JP3401273B2 - Encoding device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding apparatus suitable for application to a codec or the like for coding image data and compressing it, or decoding compressed image data to obtain original image data.

[0002]

2. Description of the Related Art Conventionally, a device called a codec has been proposed. This codec is for encoding and compressing when transmitting or recording image data,
The encoding of this image data was completed in December 1990 by the International Telegraph and Telephone Consultative Committee (CCITT), which is under the umbrella of the International Telecommunication Union (ITU).
C (coder, decoder) Recommendation 261 standardized.

The moving picture coding is applied to a standard television or a high definition (H) as a signal source.
D) It is used in fields such as broadcasting and communication as applications involving signal transmission to a remote place using a television, and storage as applications for local signal processing.

This Recommendation H. As a video format according to H.261, a common intermediate format (CIF: Common Int) that can resolve differences in television systems depending on regions (the whole world) and can communicate between CODECs
ermediate Format).

The resolution of an image based on this CIF is 35 horizontal pixels.
2, vertical 288 dots.

Generally, a coding unit of a video codec encodes input video data with an encoder, multiplex-encodes the data, stores the data in a transmission buffer once, and then encodes it with a transmission encoder. The data is transmitted as an encoded bit string, and the decoding unit decodes the transmitted encoded bit string video data with a transmission decoder, temporarily stores it in the reception buffer, multiplex-decodes it, and further decodes this data. Decode to get the original video signal.

When transmitting a large amount of image data as described above, the image data is encoded and compressed at the time of transmission, and the image data that is encoded and compressed at the time of reception is decoded.

Therefore, the video codec can be used not only for image transmission but also for recording image data in a VTR, for example.

Especially, the image data of the high definition television system, which has been rapidly advanced in recent years, has an enormous amount of data, unlike that of the standard television system. Therefore, it is naturally encoded and compressed at the time of recording and reproduced at the time of reproduction. Decoding to obtain the original image data is an indispensable subject in order to significantly reduce the recording cost.

As one of the processing of television signals in this codec, there is known a method of reducing the sampling frequency for the purpose of narrowing the transmission band, that is, a method of thinning out pixels by sub-sampling to lower the sampling frequency. Has been.

As one of the sub-sampling, it is proposed that the image data is thinned to 1/2 and the sub-sampling point and a 2-bit flag indicating the position of the sub-sampling point used at the time of interpolation are transmitted. ing. When 1 pixel data of the digital video signal (image data) is 8 bits, adding 2 bits of the flag results in 5 bits per pixel and a compression rate of 5/8.

This sub-sampling has the problem that since the sub-sampling pattern is always the same, the image quality of the restored image is conspicuously deteriorated in a portion such as the contour of the object in the image. In particular, set the subsampling rate to 1
If it is higher than / 2, there is a drawback that the image quality is significantly deteriorated.

Therefore, in order to solve such a problem, the present applicant divides one image into a large number of two-dimensional blocks, and determines the difference between the maximum value and the minimum value of a plurality of image data in this block. That is, an encoding method has been proposed in which the dynamic range is obtained and the sub-sampling cycle is changed according to the dynamic range of the block (Japanese Patent Laid-Open No. 62-2).
(See Japanese Patent No. 66924).

That is, a block having a small dynamic range is determined to be a flat image and the sub-sampling cycle is lengthened to, for example, 1/8, and a block having a relatively large dynamic range is changed. Judge that it is an image and set the subsampling cycle to 1 /
2, it is determined that a block with a very large dynamic range is an image with a large change, and sub-sampling is not performed.

However, in the high-efficiency coding apparatus that selectively switches the sub-sampling cycle according to the dynamic range in this way, the sub-sampling cycle is set in block units, and thus the restored image is in block units. However, there was a defect that the distortion of the block was noticeable due to the quality of the image.

Further, there is a limit to the types that can be selected as the sub-sampling period, and the adaptability to the image characteristics is insufficient.

Therefore, the Applicant further confirms that one basic pixel for each pixel (for example, 4 samples × 4 lines) of the digital video signal is always transmitted without being thinned out, and pixels other than the basic pixel are thinned out by subsampling. , The transmission is carried out as it is, and this judgment is made according to the magnitude of the predicted error when the pixels thinned out on the receiving side are interpolated by the average value of two pixels above and below or left and right, and a prediction error is made. If the prediction error is larger than the threshold value, it cannot be decimated, so the original data is transmitted, and if the prediction error is smaller than the threshold value, decimation is possible. , Does not transmit the raw data, replaces the raw data with the mean value, while subsampling
A high-efficiency coding device is proposed in which the density of coarse sub-sampling is changed to the density of fine sub-sampling in order, and thereby data of pixels whose transmission / thinning is controlled and data of basic pixels are transmitted. There is.

