KR20130079090A

KR20130079090A - Deep map transmitting apparatus and method for skipping dct and deep map reproduction apparatus and method for skipping dct

Info

Publication number: KR20130079090A
Application number: KR1020120053223A
Authority: KR
Inventors: 위호천; 이재준; 이천; 호요성
Original assignee: 삼성전자주식회사; 광주과학기술원
Priority date: 2011-12-30
Filing date: 2012-05-18
Publication date: 2013-07-10

Abstract

Disclosed are an image encoding apparatus and method, and an image decoding apparatus and method that do not perform a discrete cosine transform (DCT) transformation. The apparatus for encoding an image may reduce the bit size to be encoded without performing DCT transformation by grouping the residual information and encoding the residual information by using the representative value of each group and the group identification information for identifying the group.

Description

DEP MAP TRANSMITTING APPARATUS AND METHOD FOR SKIPPING DCT AND DEEP MAP REPRODUCTION APPARATUS AND METHOD FOR SKIPPING DCT}

Embodiments of the present invention relate to an apparatus and method for encoding an image and an apparatus and method for decoding an image. The present invention relates to an apparatus and method for encoding and decoding an image without a discrete cosine transform (DCT) transformation process.

The depth image is a color image representing distance information of an object and a camera to be captured in the form of a color image, and is used as auxiliary data for synthesizing a color image of a virtual viewpoint in various 3D video applications.

The conventional DCT / quantization based image compression method used as a method of encoding a depth image has a problem in that loss of high frequency components of a boundary line causes deterioration of image quality of a synthesized image. In addition, performing the DCT transform and then performing quantization according to the distribution of components included in the residual information may increase the size of the encoded information than directly quantize the residual information.

Therefore, there is a need for a method of encoding an image by an encoding method optimized according to the extent to which information loss due to direct quantization of residual information affects the image quality of image synthesis.

An image encoding apparatus according to an embodiment of the present invention includes a residual information classification unit for classifying residual data, which is a difference between an original image and a predicted image, into a plurality of groups; Representative value determination unit for determining a representative value that can represent the value of the remaining information included in the classified groups; A group identification information determining unit which determines group identification information identifying the groups from the remaining information; And an encoding unit encoding the representative values of the groups and the group identification information.

In the apparatus for encoding an image according to an embodiment of the present invention, the group identification information may be information representing values of elements included in the residual information in a map form for identifying a group corresponding to each element.

The encoder of the image encoding apparatus according to an embodiment of the present invention may quantize representative values of groups, and encode quantized representative values and group identification information.

An image encoding apparatus according to an embodiment of the present invention may include a first encoding cost determiner configured to determine a cost of a first encoding method for performing a discrete cosine transform (DCT) transformation; A second encoding cost determiner that determines a cost of a second encoding method that does not perform a DCT transform; An encoding method selection unit for selecting one of the first encoding method and the second encoding method by comparing the cost of the first encoding method with the cost of the second encoding method; And an image encoding apparatus encoding the original image using the selected encoding method.

The encoding method selection unit of the image encoding apparatus according to an embodiment of the present invention may transmit encoding identification information indicating the encoding method selected from the first encoding method or the second encoding method to the image decoding apparatus.

An apparatus for decoding an image according to an embodiment of the present invention includes: a representative value decoder which restores representative values of groups by decoding encoded information encoded by an encoding method that does not perform DCT transformation; A group identification information decoder which decodes the encoded information encoded by an encoding method that does not perform DCT transform, and restores the group identification information; A residual information determination unit for reconstructing residual data which is a difference between the original image and the predicted image using the representative values of the reconstructed groups and the group identification information; And an image generator for restoring the original image using the predicted image and the reconstructed residual information.

The residual information determiner of the image decoding apparatus according to an embodiment of the present invention may determine the residual information by converting the identification information corresponding to each group from the group identification information into representative values of the decoded groups.

The representative value decoder of the image decoding apparatus according to an embodiment of the present invention may decode information encoded by the representative value of the encoded information by using entropy decoding, and inversely quantize the decoded information to restore the representative value of each group. Can be.

An image encoding method according to an embodiment of the present invention includes the steps of: classifying residual data, which is a difference between an original image and a predicted image, into a plurality of groups; Determining a representative value that can represent a value of residual information included in the classified groups; Determining group identification information identifying the groups in residual information; And encoding representative values of the groups and group identification information.

