JP2943870B2 - Method and apparatus for transmitting motion amount of image transmission apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] For example, with regard to a motion amount transmission method in an image transmission device such as a high-efficiency encoding device for a moving image, the purpose of the transmission side is to further reduce the amount of motion information to be transmitted. An initial deviation of the motion amount of the target block is obtained from the detected motion amount of the peripheral block of the target block based on a predetermined algorithm, and a search range smaller than the maximum motion detection range is set based on the initial deviation. The deviation of the motion amount from the initial deviation within the search range is obtained, and this deviation information is sent to the receiving side. The deviation is obtained, and the initial deviation and the received deviation are combined to obtain the motion amount of the block of interest.

[Industrial applications]

The present invention relates to a motion amount transmission method and apparatus in an image transmission device such as a high-efficiency video encoding device.

In recent image transmissions such as video conference telephone systems, various methods have been used to reduce the bit rate by compressing the amount of transmission information as much as possible in order to use transmission lines effectively and reduce terminal costs. Proposed.

Above all, a so-called block matching type motion compensation prediction method has come into the spotlight as this kind of information amount compression method. In this method, the motion of the transmission screen is detected, and the prediction screen is compensated for the motion to obtain a prediction error (difference signal). The bit rate has been increased.

In such a motion-compensated predictive coding method, not only the signal amount of the prediction error is reduced but also the signal amount of the motion information to be transmitted is further reduced in order to further reduce the bit rate of the transmission information. Is required.

[Conventional technology]

2. Description of the Related Art Conventionally, as a motion amount encoding method in the motion compensation prediction encoding method, there is the following method. That is, (1) the Huffman coding method of encoding with a variable length according to the frequency of occurrence because the frequency of occurrence is high when the amount of motion of the image is small. In the case of an image, there is a combination method in which run-length encoding is performed when the motion amount is zero, and Huffman encoding is performed according to the frequency of occurrence in other cases.

Here, as the motion amount information to be transmitted, a maximum range in which a block of interest (hereinafter, referred to as a target block) to be coded is assumed to move is set, and the motion of the target block within this maximum range is set. Motion vector information indicating a position is used.

[Problems to be solved by the invention]

In the conventional motion amount coding method, high prediction efficiency and high coding efficiency can be obtained when an object is moving at a constant speed. However, when a stationary object starts to move suddenly, when a moving object suddenly stops, or when the object almost stops, the prediction efficiency and the coding efficiency decrease.

The present invention has been made in view of such a problem.
In order to obtain high prediction efficiency and high coding efficiency not only when the object is moving at a constant speed, but also when the stationary object suddenly starts moving or when the object is almost stationary. The purpose is to do.

[Means for solving the problem]

 FIG. 1 is a principle block diagram according to the present invention.

According to the image encoding apparatus of the present invention, the error based on the motion vector of the block at the same position on the previous screen for the current block to be coded on the current screen, The error based on the vector and the means for obtaining the error based on the motion vector already obtained for the peripheral block around the block of interest on the current screen are compared with the three errors obtained by the means for obtaining the error. Means for selecting a motion vector having a small error from the selected motion vector, and using the selected motion vector as an initial displacement vector to search for an optimal motion vector from a predetermined search range, and A deviation calculating means for obtaining deviation vector information from the position vector, and a motion vector having a small error by the selecting means. Along with the selection information indicating which one has been selected as the motion vector, at least the deviation vector information of the block at the same position on the previous screen selected as the motion vector having a small error by the selection means is transmitted to the receiving side as motion amount information. Means.

Further, the image decoding apparatus according to the present invention uses the motion vector of the block at the same position on the previous screen and the vector of motion zero as the motion vector having a small error with respect to the target block to be encoded on the current screen on the encoding side. Selection information indicating which one of the motion vectors of the peripheral blocks of the target block has been selected, and searching for an optimal motion vector from a predetermined search range using the selected motion vector as an initial deviation vector. Receiving means for receiving motion amount information including the obtained optimum motion vector and deviation vector information of the initial deviation vector, wherein the selection information selects a motion vector of a block at the same position on the previous screen The motion vector of the block at the same position on the screen before decoding, which has already been decoded on the receiving side, is set as the initial deviation vector. First means for obtaining a motion vector of the block of interest from the received deviation vector information by synthesis, and when the selection information indicates that the vector with zero motion is selected, the reception means is prepared in advance. A second means for obtaining a motion vector of the block of interest based on a vector of zero motion; and, when the selection information indicates that the motion vector of the peripheral block has been selected, the reception side decodes the current picture already decoded on the receiving side. Third means for obtaining a motion vector of the block of interest based on a motion block of a peripheral block at a position corresponding to the peripheral block on the transmission side.

