JPH0793583B2 - Predictive coding device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a predictive coding apparatus for predictively coding and transmitting an information signal.

[Prior Art] In a conventionally known predictive coding method, for example, a predictive prediction D-PCM (Differential-Pulse Code Modulatio)
This is an encoding method called n), which will be described below. FIG. 2 shows a block diagram of an apparatus for realizing this system.

As shown in FIG. 2, an information signal such as an image signal input from the input terminal 1 is converted into an X-bit digital signal by the A / D converter 2, and the subtractor 3 decodes the decoded signal one sample before, which will be described later. Of the difference value, and the signal of the difference value is compressed by being converted into xbit (X> x) by the non-linear quantizer 4, and further is subjected to variable length coding by the code converter 8 and output from the output terminal 9. To be done.

The output signal of the non-linear quantizer 4 is also input to the representative value setting device 5, and the signal of the representative value selected according to the input signal by the representative value setting device 5 is added by the adder 6 to the decoded signal. , Form a new decoded signal.

The decoded signal from the adder 6 is delayed by one sample by the one-sample delay unit 7 and input to the subtractor 3 as a predicted value of the next sample point, and also to the adder 6 again.

[Problems to be Solved by the Invention] However, the predictive coding apparatus as described above has the following problems. That is, the non-linear quantizer 4 in FIG. 2 has the quantizing characteristics as shown in FIG. 3, and in general, in non-linear quantizing, the quantizing step becomes coarser as the input level becomes larger, and the representative value is accordingly increased. The level difference between them also becomes large. For example, when the prediction error is a portion such as an edge portion of an image signal where the luminance change is sharp and the prediction error is near A in FIG. 3, the nonlinear quantizer 4
Is a representative value with a slight change in the quantization error between frames, and takes C in FIG. 3 or C as a representative value, and the position of the edge fluctuates from frame to frame, which is referred to as edge business. Occurs, resulting in an image that is very difficult to see visually.

The present invention has been made in view of the above problems, and it is an object of the present invention to provide a predictive coding apparatus capable of suppressing signal deterioration even in a portion where an input signal level changes abruptly and performing efficient quantization. To aim.

[Means for Solving Problems] A predictive coding apparatus according to the present invention comprises a quantizing means for quantizing a difference value, a polarity of a quantization error by the quantizing means, and the difference in an edge portion immediately before an input value. A determination unit that determines whether or not the polarities of the values match, and the polarity of the quantization error and the polarity of the difference value at the edge portion immediately before the input value match according to the output of the determination unit. And correction means for correcting the quantized signal obtained by the quantization means.

[Operation] With the configuration described above, by controlling the generation polarity of the quantization error in the portion where the input signal changes abruptly according to a predetermined rule, it is possible to suppress the deterioration of the signal in this portion.

[Examples] Hereinafter, the present invention will be described with reference to Examples.

FIG. 1 is a block diagram showing an embodiment of a prior value prediction D-PCM encoder to which the present invention is applied. The same parts as those in FIG. 2 have the same coding.

The analog image signal input from the input terminal 1 is converted into a digital signal by the A / D converter 2, and the A / D converter 2
The output signal of is input to the + input terminal of the subtractor 3 and at the same time to the-input terminal of the subtracter 13. A decoded signal for edge correction determination, which will be described later, is further input to the + input terminal of the subtracter 13, and a signal indicating the difference between the two is input to the correction circuit 14 (input 2).

In the subtractor 3, the difference between the output signal (+ input) of the A / D converter 2 and the signal (-input) from the 1-sample delay unit 7 is formed. The output signal of the subtractor 3 is input to the non-linear quantizer 4 and simultaneously to the correction circuit 14 (input 1).

The output signal of the correction circuit 14 becomes a quantization level correction signal and is input to the adder 15, where the nonlinear quantization level in the output signal of the nonlinear quantizer 4 is corrected.

The output signal of the nonlinear quantizer 4 is a representative value setting circuit ▲ Q ^-1 ₂
▼ Input to 10 and add representative value signal obtained there to adder 11
By adding to the output signal of the delay circuit 12 which is the previous decoded signal at, a decoded signal for edge correction determination is obtained,
The decoded signal for this edge correction determination is a 1-sample delay unit 12
And to the + input terminal of the subtractor 13. Therefore,
In the subtractor 13, the difference signal between the decoded signal of the signal quantized by the quantizer 4 and the original signal output from the A / D converter 2 is quantized as indicated by a in FIG. 6 (A). It is formed as a signal indicating the amplitude of the error and is input to the input 2 of the correction circuit 14.

Next, FIG. 5 shows the configuration of the correction circuit 14, which is a feature of the present invention. As shown in FIG. 5, the output signal of the subtractor 3 is an input terminal.
It is input from 20 to the comparator 22 and the code detector 23. The comparator 22 compares the absolute value of the signal from the input terminal 20 with a predetermined value TH1 to output the following determination signal for the input image signal input to the input terminal 1 in FIG. Switch 24 is switched to a terminal or b terminal.

| Input 1 | ≧ TH1 (edge) a terminal | input 1 | <TH1 (other than edge) b terminal That is, when the comparator 22 determines that the input image signal is an edge, the switch 24 is connected to the a terminal and the code detector twenty three
Sign signal indicating positive or negative of the differential input signal detected in (positive =
0, negative = 1) is output to the exclusive OR 30. Here, positive indicates an up edge and negative indicates a down edge.

