JP2835740B2 - Encoding device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding device that encodes a difference between a sample value and its predicted value.

[Prior Art] When digitizing image information and audio information and transmitting the same, encoding is performed to reduce the amount of transmission information.

One of the encoding methods is differential encoding (Differential Pulse).
Code Modulation (hereinafter referred to as DPCM), and the amount of information is compressed by utilizing the correlation between adjacent sample values. Specifically, the coded sample value is decoded once, a predicted value for the next sample value that has been subsampled is obtained using the decoded value, and the difference between the predicted value and the input sample value is quantized. Encoding method.

FIG. 5 is a block diagram showing a configuration example of a general encoding device based on the above principle. FIG. 6 is an explanatory diagram showing a sampling pattern by the subsampler 3 in FIG.

Reference numeral 1 denotes an input terminal to which input data X _i (for example, a digital television signal) is applied. 2 is the input data X _i
A pre-filter that attenuates the two-dimensional spatial frequency domain inside, 3
Is a subsampler for applying an offset subsample to the output signal of the prefilter 2, and 4 is a DPCM encoder for encoding the output signal of the subsampler 3. 5 is a transmission path, 6 is a DPCM decoder that decodes the coded signal from the DPCM encoder 4 and performs band expansion processing, 7 is an interpolation filter that performs interpolation on the signal expanded by the DPCM decoder, Reference _numeral 8 denotes an output terminal for outputting output data Vi.

In the above configuration, for example, an 8-bit digital television signal is applied to the input terminal 1, a predetermined two-dimensional spatial frequency region is attenuated by the pre-filter 2, and offset / sub-sampling is performed by the sub-sampler 3. This offset subsampling is
Processing is performed as in the pattern of FIG.

In FIG. 6, the solid line represents the scanning line of the first field, and the broken line represents the scanning line of the second field. still,
Here, it is assumed that the input signal is a television signal that is two-to-one interlaced. The television signal is sampled in a grid at points indicated by ○ and ×, and the subsampler 3 transmits only the sample points (pixels) indicated by ○ (where f _s is the sampling frequency). Is). In this manner, by performing well-known offset subsampling in a staggered manner and interpolating the sample points not transmitted by the interpolation filter 7 in the subsequent stage on the receiving side, the image can be reduced without lowering the horizontal and vertical resolution. Can be transmitted. However, although the resolution in the oblique direction is deteriorated, the visual characteristics of a person are not so much deteriorated because the resolution in the oblique direction is smaller than that in the horizontal and vertical directions. The characteristics of the prefilter 2 are designed to attenuate the oblique direction of the spatial frequency, thereby preventing the aliasing distortion due to the above-described sub-sample.

The signal subsampled by subsampler 3 is
For example, a signal of 8 bits per sample point (pixel) is compressed into 4 bits by the DPCM encoder 4 and transmitted to the transmission path 5. The 4-bit coded signal from the transmission line 5 is DPCM
The signal is decoded by the decoder 6 and expanded to the original 8-bit signal. Next, the transmission path 5
The sample points indicated by the crosses in FIG. 6 that have not been transmitted are interpolated based on the surrounding transmitted sample points. Interpolated output data V _i is output from the output terminal 8.

Next, details of the configuration of the DPCM encoder 4 shown in FIG. 5 will be described. FIG. 7 is a block diagram showing a configuration example of a general DPCM encoder 4. In the encoding here, the sample value encoded as described above is once decoded (local decoding).
Then, a method is used in which a predicted value for the next sample value is obtained using the decoded value, and an error between the predicted value and the actual value is quantized and encoded.

10 is an input terminal to which the sample value X _i is input, 12 is a subtractor for subtracting the sample value X _i from the predicted value P _i (predecoded value), and 14 is a difference value output from the subtractor 12. a quantizer to be sent to the output terminal 16 as the DPCM code Y _i is quantized. 18 is an inverse quantizer that combines the DPCM code Y _i into a difference value, 20 is an adder that adds the prediction value P _i and the output signal of the inverse quantizer 18 to restore the difference value to a sample value, and 22 is the amplitude of the output signal of the adder 20 is limited to a predetermined range, a limiter to obtain a local decoded value V _i.
24 is a D-type flip-flops to be applied to the subtractor 12 the locally decoded value V _i output from the limiter 22 as the predicted value P _{i + 1} of the next sample values X _{i + 1.}

Next, the operation of the DPCM encoder 4 having the configuration shown in FIG. 7 will be described.

