JP2003047009A - Hierarchical encoding method and decoding method for digital image signal - Google Patents

Abstract

(57) [Summary] [PROBLEMS] Despite employing a hierarchical representation, the amount of transmission data does not increase, the scale of hardware is small, and high-speed decoding is enabled. SOLUTION: A first layer is constituted by an input digital image signal. 2 × 2 = 4 pixels (a, b,
An average value m1 of c and d) is formed. The second layer is composed of the average values thus formed. The transmission of one pixel d among the four pixels is omitted. Therefore, the transmission data amount is the average value data and three pixels, and is equal to the data amount of the original four pixels. An average value M1 is formed from 2 × 2 data of the second hierarchy. M1 to M4 are the third hierarchy. Further, an average value M of M1 to M4 is formed. M is the fourth hierarchy. Not only the average value, but also the difference data with respect to the average value of the upper layer, excluding the highest layer, may be transmitted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention, for example, divides a digital image signal into a plurality of signals expressing different resolutions, encodes each signal, and transmits the digital image signal. Regarding the method of conversion.

[0002]

2. Description of the Related Art A high-resolution image signal is used as a first layer,
An image signal of the second layer having a lower resolution than this, an image signal of the third layer having a lower resolution than the image signal of the second layer,
.. has been proposed (referred to as hierarchical coding). According to this encoding, image signals of a plurality of layers can be transmitted in one transmission path (communication path, recording / reproducing process).
The transmission image data can be reproduced on the receiving side by any one of the television monitors corresponding to the plurality of layers.

More specifically, a standard definition video signal,
High-resolution video signals such as high-definition signals, image data of computer displays, low-resolution video signals for high-speed retrieval of image databases, etc. exist as video signals of different resolutions. In addition to high and low resolutions, such hierarchical coding can be applied to image reduction.

FIG. 8 shows an example of the hierarchical coding apparatus proposed previously. This example is a three-layer encoding, and the first
The image signal of the second layer in which the number of pixels is 1/4 with respect to the image signal of the layer, the third layer (the highest layer) in which the number of pixels is 1/16
The image signal is transmitted. As shown in FIG.
A digital image signal is supplied to an input terminal 41. This input image signal is the first layer signal.
The input image signal is supplied to the thinning circuit 42 and the subtraction circuit 43. The coding circuit 45 is connected to the thinning circuit 42 via the thinning circuit 44, and the output terminal 53 for the third layer is derived from the coding circuit 45.

Each of the thinning circuits 42 and 44 is
The number of pixels is halved in the horizontal and vertical directions, and the number of pixels is reduced to ¼. Thinning circuit 44 in which the number of pixels is 1/16 with respect to the input image signal
The output image signal of is encoded by the encoding circuit 45 and is output to the output terminal 53. Usually, thinning circuits 42 and 4
For 3, a thinning filter is used.

The output image signal of the thinning circuit 42 is supplied to the interpolation circuit 46 and the subtraction circuit 47. Interpolation circuit 46
Interpolates the thinned pixels, and the output image signal of the interpolation circuit 46 is supplied to the subtraction circuit 43. In the subtraction circuit 43,
The difference between the input image signal and the output signal of the interpolation circuit 46 is calculated for each pixel. The difference signal from the subtraction circuit 43 is encoded by the encoding circuit 48. From the encoding circuit 48 to the first
The output terminal 51 for the hierarchy is derived.

The output signal of the thinning circuit 44 is an interpolation circuit 49.
Is supplied to the subtraction circuit 47 via. The subtraction circuit 47
The difference value between the output signal of the thinning circuit 42 and the output signal of the interpolation circuit 49 is calculated for each pixel. The difference signal from the subtraction circuit 47 is supplied to the encoding circuit 50. Encoding circuit 50
From the output terminal 52 for the second layer. The interpolation circuits 46 and 49 are usually composed of interpolation filters.
The encoding circuits 45, 50 and 48 perform encoding for compressing the transmission data amount.

The signal generated at the output terminal 51 is a first layer encoded differential signal, and the signal generated at the output terminal 52 is the second differential signal.
This is a layered encoded difference signal, and the signal generated at the output terminal 53 is a third layer encoded difference signal. As described above, the conventional hierarchical encoding device forms an upper layer by thinning out, and in each layer, generates difference data from data in which the data of the upper layer is interpolated once, and the data of the highest layer and other layers are generated. And the difference data of are encoded respectively.