In this device, 1 bit is added to each sample of pixel data other than the basic pixel in order to control transmission / thinning. On the receiving side,
The added bit is detected to determine whether to use the received original data or the average value.

[0019]

When fixed pattern sub-sampling and block coding such as ADRC are combined, the number of pixels in a block is uniquely determined, but the sub-sampling that adaptively changes the density is performed. Since the number of pixels in a block is unknown, for example, if fixed block coding is performed as it is, coding is performed and even redundant information such as additional bits is transmitted even if there are few transmission pixels. There was an inconvenience that it would end up.

When variable density sub-sampling is performed first and then block coding is performed, the interpolation error at the time when the variable density sub-sampling is performed and the interpolation error after coding are at a considerable level. Since the difference between the original image data and the decoded image data is significantly different, there is the inconvenience that the image level fluctuation during encoding adversely affects the image quality after interpolation. It was

The present invention has been made in view of the above points, and improves compression efficiency when compressing pixel data by combining sub-sampling and encoding processing, and restores the compressed pixel data when restored again. The present invention is intended to propose an encoding device capable of improving the image quality of the image.

[0022]

SUMMARY OF THE INVENTION The present invention relates to an encoding device for encoding input pixel data into transmission data having a smaller amount of data, and a plurality of pixel data having a temporal or spatial arrangement. and subsampling means for variably density subsampling, the output of the sub-sampling means, encoding means for encoding in response to the dynamic range, or the output of the sub-sampling means
Or the output of the encoding means is transmitted within a predetermined block
And a selecting means for selecting and outputting according to the amount of information to be output.

[0023]

According to the present invention, in the above-mentioned encoding device of the present invention, the selecting means transmits within a predetermined block.
The encoding device selects the output of the sub-sampling means when the amount of information to be transmitted is small, and selects the output of the encoding means when the amount of information to be transmitted within a predetermined block is not small.

[0025]

[0026]

[0027]

[0028]

[0029]

According to the present invention, the sub-sampling means performs variable density sub-sampling of a plurality of pixel data having a temporal or spatial arrangement, and the encoding means outputs the output of the sub-sampling means to its dynamic range. coded in accordance with, by the selecting means, the output of the output or the coding means of the sub-sampling means, where
It is selected and output according to the amount of information to be transmitted in the fixed block .

[0030]

[0031] In addition, according to the present invention, the selection means, place
When the amount of information transmitted in a fixed block is small,
Select the output of the sub-sampling means, and within the predetermined block
If the amount of information to be transmitted is not small, the output of the encoding means is selected.

[0032]

[0033]

[0034]

[0035]

[0036]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of the high efficiency coding apparatus of the present invention will be described in detail below with reference to FIG.

In FIG. 1, reference numeral 1 denotes an input terminal to which image data (digital television signal) in which, for example, one sample is quantized into 8 bits from a recording system of a VTR (not shown) is supplied. The image data from 1 is supplied to a subsampling circuit (variable density subsampling circuit) 2.

As described above, the sub-sampling circuit 2 transmits the basic pixels without thinning out, and the pixels other than the basic pixels are compared with the average value of the two pixels, and the original pixel data and the average value are compared. If the difference (prediction error) exceeds the threshold value, the original pixel data is transmitted. If the prediction error is less than the threshold value, the original pixel data is thinned out, and the original pixel data stored in the memory (not shown) is replaced with the average value. To do.

The blocking circuit 3 converts the input image data into a continuous signal for each two-dimensional block which is a unit of coding. In this example, the size of one block is, for example, (4 lines × 4 pixels = 16 pixels). The image data (in pixels) processed by the blocking circuit 3 is supplied to the dynamic range (DR) detection circuit 4 and the addition circuit 5, respectively.

The dynamic range detecting circuit 4 detects the dynamic range and the minimum value of the pixel data from the blocking circuit 3 for each block, and supplies the minimum value data to the adding circuit 5 and the framing circuit 9, respectively. The dynamic range is supplied to the bit length determination circuit 6 and the framing circuit 9, respectively.

The adding circuit 5 subtracts the minimum value data from the dynamic range detecting circuit 4 from the pixel data from the blocking circuit 3 and supplies the subtraction result to the quantizing circuit 7.

