JP2860970B2 - Image coding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a discrete cosine transform (Di
The present invention relates to an image encoding device having a screte cosine translation (hereinafter abbreviated as DCT) encoder, and more particularly to an image encoding device for compressing an image signal with high efficiency.

[0002]

2. Description of the Related Art Conventionally, as a means for compressing image information when encoding an image signal, there is a method of reducing the resolution.

[0003] For example, when transmitting a certain image, information of all pixels is not encoded, but one by one in the vertical and horizontal directions.
If encoding is performed by thinning out every pixel, the information amount is reduced to 1/4. This lowers the resolution in the spatial direction.

However, in some cases, the number of coded pixels is determined in advance, and it is not possible to thin out the pixels. In such a case, a method of performing compression by reducing the resolution in the amplitude direction or the time direction has been conventionally used. Used.

Here, as means for reducing the resolution in the amplitude direction, there is control of a quantizer. For example, if the value of a pixel expressed by 8 bits is quantized by 6 bits, the information amount becomes 3/4. Also, reducing the resolution in the time direction means
In other words, it means "frame drop", and the information amount can be compressed also by this.

As a conventional image encoding apparatus, for example,
There is one as shown in FIG. In FIG. 8, reference numeral 1 denotes a DCT transform encoder (DCT encoder) for encoding input image information using DCT transform.
It is configured as a predictive encoder that encodes a prediction error, and as shown in FIG. 8, a DCT transform unit 5, a quantizer 6, an inverse quantizer 7, an inverse DCT transform unit 8, a frame memory 9, It comprises a low-pass filter 10 and a motion compensator 11.

The difference between the digital image signal applied to the encoding unit 1 and the output of the frame memory 9 is obtained to obtain an inter-frame difference value (prediction error). After the DCT transformation is performed, quantization is performed by the quantizer 6 in a predetermined quantization step. The output of the quantizer 6 is output to the variable-length encoding unit 2 described later, processed by the inverse quantizer 7 and the inverse DCT transform unit 8, and the output of the frame memory 9 is added to the output. The value is obtained and stored in the frame memory 9. Then, the low-pass filter 10 in the loop includes the frame memory 9 to improve the prediction efficiency.
Is filtered, and the difference between the predicted value obtained through the low-pass filter 10 and the original signal is encoded as a prediction error.

[0008] Reference numeral 2 denotes a variable length coding unit for performing variable length coding on the output from the coding unit 1 and outputting it as a transmission line code. A buffer memory 4 for accumulating and smoothing the data to be output in step (4). To change the controller.

When the controller 4 increases the BOC, that is, when the amount of information from the encoder 1 is larger than the amount of information read out corresponding to the transmission rate, the quantization step size in the quantizer 6 To reduce the amount of information input to the buffer memory 3 while reducing the amount of data occupied, that is, when the amount of information from the encoding unit 1 is smaller than the amount of information read corresponding to the transmission rate. Reduces the quantization step size in the quantizer 6 and increases the amount of information input to the buffer memory 3.

[0010]

By the way, in consideration of coding of a moving image when the transmission rate is low, the coding unit 1
It is necessary to make the quantization step size of the quantizer 6 rough in order to perform a large amount of information compression, which leads to a decrease in coding efficiency. There is a problem that the number of pixels per frame (frame rate) is reduced.

In an apparatus for compressing an input signal or an inter-frame prediction error signal by DCT,
A low-pass filter is provided in the preceding stage of the DCT encoder, and the characteristic of the low-pass filter is switched based on the buffer occupancy (see European Patent Application Publication No. 0396846). Such a conventional technique only discloses switching the characteristics of the low-pass filter based on the buffer occupancy, but does not disclose changing the characteristics of the low-pass filter with other parameters. .

The present invention was devised under such circumstances, and by finding other methods for reducing the resolution in the spatial direction without thinning out the pixels, the image quality is degraded and the number of pixels per unit time is reduced. It is an object of the present invention to provide an image encoding device that improves encoding efficiency without causing a decrease.

