JP2001128173A - Object encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object encoding device which can reduce the quantity of codes at the spot (block) of a background object which is concealed by an object having an arbitrary shape at when encoding the background object. SOLUTION: A VOP inside/outside discriminating means 30 discriminates the block of the background object corresponding to the position of the block (namely, block B) which is discriminated as a block in the VOP of an object picture having an arbitrary shape as the block which is completely concealed by the object picture having the arbitrary shape. Then the device does not encode the DCT factor of the picture element corresponding to the block of the background object picture which is discriminated to be completely concealed by the object picture having the arbitrary shape.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object encoding device using an object encoding method. It is another object of the present invention to provide an object encoding apparatus that can reduce the amount of code at the location (block) of a background object hidden by an arbitrary-shaped object when encoding a background object.

[0002]

2. Description of the Related Art In image coding, there is an object coding system in which different moving image sequences (object images) are superimposed, and this technology is also used in MPEG-4 and the like. The object coding method is a method of creating a new sequence by superimposing a rectangular object image serving as a background or an arbitrary-shaped object image serving as a foreground (for example, a person image or an object image cut out by a chroma key technique). There is a feature that various sequences can be created by changing the combination of each object.

FIG. 5 shows a block diagram of a conventional object coding apparatus. The pixel value signal encoding unit 52 of FIG.
Is a portion for encoding a pixel value signal representing the luminance and color difference of each object image (a rectangular object image serving as a background and an arbitrary-shaped object image serving as an object image). The shape signal encoding unit 53 encodes a shape signal representing the shape of the arbitrarily-shaped object image (object image). The shape signal is identification data indicating whether each pixel is included inside the object or outside the object. Since the discrimination between the inside and the outside of the object can be expressed by one bit, the shape signal is defined as a binary signal corresponding to each pixel of the luminance signal on a one-to-one basis. Further, the transparency signal encoding unit 54 encodes a transparency signal representing the transparency of the arbitrarily shaped object image. The transparency signal is also called alpha data. 0 means transparent, 255 means opaque 0
From 255 to 8 bits, one-to-one with each pixel of the luminance signal
Has a value corresponding to. The recording / transmission multiplexing unit 55 multiplexes the data encoded by each encoding unit into a stream. These constitute an object image encoding device.

If the pixel value of the background is Yb, the pixel value of the object is Yf, and the transparency of the object is Gf, the pixel value Yc of the pixel obtained by combining the object with the background is Yc = (255−Gf) × Yb + Gf during decoding. × Yf. When the encoding of the transmittance signal is omitted, the value of the shape signal is used as it is as the value of the transmittance signal.

When the shape signal encoding unit 53 encodes a shape signal of an arbitrary shape object, the block is completely VOP (ViP).
deo Object Plane) outside (the outside of the object), block is completely inside the VOP (inside the object), and block is classified into three types on the VOP boundary. In the example shown in FIG. 7, A indicates that the block is completely VO
Outside P, B is completely VOP in block, C is block in VOP
It is an example classified as being on a boundary line. The VOP inside / outside information is also used when decoding a pixel value signal of an arbitrary-shaped object.

FIG. 6 shows a block diagram of a pixel value signal encoding unit 52 using the conventional method (MPEG). First, intra-frame (intra) coding will be described. The input pixel value signal is divided into macro blocks, which are basic processing units.
One macroblock is composed of six blocks in the case of the 4: 2: 0 image format. Each block is subjected to DCT by a discrete cosine transform (DCT) unit 63 and then quantized by a quantization unit 64. The quantization width at the time of quantization is determined by the code amount control means 65. In addition, the quantized DC
The T coefficient and the quantization width are encoded by the variable length encoding means 66.

[0007] On the other hand, inter-picture coding uses motion vector detection such as block matching from another picture temporally adjacent to a picture containing a macroblock to be coded. By the method, the motion vector detecting means 61 detects a predicted macroblock having the smallest error with respect to the current macroblock. The signal indicating the motion to the predicted macroblock with the smallest error is the motion vector. An image referred to for generating a predicted macroblock is called a reference image. Based on the detected motion vector, the reference image is motion-compensated, and the motion compensation unit 62 obtains an optimal predicted macroblock. Next, the difference between the target macroblock and the corresponding predicted macroblock is determined, DCT is applied to the difference signal by the discrete cosine transform (DCT) means 63, and the DCT transform coefficient is quantized by the quantization means 64. . The quantized DCT coefficients are variable-length coded by the variable-length coding means 66 together with the motion vector and the quantization width.