On the other hand, the bit length determination circuit 6 determines the number of quantization bits (bit length) corresponding to the dynamic range. In this case, the bit length is determined in consideration of human visual characteristics. That is, the maximum distortion is increased when the dynamic range is large.

As an example, the bit length determination circuit 6 determines the bit length according to the dynamic range as follows. That is, when the dynamic range is 0 or more and 10 or less, the bit length is "0" and the maximum distortion is "5", and when the dynamic range is 11 or more and 25 or less, the bit length is "1" and the maximum distortion is "6". When the dynamic range is 26 or more and 99 or less, the bit length is “2”, the maximum distortion is “12”, and the dynamic range is 100 or more 2
When it is 55 or less, the bit length is "3" and the maximum distortion is "1".
6 ".

The bit length data thus determined is supplied to the quantizing circuit 7. The quantizing circuit 7 performs a quantizing process on the addition result from the adding circuit 5, that is, the pixel data after the minimum value removal, based on the bit length data from the bit length determining circuit 6, and performs the quantizing process. The data obtained by applying the above, that is, the encoded code is supplied to the fixed contact 8a of the switch 8.

Here, in FIG. 1, the blocking circuit 3, the dynamic range detection circuit 4, the addition circuit 5, the bit length determination circuit 6 and the quantization circuit 7 constitute, for example, an ADRC (adaptive dynamic range coding) circuit. It shall be.

The switch 8 has its movable contact 8c selectively connected to the fixed contact 8a or 8b, so that the coded code from the quantization circuit 7 or the sub-sampling circuit 2 can be obtained.
To selectively supply the framing circuit 9 with the output from the.

By the way, regarding the switching control of the switch 8, a block in which the amount of information of pixels to be transmitted in the block is smaller than the additional information necessary for block coding, that is, the minimum value data and the dynamic range and the amount of code. Only in the case, the control is performed so that the pixel is directly transmitted without performing the block coding (ADRC in this example).

For ADRC, for example, the minimum value and the dynamic range of 8 bits must be transmitted in each block. Variable density subsampling and AD
When RC is combined, there are always cases where there are as few transmission pixels as 0, 1, 2, etc. in the ADRC block. At this time, if the ADRC encoding is performed, the amount of information is rather increased.

Therefore, at the time of encoding, the number of transmission pixels is detected, and whether or not the ADRC process is performed is determined based on the detection result. That is, in the case of the configuration shown in FIG. 1, for example, between the sub-sampling circuit 2, the blocking circuit 3, the dynamic range detecting circuit 4, the bit length detecting circuit 6, the quantizing circuit 7 or the dynamic range detecting circuit and the framing circuit 9. The control circuit provided in the above, or a control unit such as a VTR main circuit (not shown) detects the number of transmission pixels and controls the switch 8 based on the detection result. When the number of transmission pixels is small, the movable contact 8c is opened. The movable contact 8c is connected to the fixed contact 8b, and when the number of transmission pixels is not small, the movable contact 8c is connected to the fixed contact 8a.

Now, the framing circuit 5 uses the dynamic range (for example, 8) from the dynamic range detection circuit 4.
Bit) and the minimum value data (for example, 8 bits 9), and the block data (encoding code) or the sub-sampling data from the quantizing circuit 7 or the sub-sampling circuit 2 via the switch 8 is subjected to error correction encoding processing, respectively. Along with this, a sync signal is added to obtain recording data or transmission data.
Supply to the recording system, etc.

The VTR recording system records the data from the framing circuit 9 on the magnetic tape so as to form inclined tracks.

The data thus recorded is reproduced by a reproducing system such as a VTR (not shown), and thereafter, in a frame decomposing circuit, dynamic range and minimum value data, and block data (encoding code) or sub-sampling data. And the error correction processing is performed on these data, and the data subjected to the error correction processing is decoded by the decoding circuit.

Since the decoded data is either ADRC processed after subsampling or only subsampling is performed,
Make that judgment, and based on the result, perform appropriate processing,
Image data (digital television signal) in which one sample is quantized into 8 bits is obtained, and this is supplied to an unillustrated reproduction system of, for example, a VTR via an output terminal 12.

By the way, on the decoding side, it is simply determined how many pixels are present in the block whether the encoded data is ADRC processed after sub-sampling or only sub-sampling is performed. It can be easily judged by simply detecting.