FIG. 9 shows the configuration of a decoding device corresponding to the conventional coding device of FIG. To the input terminals 61, 62 and 63, the above-mentioned signal of the highest layer and the coded difference signals of the other layers are respectively supplied. The signal of the highest hierarchy from the input terminal 63 is supplied to the decoding circuit 64, the output signal of the decoding circuit 64 is output to the output terminal 73 and the interpolation circuit 67.

The second layer encoded differential signal from the input terminal 62 is decoded by the decoding circuit 65, and the differential signal is supplied to the adding circuit 68. The adder circuit 68 adds the difference signal to the image signals having the same resolution by the interpolation circuit 67,
The image signal of the second layer is obtained at the output terminal 72.

The encoded differential signal from the input terminal 61 is supplied to the decoding circuit 66, and a third level differential signal is generated at the output thereof. This difference signal is supplied to the adding circuit 70. The addition circuit 70 adds the decoded difference signal to the image signal generated by the interpolation circuit 69. From the output terminal 71 connected to the adder circuit 70, the image signal of the first layer is taken out.

[0012]

The above-mentioned conventional layered coding has a drawback that the amount of information increases every time it is separated into layers. As described above, when the number of pixels is thinned out by 1/4, the number of data increases by 1.31≈ (1 + 1/4 + 1/16) times.
Further, even when the four pixels of the input image are averaged and an image of a layer above it is generated, the number of pixels to be transmitted is increased by (1 + 1/4) times because the average value is included. As the number of layers increases, the number of pixels that need to be transmitted is (1 + 1/4 + 1).
/ 16 + 1/64 + ...) Doubled.
Therefore, instead of obtaining the hierarchical representation, there is a problem that the coding efficiency is deteriorated.

Further, the conventional hierarchical coding decoding apparatus generates an image from the image data of the upper layer by the interpolation filter and adds the difference value to this image to obtain the image data of the layer. As a result, there is a problem that the scale of hardware increases and the operation time required for decoding increases.

Therefore, an object of the present invention is to prevent a decrease in coding efficiency, have a small hardware scale, and have a merit that a calculation time at the time of decoding is short, which is advantageous. And to provide a decoding method.

[0015]

According to a first aspect of the present invention, digital image data supplied with input image data is divided into at least first and second hierarchical data representing different resolutions. In an image signal hierarchical encoding method, an averaging step for forming data of a second layer by average value data of N pixel data of a first layer, and N pixels of the first layer The first encoding step of reducing the data to (N-1) pixel data and encoding the (N-1) pixel data, and the data of the second layer obtained in the averaging step are And a second encoding step for encoding, which is a hierarchical encoding method for a digital image signal.

According to a third aspect of the present invention, a hierarchy of digital image signals supplied with input image data and adapted to divide the input image data into at least first and second hierarchical data representing different resolutions. In the encoding method, an averaging step for forming data of the second layer by the average value data of N pixel data of the first layer, the average value data of the second layer and the first layer Difference data generation step for generating (N-1) difference data with respect to the pixel data, a first encoding step for encoding the (N-1) difference data, and an averaging step. And a second encoding step for encoding the obtained second layer data.

Since the data of the other layers are constructed by the average value or the difference from the average value, even if the transmission of one pixel data or one difference data is omitted, it can be restored on the receiving side. Therefore, although the data of each layer is transmitted, the number of transmission pixels does not increase. Further, there is an advantage that the calculation time is shortened on the decoder side, high-speed processing is possible, and the hardware scale is small.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration example of the encoding device. The digital image signal from the input terminal 1 is supplied to the averaging circuit 2 and the encoding circuit 8. In this example, the input signal forms the first layer, and the output terminal 12 for the first layer is derived from the encoding circuit 8. The averaging circuit 2 forms an average value of the levels of (N = 2 × 2 = 4) pixels, for example.

At the bottom of FIG. 3, the first partial image (8
× 8 pixels) are shown. In FIG. 3, one square represents one pixel. The averaging circuit 2 performs the processing of, for example, 1/4 (a + b + c + d) = m1. Therefore, in the output of the averaging circuit 2, the portion corresponding to (8 × 8 pixels) of the input image becomes (4 × 4 pixels). The output of the averaging circuit 2 is the second layer signal.