An image encoding method according to an embodiment of the present invention includes determining a cost of a first encoding method for performing a discrete cosine transform (DCT) transformation; Determining a cost of a second coding method that does not perform DCT transformation; Selecting one of the first encoding method and the second encoding method by comparing the cost of the first encoding method with the cost of the second encoding method; And encoding the original image using the selected encoding method.

An image decoding method according to an embodiment of the present invention includes the steps of: reconstructing a representative value of groups by decoding encoded information encoded by an encoding method that does not perform DCT transformation; Restoring group identification information by decoding encoded information encoded by an encoding method that does not perform DCT transform; Restoring residual data, which is a difference between the original image and the predicted image, using the representative values of the decoded groups and the group identification information; And reconstructing the original image using the predicted image and the reconstructed residual information.

According to an embodiment of the present invention, the bit size to be encoded may be reduced by grouping the residual information and encoding the residual information by using the representative value of each group and the group identification information for identifying the group.

In addition, according to an embodiment of the present invention, a first encoding method for identifying a block for performing DCT transformation on residual information in an image, and performing DCT transformation according to the identification result, and second encoding without performing DCT transformation By encoding the original image using one of the methods, the encoding may be performed by an optimized method according to the block of the image.

1 is a diagram illustrating a structure of an image encoding apparatus according to an embodiment of the present invention.
2 is a diagram illustrating a detailed structure of a second encoder according to an embodiment of the present invention.
3 is a diagram illustrating a detailed structure of an encoding cost determiner according to an embodiment of the present invention.
4 is an example of an image encoding apparatus according to an embodiment of the present invention.
5 is an example of an operation of a second encoder according to an embodiment of the present invention.
6 is an example of generating a representative value and group identification information according to an embodiment of the present invention.
7 is a diagram illustrating a structure of an image decoding apparatus according to an embodiment of the present invention.
8 is a diagram illustrating a detailed structure of a second decoder according to an embodiment of the present invention.
9 is a diagram illustrating a video encoding method according to an embodiment of the present invention.
10 is a diagram illustrating an image decoding method according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a structure of an image encoding apparatus according to an embodiment of the present invention.

The image encoding apparatus 100 according to an embodiment of the present invention identifies a block that needs to perform discrete cosine transform (DCT) on residual data in an image, and performs DCT transformation according to the identification result. The original image may be encoded by using one of a first encoding method and a second encoding method that does not perform DCT transformation.

Referring to FIG. 1, the image encoding apparatus 100 may include an image predictor 110, an encoding cost determiner 120, an encoding method selector 130, a first encoder 140, and a second encoder 150. ).

The image predictor 110 may determine residual data by predicting a value of a next frame of the original image. In detail, the image predictor 110 generates a predicted image by predicting a value of a next frame of the original image in at least one of prediction modes, and compares the predicted image and the original image to provide residual information that is a difference between the predicted image and the original image. You can decide. In this case, the prediction mode is a macroblock of various sizes, and each macroblock may include at least one of an intra mode or an inter mode.

In addition, the image predictor 110 may select an optimal prediction mode to use when generating a prediction image from among prediction modes, and generate the prediction image using the optimal prediction mode.

The encoding cost determiner 120 uses the first encoding cost, which is a cost when the original video is encoded by the first encoding method using the bit rate distortion optimization method, and the second encoding method. A second encoding cost, which is a cost, may be determined.

A detailed process of determining the first encoding cost and the second encoding cost by the encoding cost determiner 120 will be described in detail with reference to FIG. 3.

The encoding method selector 130 may compare the first encoding cost and the second encoding cost, and select one of the first encoding method and the second encoding method according to the comparison result. In detail, the encoding method selector 130 may select an encoding method corresponding to a cost having a small value among the first encoding cost and the second encoding cost. For example, when the first encoding cost is smaller than the second encoding cost, the encoding method selector 130 may select the first encoding method.

In this case, the encoding method selection unit 130 transmits the residual information to the first encoding unit 140 when the first encoding method is selected, and transmits the residual information to the second encoding unit 150 when the second encoding method is selected. Can be sent to.

In addition, if the optimal prediction mode selected by the image predictor 110 is not a 4 × 4 block, the encoding method selector 130 skips the operation of the encoding cost determiner 120 and selects the first encoding method. The residual information may be transmitted to the first encoder 140.

In this case, since the 4 × 4 block has a quantization coefficient when encoding, a coding gain may be obtained using the second encoding method. However, another block without the quantization coefficient is a simpler region than the 4 × 4 block. It may be difficult to obtain an encoding gain by an encoding method.