As described above, in this example, the sub-sampling process is performed on the image data, and then the ADR is performed.
The coding by C is performed, the number of pixels in the block is detected, and when the number of pixels in the block is small, the switch 8
The sub-sampling process is performed to output the data that has not been encoded by the ADRC. When the number of pixels in the block is large, the switch 8 performs the sub-sampling process, and then the encoding by the ADRC is performed. Since the data is output, the amount of transmitted information can be reduced and the image quality can be improved.

Next, referring to FIG. 2, a reference example of the high efficiency coding apparatus of this example will be described. In FIG. 2, FIG.
The same reference numerals are given to the portions corresponding to, and detailed description thereof will be omitted.

In FIG. 2, unlike FIG. 1, the input terminal 1 is connected to the fixed contact 12b of the switch 12, and the output end of the quantization circuit 7 is connected to the fixed contact 12a of the switch 12 and the input end of the decoder 11. The output terminal of the decoder 11 is connected to the input terminal of the sub-sampling circuit 2, and the movable contact 12c of the switch 12 is connected to the sub-sampling circuit 2.
, And the output end of the sub-sampling circuit 2 is connected to the input end of the framing circuit 9.

That is, in the high-efficiency coding apparatus shown in FIG. 2, sub-sampling is performed after processing by ADRC. In the switch 12, data from the input terminal 1 or the quantizing circuit 7 is used. ADR
The data encoded by C is selectively supplied to the sub-sampling circuit 2.

Regarding the switching control of the switch 12, a screen in which the amount of information of pixels to be transmitted in a block or one screen is smaller than the additional information necessary for encoding, that is, the minimum value data and the dynamic range and the amount of code. Only in this case, the pixel is controlled to be transmitted as it is without performing the encoding (ADRC in this example).

In ADRC, for example, the minimum value, the dynamic range of 8 bits each, or a predetermined number of bits must be transmitted in each block or screen. When variable density subsampling and ADRC are combined, there are no transmission pixels in the ADRC block or screen.
There are always cases where there are as few as one, two, and so on.
At this time, if the ADRC encoding is performed, the amount of information is rather increased.

Therefore, at the time of encoding, the number of transmitted pixels is detected.
Whether or not to perform ADRC processing based on the detection result
Make a decision. That is, in the case of the configuration shown in FIG.
In, for example, subsampling circuit 2, dynamic
Range detection circuit 4, bit length detection circuit 6, quantization times
Path 7 or dynamic range detection circuit and framing
A control circuit provided between the circuits 9, or an example not shown
For example, the control unit such as the VTR main circuit detects the number of transmission pixels,
The switch 12 is controlled based on the detection result, and the transmission image is
When the number of primes is small, the movable contact 12c is fixed to the fixed contact 12b.
The movable contact 12c is connected to the fixed contact 12a when the number of transmission pixels is not small.

Further, the data encoded by ADRC is decoded by the decoder 11, and whether or not to perform subsampling in the subsampling circuit 2 is determined based on the data obtained by the decoding. In this example, the ADRC process is performed first in order to reduce the influence of a quantization error in encoding during decoding.

Next, the operation will be described. For example, image data for one screen supplied from a VTR main circuit (not shown) through the input terminal 1 is the dynamic range detection circuit 4
The minimum value and the dynamic range are detected by this processing, the minimum value data is supplied to the framing circuit 9 and the adding circuit 5, respectively, and the dynamic range is supplied to the framing circuit 9 and the bit length determining circuit 6, respectively. To be done.

In the adding circuit 5, the minimum value data detected by the dynamic range detecting circuit 4 is subtracted from the image data for one screen supplied through the input terminal 1,
The result is supplied to the quantization circuit 7.

On the other hand, in the bit length determination circuit 6, as already described with reference to FIG. 1, the quantization bit is determined corresponding to the dynamic range from the dynamic range detection circuit 4, and the determined quantization is performed. Bit data Data is supplied to the quantization circuit 7.

The quantizing circuit quantizes the output from the adding circuit 5 on the basis of the quantized bit data from the bit length determining circuit 6, and quantizes the quantized data, that is, the ADRC encoded code, by the decoder 11 and the switch 12. Are supplied to the fixed contacts 12a.

The decoder 11 decodes the encoded data from the quantization circuit 7 and controls the subsampling circuit 2 to turn on or off the subsampling process based on the value obtained by the decoding.

On the other hand, the fixed contact 12b of the switch 12 is supplied with data from the VTR main circuit (not shown) or the like via the input terminal 1, and the fixed contact 12a is encoded by the ADRC from the quantizer 7. Is supplied.