[0008] An image in which all blocks are intra-coded is an I-VOP, an image predicted only from a VOP that precedes it in the display order of the image is a P-VOP, and an image that precedes and follows the image in the display order of the image is displayed. An image predicted from both is called a B-VOP. The I-VOP and the P-VOP may be reference images for encoding the subsequent image in the encoding order. Therefore, the quantized DCT coefficient is inversely quantized by the inverse quantization means 67, and inverse DCT is performed by the inverse DCT means 68. When intra-coded, the decoded image is stored in the memory 69, and is used as a reference image for coding the subsequent image. In the case of inter-coding, a decoded image is created by adding the motion-compensated image and stored in a memory.

[0009]

In a conventional object encoding apparatus, all background objects are encoded regardless of the presence, absence, shape, and size of an arbitrary shape object, so that the object is hidden by the arbitrary shape object in the background object. All the parts (the parts that cannot be seen on the display screen) are also encoded. Therefore, in a sequence in which an arbitrary-shaped object exists, even unnecessary data (that is, a portion hidden by the arbitrary-shaped object in the background object) is encoded, and there is a problem that the information amount increases. When encoding a background object, the present invention
The location (block) of the background object hidden by the arbitrary shape object based on the information of the arbitrary shape object
It is an object of the present invention to provide an object encoding device capable of reducing the amount of code of the object.

[0010]

In order to solve the above-mentioned problems, the present invention encodes at least two pieces of object information of an arbitrary-shaped object image and a background object image using DCT (Discrete Cosine Transform). In the object encoding apparatus, a block of a background object image corresponding to a position of a block determined to be a block in a VOP (Video Object Plane) of an arbitrarily shaped object image is a block completely hidden by the arbitrarily shaped object image. DCT of the block of the background object image determined to be present and determined to be a block completely hidden by this arbitrary shape object image
Do not encode the coefficients or D in the block
An object encoding apparatus is provided which encodes a DC coefficient in a CT coefficient and does not encode an AC coefficient, or includes means for increasing a quantization width of a block than usual.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the object encoding apparatus according to the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a first embodiment of an object encoding device, and FIG. 2 is a block diagram illustrating a configuration of a pixel value signal encoding unit 11 in the first embodiment. The difference from the conventional example is that the shape signal is input to the pixel value signal encoding unit 11 in FIG. 1 and that the VOP inside / outside determination unit 30 is added to the pixel value signal encoding unit in FIG. Since the configuration and operation other than the pixel value signal encoding unit 11 are almost the same as those of the related art, the description will be focused on the pixel value signal encoding unit 11 here. (Note that the pixel value signal encoding unit 11
The operations of the blocks 21 to 29 other than the DCT means 23 and the VOP inside / outside determination means 30 are the same as those of the blocks 61 to 69 shown in FIG. 6, and the description thereof is omitted here. )

First, when encoding a rectangular (background) object image, a rectangular (background) object image is encoded based on the shape signal of the arbitrary-shaped object.
The VOP inside / outside determination means 30 classifies the blocks of the arbitrary shape object into three types completely outside the VOP (Video Object Plane) (outside the object), the block completely inside the VOP (inside the object), and the block on the VOP boundary. judge. FIG. 7A illustrates an example in which a block is classified as completely outside the VOP, B illustrates an example in which the block is completely within the VOP, and C illustrates an example in which the block is classified as being on the VOP boundary line. Block A
And in the case of block C, the pixel of the background object corresponding to that position is displayed by the display device when decoded. However, the pixels of the background object corresponding to block B are not displayed. Assuming that the pixel value of the background is Yb, the pixel value of the object is Yf, and the transmittance of the object is Gf, the pixel value Yc of the pixel obtained by combining the object with the background is Yc = (255−Gf) × Yb + Gf × Yf at the time of decoding. In the case of block B, Gf is 255, which is equivalent to Yc = Yf. When decoding the block B, the background pixel value Yb is unnecessary.