[Action]

In the image coding apparatus on the transmission side, for the target block to be coded on the current screen, the means for obtaining the error is based on the error based on the motion vector of the block at the same position on the previous screen, and Then, an error based on the motion vector which has been obtained for the peripheral block around the block of interest on the current screen is obtained. The selection means compares these errors and selects a motion vector having a small error from among them. The deviation calculating means searches for an optimal motion vector from a predetermined search range using the selected motion vector as an initial deviation vector, and calculates deviation vector information between the retrieved optimal motion vector and the initial deviation vector. Ask for. The transmitting means receives, as the motion amount information, at least the deviation vector information of the block at the same position on the previous screen selected as the motion vector having the small error, together with the selection information indicating which one has been selected as the motion vector having the small error. Send to the side.

In the image decoding device on the receiving side, the receiving means receives the motion amount information. The first means, when the selection information indicates that the motion vector of the block at the same position on the previous screen is selected based on the received motion amount information, has already decoded on the reception side. The motion vector of the block of interest is obtained by combining the motion vector of the block at the same position on the previous screen as the initial deviation vector from the received deviation vector information. The second means obtains a motion vector of the block of interest based on a vector of zero motion prepared in advance on the receiving side when the selection information indicates that the vector of zero motion is selected. When the selection information indicates that the motion vector of the peripheral block has been selected, the third means determines whether or not the peripheral block at a position corresponding to the peripheral block on the transmitting side in the current screen already decoded on the receiving side. A motion vector of the block of interest is obtained based on the motion vector.

〔Example〕

 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 2 shows an image encoding apparatus according to an embodiment of the present invention. In this embodiment, a motion detection type motion compensation predictive coding circuit having an initial deviation and a motion amount coding unit are shown.

In FIG. 2, a subtractor 11 is a one-block input screen X.
And a motion compensation prediction screen output from the variable delay unit 17 and outputs the result as a prediction error screen ε. Quantizer
12 quantizes the prediction error screen ε. Huffman coding circuit
13 performs Huffman coding on the output q of the quantizer 12.

The adder 14 spreads the output q of the quantizer 12 and the prediction screen to generate a local decoding screen X '. The frame memory 15 stores the previous picture obtained by delaying the local decoded picture X 'of the adder 14 by about one frame. The variable delay unit 17 is a frame memory 15
Is shifted according to the motion vector from the motion compensator 16 to generate a motion compensated prediction screen.

The motion compensator 16 receives the input screen X and the previous screen of the frame memory 15 as input, detects a motion amount (motion vector V _D ) of the input screen X based on these, and detects the detected motion vector V _D
Is sent to the variable delay unit 17, and various kinds of motion amount information are sent to the motion amount encoding unit 18. The motion amount information includes 1-bit motion zero information for determining whether the motion is zero or not.
B ₁ , motion deviation information ΔV, and 1-bit specific motion amount selection information B ₂ for determining whether a specific motion amount has been selected.

The motion amount information output from the motion compensator 16 is encoded by the motion amount encoding unit 18 and output. The motion amount encoding part 18 includes a run length encoding circuit 181 and a Huffman encoding circuit 182.

Run-length coding circuit 181 sends a high motion zero information B ₁ occurrence frequency to a decoder side by run length encoding.
On the other hand, Huffman coding circuit 182 sends a plurality of different values and the possible motion deviation information ΔV to a specific motion amount selection information B ₂ to the decoder side by Huffman coding as a group.

The motion detection operation in the motion compensator 16 will be described in detail below. First, the target block X is an input screen of interest be currently coded, already on the basis of the detection motion vectors of neighboring blocks of the coding already, the determination of initial vector V ₁ in accordance with a predetermined algorithm .
For example, as shown in Figure 4, the motion vector V _M of the block M in the same position in the previous frame with respect to the target block X, the motion vector V _K of blocks K immediately above the target block X in the current screen, or, Zero motion vector V _D (0,
0) of the cumulative error sum of the target block X after deflection to the minimum motion vector and initial vector V _1.

When initial vector V ₁ is determined, then the difference vector ΔV by searching within a predetermined search range E, to determine the best motion vector V _D. That is, as shown in FIG. 5, the target block X to set the initial vector V ₁ in accordance with excursions to predetermined search range E about the block X ₁ was,
Performs block matching while moving the block X ₁ within the search range E, finding a small block X _D The most error accumulation sums. Then the direction vector from the block X ₁ to block X _D and deviation vector [Delta] V. In this case, optimum motion vector V _D of the block of interest X is the vector sum of the initial vector V ₁ and the deviation vectors [Delta] V. That is, V _D = V ₁ + ΔV.