The output signal of the averaging circuit 2 is supplied to the averaging circuit 3 and the encoding circuit 7. From the encoding circuit 7 to the second
The output terminal 11 for the hierarchy is derived. The averaging circuit 3
The average value of the four (2 × 2) average values of the second layer is formed, and the data of the third layer is generated. For example, 1/4 (m1
+ M2 + m3 + m4) = M1 is performed. Therefore, in the output of the averaging circuit 3, the (8 × 8 pixel) area of the input image becomes the (2 × 2 pixel) area. The output of the averaging circuit 3 is the signal of the third layer.

Further, the output signal of the averaging circuit 3 is supplied to the averaging circuit 4 and the encoding circuit 6. Encoding circuit 6
From the output terminal 10 for the third layer. The averaging circuit 4 forms an average value from four (2 × 2) average values of the third hierarchy. That is, 1/4 (M1 + M2 + M3 +
M4) = M processing is performed. Therefore, in the output of the averaging circuit 4, (8 × 8) of the input image becomes one pixel. The output of the averaging circuit 4 is the signal of the fourth highest layer. The output signal of the averaging circuit 4 is supplied to the encoding circuit 5, and the output terminal 9 for the fourth layer is derived from the encoding circuit 5. Figure 3
As can be seen from the above, as the hierarchy changes, the number of pixels decreases to 1/4, 1/16, and 1/64. Therefore, if the area of the image is constant, the resolution will decrease at the same rate. If the distance between pixels is constant, the size of the image is reduced at the same rate.

The coding circuits 5, 6, 7, 8 perform coding for compressing the transmission data, and the other coding circuits 6, 7, 8 except the coding circuit 5 for the uppermost fourth layer, The process of extracting the data to be transmitted is also performed. That is, transmission of one pixel data among the data of four pixels is omitted. As a result, it is possible to prevent an increase in the amount of transmission data. For example, in the encoding circuit 7, the average values m1, m2 from the averaging circuit 2,
m3 and m4 are supplied. If these four average value data are transmitted, the amount of transmission data increases and the transmission efficiency decreases. Therefore, the encoding circuit 7 performs a process of omitting the transmission of one of these data, for example, m4. The other encoding circuits 6 and 8 also perform the processing of omitting the transmission of one data, as described above. In Figure 3, (4 x 4)
This is data in which the pixel data at the lower right with a line or the average value data is omitted among the pixels.

The number of pixels to be transmitted in such a hierarchical representation is calculated. 48 + 12 + when adding from the lower hierarchy
3 + 1 = 64 pixels, which is the same as the number of input pixels. Although the hierarchical representation is realized, the number of pixels to be transmitted does not increase.

Next, a decoding device corresponding to the coding device of FIG. 1 will be described with reference to FIG. In FIG. 2, 2
Input terminals 1, 22, 23, and 24 are supplied with the fourth, third, third, and first layer data transmitted from the above-mentioned encoder, respectively. Each input terminal 21-2
Decoding circuits 25, 26, 27, and 28 are connected to the circuit 4. The decoding circuits 25 to 28 decode the compression processing performed by the encoding circuits 5 to 8.

The output signal of the decoding circuit 25 is taken out to the output terminal 32 for the fourth layer, and the data reproducing circuit 29 is also used.
Is supplied to. The output signal of the decoding circuit 26 is also supplied to the data reproducing circuit 29. The data reproducing circuit 29 reproduces data that is not transmitted. That is, assuming that the data M4 that is not transmitted is M = 1/4 (M1 + M2 + M3 +
M4), the data reproduction circuit 29 determines that M4 = 4M
M4 can be reproduced by the calculation of − (M1 + M2 + M3). The output signal of the data reproducing circuit 29 is taken out to the output terminal 33 for the third layer.

The output signal of the decoding circuit 27 and the third layer signal from the data reproducing circuit 29 are supplied to the data reproducing circuit 30. The data reproducing circuit 30, like the above-mentioned data reproducing circuit 29, carries out a process of restoring data whose transmission is omitted. That is, m4 = 4M1- (m1
+ M2 + m3) makes it possible to reproduce m4. The output signal of the data reproducing circuit 30 is the output terminal 34 for the second layer.
Taken out.