Accordingly, the encoding method selector 130 determines whether the optimal prediction mode selected by the image predictor 110 is a 4 × 4 block, and the optimal prediction mode selected by the image predictor 110 is a 4 × 4 block. In this case, the encoding method selection speed may be improved by directly transmitting the remaining information to the first encoder 140 without operating the encoding cost determiner 120.

In addition, the encoding method selection unit 130 generates encoding identification information for identifying the encoding method selected from the first encoding method or the second encoding method, and the first encoding unit 140 or the second encoding unit 150 Before encoding the residual information, the encoding identification information may be transmitted to the image decoding apparatus in macroblock units.

For example, the encoding method selector 130 may define encoding identification information as 'NoDCT_flag', which is binary information consisting of 0's and 1's. In this case, when the first encoding method is selected, the encoding method selection unit 130 may determine 'NoDCT_flag' as 0, and when selecting the second encoding method, the encoding method selector 130 as 1.

The first encoder 140 may encode the residual information by using a first encoding method of performing DCT transformation on the residual information.

In detail, the first encoder 140 may perform DCT transformation on the residual information to convert the residual information into the frequency domain. In detail, the first encoder 140 may express the residual information having high repeatability spatially by DCT, and may express it as several coefficients in the frequency domain.

Next, the first encoder 140 may encode residual information converted into the frequency domain. In detail, the first encoder 140 may encode residual information transformed into coefficients according to a DCT transform by using quantization and entropy encoding.

The second encoder 150 may encode residual information by using a second encoding method that does not perform DCT transformation.

In detail, the second encoder 150 may classify the residual information into a plurality of groups and determine a representative value of each of the classified groups. In addition, the second encoder 150 may generate group identification information for identifying the classified groups.

In this case, the representative value is a value capable of representing the corresponding values based on the values of the information included in the classified group. For example, the second encoder 150 may determine an average value of the information included in each group as a representative value.

In addition, the group identification information may be information representing values of components included in the residual information in the form of a map capable of identifying a group corresponding to each component. In this case, the group identification information may be identified using a binary number or an integer. For example, the second encoder 150 may correspond to the residual information and determine the group identification information in the form of a binary map. At this time, 0 is a value representing the first group among the groups classified by the residual information classifying unit 210 among the components of the binary map composed of 0 and 1, and 1 is the first of the groups classified by the residual information classifying unit 210. It may be a value representing 2 groups. In addition, 0 may match the representative value of the first group, and 1 may match the representative value of the second group.

The second encoder 150 may encode residual information with fewer bits than the first encoding method by encoding the representative value and the group identification information of each classified group.

For example, when the remaining information consists of 10 '0' and 6 '7', the first encoder 140 may have 10 '0' and 6 's according to the arrangement order of' 0 'and' 7 '. You must encode '7'. In this case, since one bit is required to represent '0' and three bits are required to represent '7', the remaining information to be encoded is 53 bits.

On the other hand, the second encoder 150 may classify '0' and '7' into respective groups and generate group identification information including an arrangement order of the '0' and '7' groups. In this case, the group identification information may be information indicating which group of each component of the residual information corresponds to a group '0' and a group '7'. That is, by selecting one of 'yes' and 'no', it is possible to indicate which group belongs to the '0' group and the '7' group. Therefore, the bit size of the group identification information is 16 bits, which is the number of components. do.

In addition, since the '0' group and the '7' group are groups in which '0' and '7' are component values of residual information, the representative values may be '0' and '7', respectively. That is, the size of the representative value of each group is 1 bit and 3 bits.

Therefore, since the second encoder 150 encodes the 16-bit group identification information, the representative value of the 1-bit '0' group and the 3-bit '7' group representative value, the second encoder 150 encodes 20 bits of information.

That is, the second encoder 150 according to the present invention groups the residual information and encodes the residual information by using the representative value of each group and the group identification information for identifying the group, thereby reducing the bit size to be encoded. Can be.

In addition, the image encoder 100 according to the present invention identifies a block to which DCT transformation should be performed on residual information in an image, and performs a first encoding method for performing DCT transformation according to the identification result and a non-DCT transformation. By encoding the original image using one of two encoding methods, the encoding may be performed by an optimized method according to a block of the image.

2 is a diagram illustrating a detailed structure of a second encoder according to an embodiment of the present invention.