Therefore, when the background object image is encoded, the VOP inside / outside determination means 30 determines the block of the background object image corresponding to the position of the block (ie, block B) determined to be the block in the VOP of the arbitrarily shaped object image. Is a block completely hidden by the arbitrary-shaped object image. Then, pixels corresponding to blocks of the background object image determined to be blocks completely hidden by the arbitrary-shaped object image do not encode DCT coefficients. That is, D
The DCT processing is not performed by the CT means 23. Alternatively, the DC coefficient in the DCT coefficient of the pixel corresponding to the block is encoded and A
Do not encode C coefficients. Thereby, when encoding the background object, the code amount of the background object block hidden by the arbitrary shape object can be reduced based on the information of the arbitrary shape object. In addition, VOP
Instead of the VOP inside / outside determination result of the inside / outside determination means 30, a result of coordinate conversion of the VOP inside / outside determination result obtained by the shape signal encoding unit 12 may be used.

Next, FIG. 3 shows the configuration of a pixel value signal encoding unit according to a second embodiment. The first embodiment corresponds to the VOP in FIG.
The difference is that the information determined by the inside / outside determination means 40 is input to the code amount control means 35. The blocks other than the VOP inside / outside determination means 40 and the code amount control means 35 are the same as the conventional ones.

When encoding a background object image, VO
The P inside / outside determination means 40 performs the VOP inside / outside determination result as in the first embodiment. Then, for a pixel corresponding to a block of the background object image determined to be a block completely hidden by the arbitrary-shaped object image,
In the code amount control means 35, the quantization width is increased as compared with the pixels in other blocks. Thereby, when encoding the background object, based on the information of the arbitrary shape object,
It is possible to reduce the code amount of the background object block hidden by the arbitrary-shaped object.

Next, FIG. 4 shows the configuration of a pixel value signal encoding unit according to a third embodiment. The difference from the first and second embodiments is that a reference image determining means 51 is added. Reference image determination means
According to 51, it is determined whether or not the frame of the background object image that is currently being encoded becomes a motion compensation reference image of a subsequent image in the encoding order. FIG. 8 is a diagram showing, in display order, whether or not to become a motion compensation reference image of a subsequent image in encoding order.
The white VOP serves as a motion-compensated reference image for the subsequent image, and the gray VOP is not used as a motion-compensated reference image for the subsequent image.
In the case of encoding all using the I-VOP, and in the case of the B-VOP, it does not become the motion compensation reference image of the subsequent image. But I-VO
When P, P-VOP, and B-VOP are mixed, the I-VOP and P-VOP may or may not be a reference image of a subsequent image depending on the type of image in the display order. The image immediately after in the display order is P-VO
In the case of P or B-VOP, it may be a reference image, and when the immediately preceding image is a B-VOP, it is also a reference image.

In the case of a frame of a background object image serving as a reference image of a later image in the coding order, the frame is determined to be block B by the VOP inside / outside determination means 50 (when the frame is completely concealed by an arbitrary shape object image). Even in the case of the (determined) block, since there is a possibility that this block will be referred to at the time of motion compensation of the subsequent image, ordinary coding is performed in the same manner as blocks A and C.

For the pixels of the block determined to be block B in the frame of the background object image that is not the motion compensation reference image, the code amount control means 45 increases the quantization width as compared with the pixels of the other blocks. Thereby, when encoding the background object, the code amount of the background object block hidden by the arbitrary shape object can be reduced based on the information of the arbitrary shape object. Hereinafter, the same processing as in the second embodiment is performed. Note that, instead of increasing the quantization width, the pixel of the block whose quantization width is to be increased does not encode the DCT coefficient, or encodes the DC coefficient in the DCT coefficient and does not encode the AC coefficient. It may be.

Here, there are many cases where the block boundaries at the time of encoding an object image of an arbitrary shape do not coincide with the block boundaries at the time of encoding a background object image. Therefore, as another embodiment, means for matching a block boundary at the time of encoding an arbitrary shape object image with a block boundary at the time of encoding a background object image,
The block of the background object image at the position of the block determined to be a block in the VOP is determined to be a block completely hidden by the arbitrary-shaped object image, and is a block completely hidden by this arbitrary-shaped object image. Do not encode the DCT coefficient of the block of the background object image determined to be, or encode the DC coefficient in the DCT coefficient of the block and do not encode the AC coefficient, or set the quantization width of the block larger than normal The object encoding device may be provided with means for performing the above.