The output signal of the decoding circuit 28 and the second layer signal from the data reproducing circuit 30 are supplied to the data reproducing circuit 31. The data reproducing circuit 31, like the data reproducing circuits 29 and 30 described above, carries out a process of restoring data whose transmission has been omitted. That is, d = 4m1-
D can be reproduced by the calculation of (a + b + c). The output signal of the data reproducing circuit 30 is taken out to the output terminal 34 for the second layer.

In order to obtain a desired hierarchical image in the decoding device, it is necessary that the images in the upper layers be decoded. Even if the corresponding data of the upper layer is a thinned pixel that is not transmitted, the thinned pixel can be reproduced by the data of the higher layer, so that no problem occurs.

Next, another embodiment of the present invention will be described. Another embodiment is an example in which, as shown in FIG. 4, among the first to fourth layers of the above-described embodiment, three layers except the fourth layer are handled.

FIG. 5 shows the configuration of an encoding device according to another embodiment of the present invention. The first layer digital image signal from the input terminal 1 is supplied to the averaging circuit 2 and the subtracting circuit 13. The subtraction circuit 13 generates difference data for the average value from the averaging circuit 2. Similar to the above-described embodiment, the averaging circuit 2 generates an average value m1 of the values of the first layer (2 × 2 pixels), for example, 4 pixels of a, b, c, and d.
The subtraction circuit 13 forms difference data of three pixels other than the pixel d that is not transmitted. That is, difference data of Δa = a-m1, Δb = b-m1, and Δc = c-m1 is formed.

The difference data from the subtraction circuit 13 is taken out to the output terminal 12 for the first layer via the encoding circuit 8. The output signal of the averaging circuit 2 is supplied to the averaging circuit 3 and the subtracting circuit 14. The subtraction circuit 14 is the subtraction circuit 13
Similarly, the difference data between the average value from the averaging circuit 3 and the average value from the averaging circuit 2 is formed. That is, m4
Difference data other than the difference data Δm1 = m1-M1, Δm2 = m2-M1, Δm3 = m
3-M1 is formed.

The difference data from the subtraction circuit 14 is taken out to the output terminal 11 for the second layer via the encoding circuit 7. Further, the data of the highest (third) layer from the averaging circuit 3 is supplied to the encoding circuit 6 without being converted into the difference and is taken out to the output terminal 10 of the encoding circuit 6. Also for the third layer, transmission of data for one pixel, which is shaded in FIG. 4, is omitted.

As can be seen from FIG. 4, the number of pixels required to be transmitted in another embodiment is (48 + 12 + 4 = 64 pixels). Equal to that of

As the coding circuits 5, 6, 7, and 8, linear quantization, nonlinear quantization, or adaptive quantization represented by ADRC (Dynamic Range Adaptive Coding) can be adopted. FIG. 7 shows examples of the quantization characteristics of the linear quantizer and the non-linear quantizer, respectively. By using these quantizers, it is possible to compress the amount of transmission data by reducing the number of bits per pixel.

FIG. 6 shows a decoding device corresponding to the coding device shown in FIG. The data of each layer from the input terminals 22, 23 and 24 is decoded by the decoding circuits 26, 27 and 28. The data of the third layer from the decoding circuit 26 is taken out to the output terminal 33.

The decoding circuits 27 and 28 output difference data of the second layer and the first layer. These difference data are supplied to the difference value reproducing circuits 36 and 37. These circuits 36 and 37 reproduce one difference value whose transmission is omitted in the encoding device. Difference value reproducing circuit 3
7 will be described. Since Δa + Δb + Δc + Δd = a + b + c + d-4m1 = 0, if Δa, Δb, and Δc are known, Δd can be calculated by Δd = − (Δa + Δb + Δc). Difference value reproducing circuit 36
Similarly, the thinned out difference data can be reproduced.

The difference data from the difference value reproducing circuit 36 is supplied to the adding circuit 38. The average value data of the third layer is supplied to the adder circuit 38. The data of the second hierarchy is obtained from the adder circuit 38 and is taken out to the output terminal 34. The difference data from the difference value reproducing circuit 37 is supplied to the adding circuit 39. The average value data of the second layer is supplied from the adding circuit 38 to the adding circuit 39. Therefore, the adder circuit 39 generates the data of the first layer and outputs it to the output terminal 35.

In the above-mentioned other embodiment, when the high-definition television still image database is constructed,
The hierarchical output is reproduction data of original image (high-definition) resolution, the second hierarchical output is reproduction data of resolution equivalent to standard television, and the third hierarchical output is low resolution image for high-speed search.