At this time, as described above, the switch 12
Is the control of the circuit shown in FIG.
Section or the dynamic range detection circuit 4 and the frame
For the control of the control circuit that can be provided between the conversion circuits 9.
Therefore, when the number of pixels of blocks or one screen is small to connect the movable contact 12c to the fixed contact 12b, which following
In the case of the outside, the movable contact 12c is connected to the fixed contact 12a.

Therefore, when the number of pixels to be transmitted is small, the data supplied through the input terminal 1 is supplied as it is to the sub-sampling circuit 2 through the switch 12, and in this sub-sampling circuit 2, sub-sampling, that is, decimation processing is performed. And then supplied to the framing circuit 9.

On the other hand, in other cases, the coded code obtained by the ADRC process from the quantization circuit 7 is supplied to the sub-sampling circuit 2 through the switch 12. At this time, the decoder 11 makes this coded code. Is decoded.

Then, the subsampling circuit 2 determines whether or not to perform subsampling based on the decode value (local decode value) from the decoder 11.

After that, the data from the sub-sampling circuit 2 is supplied to the framing circuit 9. In the framing circuit 9, error correction code and synchronization are added, and the recording system of the VTR (not shown) is output via the output terminal 10. And is recorded by this recording system so as to form inclined tracks on a magnetic tape, for example.

When the data recorded in this way is reproduced, if the data reproduced by the reproducing system of the VTR (not shown) is sub-sampled after being encoded by ADRC, ADRC decoding is performed and further interpolation is performed. Thus, only interpolation is performed in the case of data for which only sub-sampling has been performed.

As described above, the ADRC encoding process is performed on one screen, the data encoded by the ADRC is once decoded to obtain the local decode value, and the local decode value is used to determine whether or not to subsample the decoded value. Therefore, when the encoded data is reproduced by the reproducing system again, and when it is received by the receiving side, the deviation of the interpolation value can be reduced, and thereby the original data can be accurately reproduced.

In the above example, the VTR has been described as an example, but the same effect can be obtained even when applied to a data transmission system such as a television conference system.

In the above example, the case of performing ADRC has been described, but the same applies to the case of performing discrete cosine transform (DCT) or the like.

The above-described embodiment is an example of the present invention, and it goes without saying that various other configurations can be adopted without departing from the gist of the present invention.

[0079]

According to the invention described above, in a coding apparatus for coding input pixel data into transmission data having a smaller amount of data, a plurality of pixel data having a temporal or spatial arrangement are The sub-sampling means for performing variable density sub-sampling, the encoding means for encoding the output of the sub-sampling means according to the dynamic range, and the output of the sub-sampling means or the output of the encoding means are predetermined. Since it has a selecting means for selecting and outputting according to the amount of information transmitted in the block , the configuration is simplified and the compression efficiency when compressing pixel data by combining subsampling and encoding processing. The quality of the restored image can be improved when the compressed pixel data is restored again. It is possible to obtain an encoding device that can improve the transmission efficiency.

[0080]

[0081] According to the present invention, in the coding apparatus of the above-described invention, the selection means, the predetermined block smell
When the amount of information to be transmitted is small, the output of the sub-sampling means is selected, and when the amount of information to be transmitted in a predetermined block is not small, the output of the encoding means is selected. It is possible to obtain an encoding device that achieves the same effects as the above-mentioned invention.

[0082]

[0083]

[0084]

[0085]

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a high-efficiency encoding device of the present invention.

FIG. 2 is a configuration diagram showing a reference example of a high efficiency encoding device of the present invention.

[Explanation of symbols]

2 Sub-sampling circuit 3 block circuit 4 Dynamic range detection circuit 5 adder circuit 6-bit length decision circuit 7 Quantization circuit 8 and 12 switches 9 framing circuit 11 decoder

Claims (2)

(57) [Claims] 1. An encoding device for encoding input pixel data into transmission data, which is data having a smaller data amount, in which sub-sampling is performed for variable density sub-sampling of a plurality of pixel data having a temporal or spatial arrangement. Means, an encoding means for encoding the output of the sub-sampling means in accordance with its dynamic range, and an output of the sub-sampling means or the output of the encoding means according to the amount of information to be transmitted in a predetermined block. And a selecting means for selecting and outputting the selected output. 2. The encoding device according to claim 1, wherein the selecting means selects the output of the sub-sampling means and transmits the output in the predetermined block when the amount of information to be transmitted in the predetermined block is small. An encoding device, characterized in that the output of the encoding means is selected when the amount of information is not small.