The sample value X _i input to the input terminal 10 is subtracted by the subtractor 12 from the predicted value P _i . Difference value by the subtraction is quantized by the quantizer 14, the output terminal as the DPCM code Y _i
Output to 16. The output of the quantizer 14 is the inverse quantizer 18
, Where the DPCM code Y _i is decoded into a difference value, and then output to the adder 20. The adder 20 adds the prediction value P _i which is outputted from the D-type flip-flop 24 to the difference value.
As a result, the difference value (quantized representative value) is restored to the sample value (locally decoded value). The restored sample value is applied to a D-type flip-flop 24 after its amplitude is limited to a predetermined range by a limiter 22. The D-type flip-flop 24 outputs the local decoded value output from the limiter 22 in synchronization with the clock, and applies it to the subtractor 12 as a predicted value P _{i + 1} for the next sample value X _{i + 1} .

Generally, the distribution of difference values of predicted values is biased toward small values, and information is compressed and transmitted by encoding and transmitting the difference values by nonlinear quantization.

[Problems to be Solved by the Invention] In the conventional encoding apparatus as described above, since a previous value, that is, a value two sample points before, is used as a predicted value, when sub-sampling is performed, the The spatial distance becomes longer and the prediction accuracy decreases. For this reason, distortion due to quantization appears in the image, causing degradation such as edge business (image quality deterioration at an edge portion) and granular noise (roughness of a reproduced image in a flat portion), and it has been difficult to transmit high-quality images. .

As a method to solve this, not only sample values on the same horizontal scan line but also sample values of other scan lines are used,
A method of generating a predicted value, that is, performing a so-called two-dimensional prediction, and improving the prediction accuracy is conceivable.

However, in a device for transmitting a differentially coded code, it is necessary to pay attention to so-called error propagation. This error propagation occurs when not only the decoded value decoded using this differential code but also all the decoded values decoded using this decoded value when an error occurs in the differential code being transmitted. Due to the inability to do so.

Therefore, in the encoding device that performs the two-dimensional prediction as described above, if an error occurs in one decoded value, the error propagates to a plurality of other decoded values. Also,
In particular, when a magnetic recording / reproducing system is assumed as a transmission path, a so-called burst error often occurs in a code which is continuously transmitted (recorded / reproduced). Since codes are generally transmitted in raster order, this is an error of pixels that are continuous in the horizontal direction on the screen. In addition, even if a code error propagates in the vertical direction of the screen, even a situation in which most of the screen cannot be reproduced is a concern.

The present invention has been made in order to solve such a problem, and an object of the present invention is to provide an encoding device that can improve the prediction accuracy and minimize the influence of the propagation of a code error. It is.

[Means for Solving the Problems] To achieve the above object, an encoding apparatus according to the present invention comprises an input unit for inputting image signals in a raster scanning order, and a sampling unit for sub-sampling the image signal input by the input unit. Means, and predictive coding means for predictive coding the sample values sampled by the sampling means, wherein the sampling means is arranged such that pixels sub-sampled on adjacent horizontal lines of the same field are on a vertical line of the screen. And the sub-sampling is performed so that the distance between the sub-sampled adjacent sample values of the same field is greater in the horizontal direction than in the vertical direction of the screen, and the predictive encoding means performs the sample value and the sample value And encodes a difference value with a prediction value corresponding to the prediction value, and the prediction value is a sample value of a pixel to be encoded. Is generated using pixels aligned in the vertical direction of the screen.

Embodiment FIG. 1 is a block diagram showing a schematic configuration of an encoding apparatus as one embodiment of the present invention. FIG. 2 is a detailed block diagram of the DPCM encoder 9 shown in FIG. The first
In the figure, the same components as those in FIG. 5 are denoted by the same reference numerals, and the description thereof will not be repeated below.

As shown in FIG. 1, in this embodiment, in FIG.
In place of the DPCM encoder 4 and the DPCM decoder 6, a DPCM encoder 39 and a DPCM decoder 42 having the configuration shown in FIG.
It is characterized in that a transmission order converter 40 and an order reverse converter 41 are provided between the PCM encoder 39 and the DPCM decoder 42.