When compression coding is used for the purpose of reducing the amount of information, the reproduced image data obtained by the decoding device does not always match the input original image data, but it is visually deteriorated. Can be made undetectable. Further, the average value is not limited to the simple average value, but a weighted average value may be formed.

[0040]

According to the present invention, even if the hierarchical representation is introduced into the image, the number of pixels to be coded does not increase, so the coding efficiency does not decrease. Further, according to the present invention, since the image of a certain layer is restored based on the one pixel data of the upper layer, the delay is small. Further, since the present invention performs the averaging process, it does not require an interpolation filter and the scale of hardware can be simplified.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an apparatus corresponding to an encoding method according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an apparatus corresponding to the decoding method according to the embodiment of the present invention.

FIG. 3 is a schematic diagram for explaining an embodiment of the present invention.

FIG. 4 is a schematic diagram for explaining another embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of an apparatus corresponding to an encoding method according to another embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of an apparatus corresponding to a decoding method according to another embodiment of the present invention.

FIG. 7 is a schematic diagram for explaining compression encoding that can be adopted in the present invention.

FIG. 8 is a block diagram showing an example of a conventional hierarchical encoding device.

FIG. 9 is a block diagram showing an example of a conventional hierarchical encoding decoding device.

[Explanation of symbols]

2, 3, 4 ... Averaging circuit, 5, 6, 7, 8 ... Encoding circuit, 13, 14 ... Subtraction circuit, 29, 30, 3
1 ... Data reproduction circuit, 38, 39 ... Addition circuit

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Kunio Kawaguchi             6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Soni             -Inside the corporation F-term (reference) 5C059 KK06 KK15 LB05 MA28 MA32                       MD02 MD03 PP01 SS06 SS11                       UA01 UA02 UA04 UA05 UA11                 5J064 AA02 BB04 BC08 BC21 BD01

Claims (6)

[Claims] 1. A hierarchical encoding method for a digital image signal, which is supplied with input image data, and which divides the input image data into at least first and second hierarchical data, which expresses different resolutions. An averaging step for forming the data of the second layer by the average value data of N pixel data of the first layer, and (N-1) N pixel data of the first layer. First pixel step of encoding the (N-1) number of pixel data, and second encoding of the second layer data obtained in the averaging step. And a coding step of the digital image signal. 2. An input image data is supplied, and the input image data is divided into at least first and second layer data representing different resolutions, and an average value of N pixel data of the first layer. The data forms the data of the second layer, and reduces the N pixel data of the first layer to (N-1) pixel data, and (N-1)
A number of pixel data, and a decoding method corresponding to the layer coding method of coding the second layer data, decoding the (N-1) pixel data of the first layer. A first decoding step; a second decoding step for decoding the data of the second layer; and pixel data that is not output in the first layer of the second layer.
A method of decoding a digital image signal, comprising: a data reproducing step of restoring from the second layer data decoded in the decoding step and the first layer data decoded in the first decoding step. 3. A hierarchical encoding method for a digital image signal, which is supplied with input image data, and which divides the input image data into at least first and second hierarchical data, the method comprising: An averaging step for forming the data of the second layer by the average value data of N pixel data of the first layer, the average value data of the second layer and the pixel data of the first layer And (N-1) difference data generating step for generating difference data, a first encoding step for encoding the (N-1) difference data, and the averaging step. And a second encoding step for encoding the above-mentioned second layer data. 4. An input image data is supplied, the input image data is divided into at least first and second hierarchical data representing different resolutions, and an average value of N pixel data of the first hierarchical layer. From the data, the data of the second layer is formed, and (N-1) difference data between the average value data of the second layer and the pixel data of the first layer is generated. -1) encode the difference data,
In a decoding method corresponding to the layer coding method for coding the data of the second layer, a first decoding step of decoding the (N-1) difference data of the first layer, A second decoding step of decoding the data of the second layer, a difference data reproducing step of reproducing the non-output difference data in the first layer, the average value data of the second layer and the above A method of decoding a digital image signal, comprising: an addition step for adding the difference data obtained in the difference value reproducing step to reproduce the first layer data. 5. The method according to claim 1, claim 2, claim 3 or claim 4, wherein N = (2 × 2 pixels). 6. The method according to claim 1, claim 2, claim 3 or claim 4, wherein the average value data is generated by a weighted average.