The present invention eliminates the above-mentioned conventional drawbacks. One embodiment will be described below with reference to the accompanying drawings. As shown in FIG. 11, the configuration of the present invention includes a first arithmetic unit 1 that obtains a difference value _{e o between} an input value x _o and a first predicted value x _o for the input value x _o ; For example, as shown in Table 4, the difference value e _o
a quantizer 2 having a characteristic of a quantized value Q _o =-1 when e _o =0 or e o =-1; a first predictor 3 that obtains a first predicted value x _o for an input value x _o ; An encoder 4 that receives the quantized value Q _o as input and obtains a predetermined encoded value C _o according to the encoding characteristics shown in Table 5, for example.
The encoded value C _o obtained by the encoder 4 of the encoded block A via the transmission line B is input, and a predetermined decoding is performed according to the decoding characteristics shown in Table 6, for example. Decoder 8 that obtains the encoded value D _o
and a second arithmetic unit 9 which obtains a restored value y _o by adding the decoded value D _o and the second predicted value _{y o} to restore the encoded value R _o ; The decoding block C consists of a second predictor 10 that obtains a second predicted value y _o .

【table】

【table】

【table】

【table】

[Table] In the present invention, point _C is obtained when the staircase wave video signal a shown in FIG. A nearby time chart is shown in FIG. 12 and will be explained. Due to the quantization characteristics of quantizer 2 (see Table 4), when the input value x _o suddenly changes from a continuous constant value, for example x _i = ₅₉ , to The value y _o is y _i =9, y _i+1 =2,
Although it gradually changes to y _i+2 =0, it does not affect the displayed image. Next, a case where a transmission error occurs in the encoded value _Co will be described. Coded value C _o (C ₄ , C ₃ , C ₂ ,
The case where an error occurs in any of the bits of C ₁ ) and the bit contents are inverted will be explained with reference to FIG. For example, if an error occurs three times in succession in the C ₃ bits of the encoded value Co ₌ 7 = binary number (0111), and the content becomes _Co = 3 = binary number (0011), the encoded value The information that means Co ₌ 7 is the decoded value _Do = -1, and the information that means the coded value Co ₌ 3 is the decoded value _Do = 20, so the restored value _yo = 79, which is the normal A difference of "+21" occurs between the restored value y _o =58, and the subsequent restored value y _o is biased by "+21" from the normal restored value. However, during the period when the video signal is flat (including the synchronization signal period t), the state of the difference value e _o =0 continues, so the quantized value Q _o =-1 continues.
In the restoration block C, this is the decoded value D _o
=-1 will continue, and the restored value _yo biased to "+21" will decrease by one level for every three points. If there is no flat portion in the video signal period of the video signal, the required operation is performed during the synchronization signal period t. The time required for convergence to the restored value y _o =0 (0 level of the video signal) varies depending on the biased level, but the input value x _o =0 during the synchronization signal period t
The number of consecutive times (sampling number) is about 50, and if the biased level is B1, the normal state is restored in the B1/50/3rd synchronization signal period. The maximum time required for convergence is (255/50/3)≈16th synchronization signal period, but it can be minimized by assigning the code of the encoded value _Co. Table 7 shows the difference value when an error occurs in each bit with respect to the encoded value _Co.

[Table] If an error occurs in C ₄ bits when encoded value _Co = 9 = binary number (1001), encoded value _Co = 1 = 2
The result is a base number (0001), and the difference value is "+98", which is the worst situation. Therefore, the time required for convergence is (98/50/3)≈6th synchronization signal period. If the difference value becomes a “negative value”, the next restoration value y
Since _o becomes a negative value, the circuit is reset, and the restored value y _o becomes 0, and the normal state is restored in the next synchronizing signal period. As described above, according to the present invention, when the difference value e o between the input value x _o and the first predicted value x _{o is} 0 in the quantization characteristic of the quantizer 2, the quantized value Q _o = -1
By assigning this to B1, when a transmission error occurs, the level B1 biased to a positive value can be approximated to the normal level using the flat part of the video signal (including the synchronization signal period t). Therefore, the synchronization signal extraction circuit 5, the synchronization code generator 6, the signal switching element 7, and the synchronization signal regeneration circuit 11 can be removed, reducing costs and eliminating the need to specially provide a synchronization code in the encoded value _{Co to} be transmitted. Since there is no coded value C _o is not limited to 15 levels, but 16
The quantization characteristic becomes denser, and a transformed value y _o that is much closer to the input value x _o can be obtained.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing the relationship between sample points and prediction points in the one-dimensional one-point prediction method, Fig. 2 is a block diagram showing the configuration of a conventional DPCM device, and Fig. 3 is a synchronization signal extraction circuit 5 in the conventional configuration. A diagram showing a specific example of
Figure 4 shows a synchronous code generator 6 in a conventional configuration.
FIG. 5 is a diagram showing a concrete example of the synchronous signal reproducing circuit 11 in a conventional configuration, FIG. 6 is a time chart showing the basic operation of the DPCM one-dimensional one-point prediction method, Figure 7 shows enlarged views b and c of quantization level values a and points b and c when a staircase wave video signal is A/D converted, and a waveform showing the bias when a transmission error occurs in the conventional configuration. Figure d, Figures 8 and 9 are time charts of points b and c when a staircase wave video signal is A/D converted to the input value x _o in the conventional configuration, and Figure 10 is the time chart of Figure 9. Fig. 11 is a block diagram showing the configuration of a DPCM device in an embodiment of the present invention, and Fig. 12 is a time chart when a transmission error occurs in the present invention. The time chart at point c when the input value x _o is shown in Figure 13.
It is a time chart showing the operation when a transmission error occurs in the figure. A... Encoding block, B... Transmission path, C... Decoding block, 1, 9... Arithmetic unit, 2... Quantizer, 3,
10...Predictor, 4...Encoder, 8...Decoder, _d1 to _d7
...Delay element.

Claims (1)

[Claims] 1. A first arithmetic unit that obtains a difference value between an input value and a first predicted value for the input value, a quantizer that receives the output of the first arithmetic unit as input and obtains a predetermined quantized value, and an input value. a first predictor that obtains a first predicted value for the quantizer; and an encoder that performs predetermined encoding using the output of the quantizer as input, and is connected to the encoding block via a transmission path. a decoder that receives the output of the encoder of the encoded block as input and decodes the output of the quantizer to obtain a decoded value;
Band compression of a video signal consisting of a decoding block consisting of a second arithmetic unit that obtains a restored value by adding a decoded value and a second predicted value, and a second predictor that obtains the second predicted value. In a DPCM device which performs A DPCM device characterized in that a biased restoration value is approximated to a normal level using a flat part of a video signal.