The transmission order converter 40 has a function of converting the transmission order from horizontal to vertical.
Has the function of converting the transmission order of signals from vertical to horizontal.

Here, the configuration of the DPCM encoder 39 whose configuration is shown in FIG. 2 will be described. Note that, in FIG. 2, the same reference numerals are used for the same components as those in FIG. This DPCM encoder 39 is different from the DPCM encoder 4 in FIG. 7 in that a 1H (one horizontal scanning period) delay line 26 is provided instead of the D-type flip-flop 24.

Next, the operation of the configuration shown in FIG. 2 will be described. Here, the process of generating the DPCM code Y _i, decoded using this is omitted because it is similar to the case of Figure 7.

Local decoding value V _i output from the limiter 22 is the 1H delay line 26
Is input to 1H delay line 26 is subjected to a delay of one horizontal scanning line unit to a local decoded value V _i, subtractor 12 as the predicted value P _i
Output to This predicted value _Pi is not the previous value, but is 1 on the screen.
This is the sample point above the horizontal scanning line.

In the apparatus shown in FIG. 5-FIG. 7, the distance on the image between the predicted value P _i and sample values X _i (hereinafter, referred to as predicted distance)
Takes the previous value, 2 / f _s and a was, but in the present embodiment becomes 2 / [nu as shown in Figure 6. The way of taking the sampling frequency f _s, is set to be slightly above than twice the transmission bandwidth of the television signal. For example, in the case of HDTV (high definition television) signals, because the transmission band is 20 MHz, the sampling frequency f _s may be about 44MHz～48MHz. Normally, the transmission band of a television signal is determined by multiplying a signal converted in the vertical direction by a coefficient called a so-called Kell factor. For HDTV, Kell-factor is about 0.56, the horizontal sampling point distance 1 / f _s at the time of the sampling frequency f _s to twice the transmission band for inter-scan-line distance 1 / [nu 1 / 0.56 = 1.78 times farther. Normally, the sampling frequency f _s, it takes more than 2.2 times the transmission band considered the feasibility of the previous value filter in the case of A / D converts the television signal. Therefore, the sampling frequency f _s with respect to HDTV signals in consideration of this, if you set the f _s = 48 MHz, 1 / f _s with respect to 1 / [nu is a distance of about 1.45 times. This ratio corresponds to the ratio of the predicted distance between the conventional example and the present embodiment. Since the present embodiment has a shorter predicted distance, the prediction accuracy is increased, distortion due to quantization is suppressed, and high image quality is improved. Will be possible.

However, in the above-mentioned DPCM encoder 39, prediction is performed in the vertical direction, and output is also performed in the horizontal direction (raster scanning direction).
Have gone to. By performing the prediction in the vertical direction in this way, when a transmission error occurs on the transmission path, error propagation occurs in the vertical direction of the screen. In particular, when the complete integral type is used for the local decoder with a prediction coefficient of 1, error propagation is reset by the local decoder (for example, a 1H delay line during a vertical blanking period).
This is done by resetting the contents of 26). And in the case of the above reset method,
Errors that occur on the screen propagate to the bottom of the screen.

As shown in FIG. 3, for example, when a random error a occurs, the error propagates as shown. In a system in which recording and reproduction are performed on a magnetic tape as a transmission path, a code error is highly likely to occur in a code transmitted continuously. Burst errors as shown are likely to occur. That is, this burst error b
Is a code error that continues in the horizontal direction on the screen, and each of them takes the form of propagating in the vertical direction, so that it appears as an error covering a large area, which is extremely unsightly. On the other hand, for the random error a, for example, the occurrence of the random error a is detected by a method such as parity check, and interpolation is performed from the left and right pixels of the line where the random error a is generated.
Errors can be corrected. However, the burst error b
It is impossible to reduce the deterioration of the image due to the above-mentioned correction.

Therefore, in this embodiment, as shown in FIG. 1, a transmission order converter 10 and a transmission order inverse converter 11 are provided to reduce the adverse effect on the image of the error propagation according to the burst error occurring on the transmission path, and Is suppressed. By taking measures against the error propagation and measures to improve the prediction accuracy by the DPCM encoder 39, it is possible to transmit an image with extremely high image quality.