Referring to FIG. 2, the second encoder 150 may include a residual information classifier 210, a representative value determiner 220, a group identification information determiner 230, and an encoder 240.

The residual information classifying unit 210 may classify the prediction image generated by the image predicting unit 110 by predicting the original image of the next frame and the residual information that is a difference between the original image of the next frame into a plurality of groups. Specifically, the residual information classifier 210 stores the residual information received from the image predictor 110 in units of macroblocks, and uses a clustering method such as K-means or an image segmentation method such as mean-shift. The stored residual information may be classified into a plurality of groups.

In this case, since the residual information of the depth image of the image is composed of most '0' and a few residual values, the residual information classifying unit 210 may classify the residual information of the depth image into two groups.

The representative value determiner 220 may determine representative values of groups classified by the residual information classifier 210. For example, the representative value determiner 220 may determine the average value of the information included in the group as the representative value. In addition, when one group includes information having a plurality of values, the representative value may be determined according to the number of information having the same value among the information included in the group.

That is, the representative value determiner 220 may determine a representative value that can represent the corresponding values based on the values of the information included in the group classified by the residual information classifier 210.

The group identification information determiner 230 may determine group identification information identifying the groups classified by the residual information classifier 210.

In this case, the group identification information may be information representing values of components included in the remaining information in a map form for identifying a group corresponding to each component. For example, the group identification information determiner 230 may determine the group identification information in the form of a binary map corresponding to the residual information. In this case, 0 may indicate a first group among the groups classified by the residual information classifying unit 210, and 1 may indicate a second group among the groups classified by the residual information classifying unit 210. In addition, 0 may match the representative value of the first group, and 1 may match the representative value of the second group.

The relationship between the representative value determined by the representative value determiner 220 and the group identification information determined by the group identification information determiner 230 and the remaining information will be described in detail with reference to FIG. 6.

In addition, the group identification information determiner 230 may generate an index related to the format of the group identification information and transmit the index to the encoder 240. In this case, the group identification information determiner 230 may preset the format of the group identification information and generate an index related to the set format of the group identification information.

The encoder 240 may encode the representative value determined by the representative value determiner 220 and the group identification information determined by the group identification information determiner 230. In this case, the encoder 240 may encode the representative value and the group identification information by using context-adaptive binary arithmetic coding (CABAC). In addition, when the representative value is a signed integer, the encoder 240 may encode the representative value by the same method as that of encoding the motion vector.

In this case, the encoder 240 may quantize representative values of groups without using entropy encoding to maintain loss encoding, and then encode the entropy encoding. For example, the encoder 240 may quantize the representative values using Equation 1.

In this case, Z _i is a quantized representative value, and m _i may be each group. For example, m ₁ may be a first group among the groups classified by the residual information classifier 210, and m ₂ may be a second group among the groups classified by the residual information classifier 210. In addition, Q _step may be a value used in H.264 / AVC as shown in Table 1. Ε may be a constant for adjusting the Q _step .

In this case, QP may be a quantization parameter.

In addition, the encoder 240 may perform lossless encoding by setting the denominator of Equation 1 to one.

In addition, when the encoder 240 receives an index from the group identification information determiner 230, the encoder 240 may encode the group identification information using the received index.

3 is a diagram illustrating a detailed structure of an encoding cost determiner according to an embodiment of the present invention.

Referring to FIG. 3, the encoding cost determiner 120 may include a first encoding cost determiner 310 and a second encoding cost determiner 320.

The first encoding cost determiner 310 may determine the cost of the first encoding method for performing the DCT transform by using rate-distortion optimization.

For example, the first encoding cost determiner 310 may determine J _mode which is the cost of the first encoding method by using Equation 2.

In this case, D may be a distortion ratio between the reconstructed image and the original image, and R may be a bit amount of the original image obtained by encoding one macroblock by the first encoding method. In this case, the reconstructed image may be a reconstructed image using the first reconstructed image obtained by decoding the residual information encoded by the first encoding method.

In addition, s may be original information of an image, r may be a first reconstructed image, and M may be header information including mode type information used by the image predictor 110 to generate a predictive image. C may be a DCT transformed quantization coefficient. In this case, the mode type included in M may be a best mode.

The second encoding cost determiner 320 may determine the cost of the second encoding method that does not perform the DCT transform by using the bit rate-distortion optimization.

For example, the second encoding cost determiner 320 may determine J _NoDCT _{_} _mode , which is the cost of the second encoding method, using Equation 3.