If it is determined that the input image signal is not an edge, the previous code signal obtained by delaying with the delay device 25 is input to the exclusive OR circuit 30.

Here, when the input image signal is not an edge, the previous code signal is input to the exclusive OR circuit 30 as it is
As shown in FIG. 4B, in the edge portion of the input signal shown by the solid line, the quantization error direction of the broken line is always brought to the edge direction by the processing described later, and the solid line as shown in FIG. This is to eliminate the double contour as shown by the broken line that occurs when the edge part of the input signal is quantized. That is, the output signal of the switch 24 is a signal indicating the direction (polarity) of the edge, and when the signal from the adder 3 is the predetermined value TH1 or more, that is, when the input image signal is determined to be the up edge or the down edge. Only updated.

Next, the output signal (input 2) of the subtractor 13 is input to the comparator 28 and the code detector 29 via the input terminal 26. Comparator 28
The signal from the input terminal 26 input to is shown in FIG.
It is compared with a predetermined value TH2 as shown in (D), and three different signals of 0 and ± 1 are gated according to the following conditions.
Enter 31.

Input 2> TH2 + 1 −TH2 <Input 2 ≦ TH20 0 Input 2 ≦ −TH2-1 Also, the sign detector 29 outputs a signal indicating the quantization error, that is, positive or negative, that is, a signal indicating the direction (positive = 0, negative = 1). It is generated and input to the exclusive OR circuit 30. Exclusive OR circuit
The output signal of 30 is input to the gate circuit 31 as a control signal.

Therefore, the operation of the correction circuit 14 is summarized in Table 1.

That is, the input 1 of the exclusive OR circuit 30, that is, the signal indicating the direction of the edge immediately before the input image signal (the polarity of the difference value) and the input 2, that is, the nonlinear quantum as shown in a of FIG. 6 (A). When it is determined that the signals indicating the direction (polarity) of the quantization error at the output of the digitizer 4 match, the gate circuit
Gate the output of 31. As a result, the output signal of the non-linear quantizer 4 is directly input to the code converter 8 without being added by the adder 15 for correction. Therefore, as shown in FIGS. 6 (A) and 6 (B), a quantized signal whose edges are emphasized can be obtained.

When the input 1 and the input 2 of the exclusive OR circuit 30 do not match, the gate circuit 31 passes the output signal (0 or ± 1) of the comparison circuit 28. As a result, the output signal of the comparison circuit 28 is input to the adder 15. That is, FIG. 6 (C)
Input image signal is input at the rising edge as shown in 2> TH
When 2 is satisfied, the direction of the quantization error is −, so the +1 signal from the comparison circuit 28 is added to the quantized signal from the non-linear quantizer 4 in the adder 15, and the quantized signal Increase the quantization step by one step. Similarly, when the input signal satisfies the input 2 ≦ −TH2 at the down edge as shown in FIG. 6 (D), the −1 signal is added to the quantized signal, and the quantization step is reduced by one step. To do.

Then, at the minus input terminal of the subtracter 3, the representative value setting unit 5 and the adder 6 are added on the basis of the (corrected) signal that has passed through the adder 15.
The decoded signal obtained by the 1-sample delay unit 7 is input.

In this way, the polarity of the non-linear quantization error can be stabilized, thus reducing edge business,
It is possible to form a digital image signal capable of obtaining an image with a clear contour, perform variable-length coding on the code converter 8 and output the coded signal from the output terminal 9.

[Effects of the Invention] As described above, according to the present invention, there is provided a predictive coding apparatus capable of suppressing signal deterioration even in a portion where an input signal level changes abruptly and performing efficient quantization. can do.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is a block diagram of a prior art predictive D-PCM coding apparatus, and FIG. 3 shows a quantization characteristic of a quantizer of a non-linear quantizer. 4A and 4B are diagrams showing the quantization error at the edge of the image signal, FIG. 5 is a block diagram of the correction circuit in FIG. 1, and FIG. 6 is A and B. , (C) and (D) are diagrams respectively showing edge portions of the quantized signal. 10 …… Representative value setter, 11,15 …… Adder, 12 …… Delayer, 13 …… Subtractor, 14 …… Correction circuit.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical indication location H04N 7/32 (56) References JP-A-56-66929 (JP, A) JP-B-62- 26633 (JP, B2) Japanese Patent Publication Sho 55-38874 (JP, B2)

Claims (1)

[Claims] 1. A predictive coding apparatus for coding a difference value between an input value and a predicted value, the quantizing means quantizing the difference value, and the polarity of a quantization error by the quantizing means, Determination means for determining whether or not the polarities of the difference values at the edge portion immediately before the input value match, and the polarity of the quantization error and the edge portion immediately before the determination portion according to the output of the determination means. A predictive coding apparatus comprising: a correction unit that corrects the quantized signal obtained by the quantization unit so that the polarities of the difference values match.