Here, the search range E is set to be sufficiently smaller than the maximum range Emax for motion detection of the target block X. In this embodiment, the search range E is a range of ± 3 pixels vertically and horizontally, and the maximum detection range Emax is a range of ± 15 pixels vertically and horizontally, and the number of matching blocks in the search range E is 49. Thus, the number of reference blocks in the search range E is sufficiently smaller than the number of reference blocks in the maximum detection range Emax.

Here bit to "1" for a specific motion amount selection information B ₂ if you select the motion vector V _K of blocks K immediately above the target block X as the motion vector V _D. The motion amount is set to "1" bit of the motion information B ₁ represents the case of zero.

FIG. 3 shows a block configuration of an image decoding apparatus according to an embodiment of the present invention. This embodiment shows a motion compensation type predictive decoding circuit of a motion detection type having an initial deviation and a motion amount decoding unit.

In FIG. 3, a Huffman decoding circuit 21 performs Huffman decoding of the quantized output (A). The adder 22 restores the decoded signal X 'by adding the output signal of the Huffman decoding circuit 21 and the prediction signal from the variable delay unit 24. Frame memory
23 gives the decoded signal X 'a delay of one frame and sends it to the variable delay unit 124. Variable delay 24 is being displaced Back on the basis of the motion vector V _D obtained independently generates a motion compensated prediction picture from the transmitting side at the receiving side.

Run-length decoding circuit 25 provides the control input terminal Pc of the switching circuit 20 and run-length decoding the motion zero information B ₁ from a signal which has been transmitted (B). The Huffman decoding circuit 26 calculates the motion deviation information Δ from the transmitted signal (C).
The and the specific motion amount selection information B ₂ V and Huffman decoding, the deviation information ΔV to the adder 27, also is inputted specific motion amount selection information B ₂ to the control input terminal Pc of the switching circuit 29.

With motion memory 28 outputs to the adder 27 the final motion vector V _D applied to the variable delay unit 24, for example, by one frame time delay as initial vector V _1, the specific motion vector is delayed one block line output to the input terminal P ₁ of the switch circuit 29 as V _K.

The adder 27 is input to the input terminal P ₂ of the switch circuit 29 as initial vector V ₁ and adding to the final motion vector V _D from deviation vector ΔV and motion memory 28 from Huffman decoding circuit 26 . The output signal from the switch circuit 29 is input to the input terminal P ₁ of the switch circuit 20. For the input terminal P ₂ of the switch circuit 20. A signal of V _D (0,0) corresponding to a motion vector of zero motion is input. This switch circuit 2
The output signal from 0 is input to the variable delay unit 24 and also to the motion memory 28.

These switch circuits 20, 29 to the "0" signal to receive the input terminal P ₁ side to the control input terminal Pc, whereas "1" signal to receive and switched to the input terminal P ₂ side.

The operation of the decoding circuit will be described. When both the motion zero information B ₁ and the specific motion selection information B ₂ are “0”, the deviation vector ΔV decoded by the Huffman decoding circuit 26 is added to the adder 27.
And the initial deviation vector V ₁ read from the motion memory 28
Vectorially summed with the motion vector V _D is produced as. The motion vector V _D is inputted to the variable delay unit 24 via the switching circuit 29 and the switch circuit 20, being displaced Back frame memory 23, it generates a predictive picture that is motion compensated.

On the other hand, when the motion zero information B ₁ is “1”, the switch circuit 20 selects the zero vector V _D (0,0) and sends it to the variable delay unit 24.
Input. In the case of a particular motion amount selection information B ₂ "1", the switch circuit 29 selects a specific motion vector V _K from the motion memory 28, which via the switch circuit 20 is input to the variable delay device 24. Here particular motion vector V _K is equivalent to the motion vector V _D of the block K immediately above the target block X.