[0013]

FIG. 1 is a block diagram showing the principle of the present invention. In FIG. 1, reference numeral 1A denotes a DCT transform coding unit (DCT) for coding input image information using DCT transform.
T-type encoder), 2 is a variable-length encoding unit that performs variable-length encoding on the output from the encoding unit 1A and outputs it as a transmission line code, and 3 is a unit that keeps the data from the variable-length encoding unit 2 constant. A buffer memory for image storage for storing and smoothing the data so as to output the data at a rate, and a quantizer 4A for the encoding unit 1A according to the BOC of the buffer memory 3 (FIGS.
This is a controller having a function of changing the quantization step size (see reference numeral 6 in FIG. 8).

Reference numeral 12 denotes a low-pass filter provided before the encoding unit 1A.
The low pass filter pass band changing unit for changing the pass band of No. 2 is provided in the controller 4A. This low-pass filter pass band changing unit is set in advance.
Which changes the pass band of the low-pass filter 12 (claims 1, 2) and in accordance with the heat transmission rate and frame rate was,
It is conceivable to change the pass band of the low-pass filter 12 according to the sum of absolute values of prediction errors, the sum of absolute values of inter-frame differences, and the sum of absolute values of motion vectors (claims 3 to 5).

[0015]

In the above-described image encoding apparatus of the present invention, even if the number of pixels to be encoded is determined and sub-sample encoding or sub-line encoding cannot be applied, the encoding unit is used to reduce the spatial resolution. The low-pass filter 12 is applied to the image information input to 1A to reduce high-frequency components and perform high-efficiency compression.

Here, in the encoding unit 1A, encoding using DCT transform is performed. Since DCT transform encoding is performed by utilizing the concentration of pixel signals at low frequencies, the low-pass filter 12 is used. To reduce the high frequency components in advance, the compression efficiency increases. At this time, the pass band of the low-pass filter 12 is changed according to the transmission rate, the frame rate, the sum of absolute values of prediction errors, the sum of absolute values of differences between frames, and the sum of absolute values of motion vectors.

[0017]

Embodiments of the present invention will be described below with reference to the drawings.

(A) Description of First Embodiment FIG. 2 is a block diagram showing a first embodiment of the present invention. As shown in FIG. A DCT transform coding unit 1A for coding image information, a variable length coding unit 2 for performing variable length coding on the output from the coding unit 1A and outputting it as a transmission line code, and a variable length coding unit And a buffer memory 3 for image storage for accumulating and smoothing the data to output the data from the buffer memory 3 at a constant rate.
A controller 4A having a function of changing the quantization step size of the quantizer of the encoding unit 1A according to the OC is provided, and a low-pass filter 12 is provided at a stage preceding the encoding unit 1A.

In this embodiment, a low pass filter pass band changing unit 13A for changing the pass band of the low pass filter 12 is provided in the controller 4A. The low-pass filter pass band changing unit 13A performs the low-pass filter 12 according to a preset transmission rate.
Is configured to change the pass band of

With the above-described configuration, basically, the image information from which the high-frequency component has been removed in advance by the low-pass filter 12 is subjected to high-efficiency compression in the DCT transform coding section 1A, and the variable length coding section 2 3 and output. Here, as in the conventional case, the controller 4A
The quantization step size of the quantizer in the DCT transform coding unit 1A is adjusted according to the BOC of the buffer memory 3.

In this embodiment, the controller 4A
The characteristic of the low-pass filter 12, that is, the pass band is controlled in accordance with the transmission rate by the low-pass filter pass band changing unit 13A provided in the control unit.

By inputting the input image information to the encoding unit 1A after passing through the low-pass filter 12, it is possible to certainly compress the amount of information. Also does not improve the image quality. Further, as shown in FIG. 3, the amount of generated information and the image quality can be controlled by changing the pass band of the low-pass filter 12, so that when the transmission rate is high, the low-pass filter having a relatively wide pass band is used. When the transmission rate is low, it is effective to apply a low-pass filter 12 having a relatively narrow pass band.

Therefore, for example, as the low-pass filters 12 having different pass bands, filters having the characteristics indicated by A to X in FIG. 3 are prepared, and transmitted by the low-pass filter pass band changing unit 13A of the controller 4A. By managing the rate and providing several types of thresholds Th1 to Thn,
The pass band of the low-pass filter 12 is changed as follows.