Further, in the first to third embodiments, when there are two or more arbitrarily-shaped objects and the sequence is switched by a user instruction or the like on the decoding side, the size and shape of the switched image may be different. In this case, since the image of the background object is coded so that the image is normally displayed even after switching, when the VOP inside / outside determination means 30 (40) makes a determination, the determination is made for each arbitrary-shaped object, and any one of the objects is determined. A block determined to be block A or block C is treated as blocks A and C even in an arbitrary-shaped object. Only blocks determined to be block B in all arbitrary shape objects are treated as blocks B for which the code amount is reduced.

[0021]

As described above, the object encoding apparatus according to the present invention can reduce the code amount of a block completely concealed by an arbitrary-shaped object at the time of encoding a background object. This has the effect of reducing the overall code amount.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a first embodiment.

FIG. 2 is a configuration diagram of a pixel value signal encoding unit according to the first embodiment.

FIG. 3 is a configuration diagram of a pixel value signal encoding unit according to a second embodiment.

FIG. 4 is a configuration diagram of a pixel value signal encoding unit according to a third embodiment.

FIG. 5 is a configuration diagram showing a conventional example of an object encoding device.

FIG. 6 is a configuration diagram illustrating a conventional example of a pixel value signal encoding unit.

FIG. 7 is a diagram illustrating an example of a pixel value signal and a shape signal of an arbitrary shape object.

FIG. 8 is an explanatory diagram showing whether or not a subsequent image in a coding order will be a motion compensation reference image.

[Explanation of symbols]

 11 pixel value signal encoding unit 12 shape signal encoding unit 13 transmittance signal encoding unit 14 recording / transmission multiplexing unit

Claims (5)

[Claims] An object encoding apparatus for encoding at least two pieces of object information of an arbitrarily shaped object image and a background object image by using DCT (Discrete Cosine Transform). Plan
The block of the background object image corresponding to the position of the block determined to be the block in e) is determined to be a block completely concealed by the arbitrarily shaped object image, and is completely concealed by the arbitrarily shaped object image. An object coding apparatus, comprising: means for not coding DCT coefficients of a block of a background object image determined to be a block. 2. An object encoding apparatus for encoding at least two or more pieces of object information of an arbitrarily-shaped object image and a background object image using DCT (Discrete Cosine Transform). Plan
The block of the background object image corresponding to the position of the block determined to be the block in e) is determined to be a block completely concealed by the arbitrarily shaped object image, and is completely concealed by the arbitrarily shaped object image. An object encoding apparatus, comprising: means for encoding a DC coefficient in a DCT coefficient in a block of a background object image determined to be a block and not encoding an AC coefficient. 3. An object encoding apparatus for encoding at least two or more pieces of object information of an arbitrary-shaped object image and a background object image by using DCT (Discrete Cosine Transform). Plan
The block of the background object image corresponding to the position of the block determined to be the block in e) is determined to be a block completely concealed by the arbitrarily shaped object image, and is completely concealed by the arbitrarily shaped object image. An object coding apparatus, comprising: means for increasing a quantization width of a block of a background object image determined to be a block, as compared with a normal range. 4. An object encoding apparatus for encoding at least two or more pieces of object information of an arbitrary-shaped object image and a background object image by using DCT (Discrete Cosine Transform). Means for matching the boundary with the block boundary at the time of encoding the background object image, and a VOP (Video Object Plan)
e) It is determined that the block of the background object image at the position of the block determined as the block in the block is a block completely hidden by the arbitrarily shaped object image, and a block completely concealed by the arbitrarily shaped object image. Do not encode the DCT coefficients of the block of the background object image determined to be present, or encode the DC coefficients within the DCT coefficients of the block and do not encode the AC coefficients, or set the quantization width of the block to be smaller than normal. An object encoding device, comprising: means for increasing the size. 5. An object encoding apparatus which encodes at least two or more pieces of object information of an arbitrary-shaped object image and a background object image by using DCT (Discrete Cosine Transform), wherein a frame of the background object image is encoded in an encoding order. Determining means for determining whether or not the frame is a motion-compensated reference image of a frame to be encoded later, detecting means for detecting a block of a background object image completely hidden by an arbitrary-shaped object image, and a determination result of the determining means And encoding a DCT coefficient of a block completely concealed by an arbitrary-shaped object image in a frame of a background object image that is not a motion-compensation reference image of a frame to be encoded later in the encoding order based on the detection result of the detection unit. No, or DCT in the block A encodes the DC coefficient within a few
Means for encoding the C coefficient or increasing a quantization width of the block larger than usual.