〔The invention's effect〕

As described above, according to the present invention, when the object is moving at a constant speed, the motion vector at the same position on the previous screen is obtained, and when the object is almost stopped, the zero motion vector is obtained. When a stationary object suddenly starts moving, the motion vector of the surrounding block is selected as the motion vector of the block of interest and coding is performed. Therefore, even when a stationary object suddenly starts moving or when the object is almost stationary, a motion vector with high prediction accuracy can be selected accurately, which enables the calculation of a deviation vector. The prediction accuracy is also increased, and thus the motion detection accuracy is increased as a whole, and a high prediction efficiency and a high coding efficiency are obtained. Alternatively, in order to obtain the same prediction efficiency or coding efficiency, a motion vector with high prediction accuracy is accurately selected, so that the maximum search range can be reduced, and a predetermined calculation of a plurality of vectors can be performed when selecting an initial vector. Even if consideration is given, it can be realized with a reduced amount of calculation or a reduced circuit scale as a whole.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the principle of the present invention, FIG. 2 is a block diagram showing an image encoding apparatus as one embodiment of the present invention, and FIG. 3 is an image decoding apparatus as one embodiment of the present invention. block diagram, FIG. 4 is an explanatory view of a selection of initial vector, and, FIG. 5 is an explanatory diagram of a search range of the optimal motion vector V _D. In the figure, 11 ... subtractor 12 ... quantizer 13,182 ... Huffman coding circuit 14,22,27 ... adder 15,23 ... frame memory 16 ... motion compensator 17,24 ... variable delay Unit 18: Motion amount encoding unit 181: Run-length encoding circuit 21, 26: Huffman decoding circuit 25: Run-length decoding circuit 28: Motion memory 20, 29: Switch circuit

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Kiichi Matsuda 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (56) References JP-A-56-143776 (JP, A) (58) Fields investigated (Int.Cl. ⁶ , DB name) H04N 7/137 H04N 7/13

Claims (3)

(57) [Claims] 1. A transmitting side, for a target block to be coded on a current screen, an error based on a motion vector of a block at the same position on a previous screen, and a vector of zero motion when the target block is set to zero motion. And the errors obtained based on the motion vectors already obtained for the surrounding blocks around the block of interest on the current screen are compared, and a motion vector with a small error is selected from these, and the selected motion vector is An optimal motion vector is searched from a predetermined search range as an initial deviation vector, and deviation vector information between the searched optimal motion vector and the initial deviation vector is obtained. Along with the selection information indicating whether or not the motion vector has been selected, at least The received deviation vector information of the block at the same position on the previous screen is sent to the receiving side as motion amount information. On the receiving side, based on the received motion amount information, the selection information is converted to the same position on the previous screen. When it is indicated that the motion vector of the block has been selected, the receiving side uses the motion vector of the block at the same position on the previous screen that has already been decoded as the initial deviation vector from the received deviation vector information to obtain the target block. And the motion vector of
When the selection information indicates that the zero motion vector has been selected, the motion vector of the target block is obtained based on the zero motion vector prepared in advance on the receiving side. When the motion vector of the target block is selected, the motion vector of the target block is calculated based on the motion vector of the peripheral block at the position corresponding to the peripheral block on the transmitting side in the current screen which has already been decoded on the receiving side. Of transmitting a motion amount of an image transmitting apparatus. 2. An error based on a motion vector of a block at the same position on the previous screen for a block of interest to be coded on the current screen, an error based on a vector of zero motion when the block of interest is zero motion, Means for calculating errors based on the motion vectors already obtained for the peripheral blocks around the block of interest on the current screen are compared with the three errors obtained by the means for obtaining the errors. Means for selecting a vector, and using the motion vector selected by the selecting means as an initial displacement vector, searching for an optimal motion vector from a predetermined search range, and determining the searched optimal motion vector and initial displacement vector. The deviation calculating means for obtaining the deviation vector information of Transmitting means for transmitting, as motion amount information, deviation vector information of a block at the same position on the previous screen selected as a motion vector having a small error by at least the selecting means, together with selection information indicating whether or not the motion vector has been selected; An image encoding device comprising: 3. A motion vector having a small error with respect to a target block to be coded on the current screen on the coding side, a motion vector of a block at the same position on the previous screen, a vector of zero motion, and a peripheral block of the target block. Selection information indicating which one of the motion vectors has been selected, and the optimum motion vector obtained by searching for an optimum motion vector from a predetermined search range using the selected motion vector as an initial deviation vector. Receiving means for receiving the motion amount information including the accurate motion vector and the deviation vector information of the initial deviation vector, and the selection information indicates that the motion vector of the block at the same position on the previous screen is selected. In some cases, the motion vector of the block at the same position on the previous screen that has already been decoded on the receiving side is used as the initial deviation vector and the received deviation vector First means for synthesizing the motion vector of the block of interest from the torque information, and a motion-zero vector prepared in advance on the receiving side when the selection information indicates that the motion-zero vector has been selected. A second means for obtaining a motion vector of the block of interest based on the above, and when the selection information indicates that a motion vector of the peripheral block has been selected, the vicinity of the transmitting side in the current screen already decoded on the receiving side. Third means for obtaining a motion vector of the block of interest based on a motion block of a peripheral block at a position corresponding to the block.