The DPCM code Y _i output from the DPCM encoder 39 is input to the transmission order converter 40. The transmission order converter 40 has a memory having a capacity of one screen or more, and receives the input DPCM code.
Y _i is temporarily stored for one screen. Thereafter, the data is read out in the vertical direction of the screen and transmitted to the transmission path 5. With this operation, DP
Transmission order of CM code Y _i is converted from the horizontal to the vertical.

The horizontal / vertical converted DPCM code Y _i is supplied to the transmission order reverse converter 41 via the transmission path 5. In the transmission order reverse converter 41, the transmission order of the transmitted signal is converted from vertical to horizontal and input to the DPCM decoder 42. The output signal of the transmission order inverter 41 is expanded to 8 bits by the DPCM decoder 42, and the interpolation filter 7 interpolates the sample points not transmitted, restores the original signal, and sends it to the output terminal 8. .

Incidentally, the input side of the second view of the DPCM coder inverse quantizer 18 and input to the configuration of the DPCM decoder 42, the output of the limiter 22 and the output V _i. In other words, the DPCM encoder shown in FIG.

FIG. 4 is an explanatory diagram showing the state of an error on the screen in the present embodiment when there is a transmission error. As apparent from FIG. 4, according to the present invention, the error propagation accompanying the random error a is the same as the conventional one, but the burst error b is continuously arranged in the transmission direction, that is, in the vertical direction of the screen. Become like For this reason, only one error is propagated in the vertical direction, and interpolation can be performed from the left and right pixels as in the case of the random error a, so that image quality degradation can be reduced.

In the above embodiment, since the video signal is subjected to field offset sub-sampling and the sample points of each field are arranged in a grid pattern, pixels adjacent in the vertical direction of the screen are used to generate predicted values. When the signal is subjected to line offset subsampling, the pixels closest to the screen are arranged obliquely. In this case, the prediction accuracy can be improved by using a sample value corresponding to a pixel obliquely above the pixel to be encoded in generating the prediction value, and the code transmission order is also oblique in the screen. By performing the conversion, similarly, it is possible to minimize the adverse effect on the reproduced image due to the propagation of the error accompanying the burst error.

[Effects of the Invention] As is apparent from the above description, according to the present invention, pixels sub-sampled on adjacent horizontal lines of the same field are arranged on a vertical line of the screen, and sub-sampled in the same field. The input image signal is sub-sampled so that the distance between adjacent sample values in the horizontal direction is greater than in the vertical direction of the screen, and the sub-sampled sample value is compared with the predicted value corresponding to the sample value. Since the prediction value is encoded using a pixel that is aligned in the vertical direction of the screen with respect to the sample value of the pixel to be encoded, the prediction error for the sub-sampled image signal is extremely small. It is possible to achieve highly efficient encoding while suppressing image quality deterioration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a main configuration of an embodiment of an encoding apparatus according to the present invention, FIG. 2 is a block diagram showing a detailed configuration of a DPCM encoder in FIG. 1, and FIG. FIG. 4 is a diagram for explaining the state of error propagation in the case where transmission is not performed, FIG. 4 is a diagram for explaining the state of error propagation in the case where transmission order conversion is performed, and FIG. FIG. 6 is an explanatory diagram showing a sampling pattern by the subsampler of FIG. 5, and FIG. 7 is a DP shown in FIG.
FIG. 3 is a block diagram illustrating a configuration of a CM encoder. In the figure. 3: Subsampler 5: Transmission line 12: Subtractor 14: Quantizer 18: Dequantizer 20: Adder 22: Limiter 26: 1H delay line 39: DPCM encoder 40: Transmission order converter 41: Transmission order Inverter 42: DPCM decoder P _i : Predicted value V _i : Local decoded value X _i : Sample value

Claims (1)

(57) [Claims] An input unit for inputting image signals in a raster scanning order; a sampling unit for sub-sampling the image signal input by the input unit; and a predictive code for predictively encoding a sample value sampled by the sampling unit. The sampling means is arranged such that pixels sub-sampled on adjacent horizontal lines of the same field are arranged on a vertical line of the screen, and a distance between sub-sampled adjacent sample values of the same field is displayed on the screen. Sub-sampling so that the horizontal direction is larger than the vertical direction, the prediction encoding means encodes a difference value between the sample value and a prediction value corresponding to the sample value, and the prediction value is encoded. Make sure that the sample value of the pixel you are trying to generate is generated using pixels aligned in the vertical direction of the screen. An encoding device characterized by the following.