When the transmission rate ≤ TH1, the low-pass filter of the characteristic A. When the TH1 <the transmission rate ≤ TH2, the low-pass filter of the characteristic B.

As described above, according to the first embodiment of the present invention, the resolution in the spatial direction can be reduced without thinning out the pixels, and the image quality and the number of pixels per unit time are not reduced. In addition, the coding efficiency is greatly improved, and by changing the pass band in the low-pass filter 12 according to the transmission rate, when the transmission rate increases, a high-quality image can be transmitted accordingly. .

(B) Description of the Second Embodiment FIG. 4 is a block diagram showing a second embodiment of the present invention. As shown in FIG. 4, the device of this embodiment is also the same as the first embodiment shown in FIG. Although the configuration is almost the same as that of the example, in the device of the second embodiment, the low-pass filter pass band changing unit 13B provided in the controller 4A is adapted to control the low-pass filter 12 according to a preset transmission rate and frame rate.
Is configured to change the pass band of

Even when the transmission rate is low, the deterioration of the image quality can be suppressed by lowering the frame rate. Therefore, in this embodiment, when the transmission rate is low and the frame rate is high, the low-pass filter 12 having a narrow pass band is used. By controlling the transmission, higher-quality image transmission can be performed.

(C) Description of Third Embodiment FIG. 5 is a block diagram showing a third embodiment of the present invention. As shown in FIG. 5, the apparatus of the present embodiment also includes the first embodiment shown in FIG. The DCT transform coding unit 1B according to the present embodiment is configured substantially in the same manner as that of the conventional example shown in FIG.
A prediction error evaluator 14 is provided in the T-transform encoder 1B,
Low-pass filter pass band changing unit 1 of controller 4A
The 3D is configured to change the pass band of the low-pass filter 12 according to the absolute value sum of the prediction error from the prediction error evaluation unit 14.

In this embodiment, a hybrid predictive encoder combining motion compensation and DCT is taken as an example. In such a predictive encoder, a signal obtained from a frame memory 9 through a low-pass filter 10 in a loop is obtained. The difference between the predicted value and the original signal is encoded as a prediction error.

In general, the larger the prediction error, the larger the amount of generated information tends to be. Therefore, in this embodiment, a plurality of types of low-pass filters 12 having different pass bands are prepared, and the prediction error evaluation unit is provided. The low-pass filter 12 is switched according to the sum of the absolute values of the prediction errors from.

That is, when the sum of the absolute values of the prediction errors increases, the low-pass filter 12 having a relatively narrow pass band is used.
On the other hand, if the absolute value sum of the prediction error becomes small, the low-pass filter 12 having a relatively wide pass band is used.
Is controlled.

(E) Description of the Fourth Embodiment FIG. 6 is a block diagram showing a fourth embodiment of the present invention. As shown in FIG. 6, the apparatus of the present embodiment also includes the first embodiment shown in FIG. The DCT transform coding unit 1B according to the present embodiment is configured in substantially the same manner as the conventional one shown in FIG. 8, but the apparatus according to the fourth embodiment employs a difference between frames. That is, the prediction error in the motion vector (0, 0) is obtained by the motion compensating unit 11 and accumulated in the inter-frame difference accumulating unit 15, and the low-pass filter pass band changing unit 13E provided in the controller 4A outputs the inter-frame difference. The pass band of the low-pass filter 12 is changed according to the absolute value sum of the inter-frame difference from the accumulating unit 15.

In this embodiment, since the inter-frame difference already obtained at the time of the motion vector search is used for control, the prediction error evaluator 14 in the preceding stage of the DCT converter 5 in the third embodiment is not required.

In general, the sum of the absolute values of the inter-frame differences tends to increase and the amount of generated information tends to increase for an image having a large motion. Therefore, in the present embodiment, the low-pass filter As the sum of absolute values of the inter-differences increases, a filter having a relatively narrow pass band is selected, and control is performed so as to suppress the amount of generated information.

(F) Description of Fifth Embodiment FIG. 7 is a block diagram showing a fifth embodiment of the present invention. As shown in FIG. 7, the apparatus of this embodiment is also the same as the first embodiment shown in FIG. The configuration is substantially the same as that of the example, and the DCT transform coding unit 1B in the present embodiment is configured substantially the same as the conventional one shown in FIG. The sum of absolute values is obtained by the motion compensator 11, accumulated in the motion vector accumulator 16, and provided to the low-pass filter pass band changing unit 13 provided in the controller 4 </ b> A.
F is configured to change the pass band of the low-pass filter 12 according to the sum of absolute values of the motion vectors from the motion vector accumulating unit 16.

In general, the larger the motion of an image, the larger the sum of absolute values of motion vectors and the larger the amount of generated information. Therefore, in this embodiment, the low-pass filter As the sum of absolute values of the vectors is larger, a filter having a relatively narrow pass band is selected, and control is performed so as to suppress the amount of generated information.

[0037]

As described above in detail, according to the image coding apparatus of the present invention (claims 1 to 5), a low-pass filter is provided at a stage prior to the discrete cosine transform type coder , and a predetermined value is set.
Change the passband of the low-pass filter 12 in accordance with the heat transmission rate and frame rate, the absolute value sum of the prediction error absolute value sum of interframe differences, the low-pass filter 12 in accordance with the absolute value sum of the motion vector By providing a low-pass filter pass band changing unit that changes the pass band, the resolution in the spatial direction can be reduced without thinning out the pixels, without deteriorating the image quality or reducing the number of pixels per unit time. This has the advantage that the coding efficiency can be greatly improved.

[Brief description of the drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a block diagram showing a first embodiment of the present invention.

FIG. 3 is a graph showing frequency characteristics of a low-pass filter.

FIG. 4 is a block diagram showing a second embodiment of the present invention.

FIG. 5 is a block diagram showing a third embodiment of the present invention.

FIG. 6 is a block diagram showing a fourth embodiment of the present invention.

FIG. 7 is a block diagram showing a fifth embodiment of the present invention.

FIG. 8 is a block diagram showing a conventional example.

[Explanation of symbols]

1A, 1B DCT transform encoder (discrete cosine transform encoder) 2 Variable length encoder 3 Buffer memory 4A controller 5 DCT transformer 6 Quantizer 7 Inverse quantizer 8 Inverse DCT transformer 9 Frame memory 10 Low pass Filter 11 Motion compensating unit 12 Low-pass filter 13A, 13B, 13D, 13E, 13F Low-pass filter pass band changing unit 14 Prediction error evaluating unit 15 Inter-frame difference accumulating unit 16 Motion vector accumulating unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Mitsunori Ohno 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (72) Inventor Takashi Ito 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Fujitsu Limited (72) Inventor Kiichi Matsuda 1015 Uedanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited (56) Reference European publication 396846 (EP, A1)

Claims (5)

(57) [Claims] 1. A picture coding apparatus having a discrete cosine transform type encoder, a low-pass filter provided in front of the discrete cosine transform type encoder, of the low-pass filter according to a preset heat transmission rate a low-pass filter passband changing unit for changing the passband characterized the kite is provided, the image coding apparatus. 2. The low-pass filter pass band changing unit ,
Characterized in that it is configured to change the passband of the low-pass filter according to a preset frame rate, the image coding apparatus according to claim 1. 3. An image coding apparatus having a discrete cosine transform encoder, comprising: a low-pass filter provided before the discrete cosine transform encoder; and a low-pass filter that passes through the low-pass filter according to a sum of absolute values of prediction errors. and wherein the kite and a low-pass filter passband changing unit for changing the band is provided, the image coding apparatus. 4. An image coding apparatus having a discrete cosine transform encoder, comprising: a low-pass filter provided at a stage preceding the discrete cosine transform encoder; and a low-pass filter according to a sum of absolute values of inter-frame differences. a low-pass filter passband changing unit for changing the passband characterized the kite is provided, the image coding apparatus. 5. An image coding apparatus having a discrete cosine transform encoder, comprising: a low-pass filter provided before the discrete cosine transform encoder; and a low-pass filter that passes through the low-pass filter according to a sum of absolute values of motion vectors. and wherein the kite and a low-pass filter passband changing unit for changing the band is provided, the image coding apparatus.