In this case, D may be a distortion ratio between the reconstructed image and the original image, and R may be a bit amount of the original image obtained by encoding one macroblock by the second encoding method. In this case, the reconstructed image may be a reconstructed image using the second reconstructed image obtained by decoding the residual information encoded by the second encoding method.

Also,

Is the second reconstructed image, m ₁ and m ₂ are the residual information classifier 210 as representative values of the first group and the second group, respectively, and P identifies the arrangement of the first group and the second group in the residual information. It may be group identification information.

The encoding method selector 130 according to the present invention selects the first encoding method when the result value of Equation 2 is larger than the result value of Equation 3, and the result value of Equation 3 is the result value of Equation 2; If larger, the second encoding method can be selected.

4 is an example of an image encoding apparatus according to an embodiment of the present invention.

The image predicting unit 110 of the image encoding apparatus according to an embodiment of the present invention may determine residual information by predicting a value of a next frame of an image. In detail, the image predictor 110 is a next frame P of an image predicted using a mode selected as an optimal mode among various blocks of an intra mode or an inter mode and an original image of the frame. Residual information that is the difference between DFn can be determined.

The encoding method selector 130 compares the first encoding cost and the second encoding cost determined by the encoding cost determiner 120, and selects one of the first encoding method and the second encoding method according to the comparison result to switch the switch. Can be controlled.

For example, when the first encoding cost is greater than the second encoding cost, the encoding method selector 130 may connect a switch to the first encoding unit 140 to transmit the remaining information to the first encoding unit 140. . In addition, when the second encoding cost is greater than the first encoding cost, the encoding method selection unit 130 may connect a switch to the second encoding unit 150 to transmit the remaining information to the second encoding unit 140.

In this case, the first encoder 140 may convert the residual information into the frequency domain by performing DCT transformation on the residual information, and encode the residual information converted by the Q 412 into the frequency domain.

In addition, the second encoder 140 classifies the remaining information into a plurality of groups by the K-means 421 to determine a representative value of each group and group identification information identifying the groups, and Q 422. May encode a representative value of each group and group identification information identifying the groups. In this case, the K-means 421 is hardware including a residual information classifying unit 210, a representative value determining unit 220, and a group identification information determining unit 230, and the Q 422 is an encoding unit 240. May correspond to hardware. In addition, Q 422 may quantize the representative values and then encode the representative values.

In addition, the image encoding apparatus may decode the encoded information to perform the prediction of the image prediction unit 110 and the bit rate-distortion optimization required to determine the first encoding cost and the second encoding cost of the encoding cost determiner 120. Can be.

The second decoder 430 may determine the residual information encoded by the second encoder 150 and transmit the residual information to the image predictor 110 and the encoding cost determiner 120. In detail, the second decoder 430 restores the representative value and the group identification information by decoding the representative value and the group identification information of each group encoded by the Q ⁻¹ 432, and the K-means 431 restores the representative value and the group identification information. The residual information input to the second encoder 150 may be determined using the group identification information and the representative value. In this case, the reconstructed residual information is a second reconstructed image.

In this case, since the detailed configuration of the second decoder 430 is the same as that of the second decoder included in the image decoding apparatus, a decoding process corresponding to the second decoding method will be described in detail with reference to FIG. 8.

The first decoder 440 may determine the residual information encoded by the first encoder 140 and transmit the residual information to the image predictor 110 and the encoding cost determiner 120. In detail, the first decoder 440 decodes the residual information encoded by the Q- ¹ 442 into the frequency domain, and performs inverse DCT transform on the residual information decoded into the frequency domain by the T- ¹ 444. The residual information input to the first encoder 140 may be determined. In this case, the reconstructed residual information is the first reconstructed image.

The switch 450 receives the reconstructed residual information by connecting to the corresponding decoder according to the operation of the encoding method selector 130, and transmits the received residual information to the image predictor 110 and the encoding cost determiner 120. I can deliver it.

In this case, the image predictor 110 may restore the next frame of the image by combining P and the residual information. Also, the encoding cost determiner 120 may determine the first encoding cost and the second encoding cost by using a distortion ratio between the next frame of the reconstructed image and the next frame of the actual image.

5 is an example of an operation of a second encoder according to an embodiment of the present invention.

The K-means 421 of the second encoder 150 according to the present invention is a prediction image that is a result of prediction of the original image 510 of the corresponding frame and the image predictor 110 based on the original image of the previous frame. Residual information 530 that is the difference between 520 may be received.

In this case, the K-means 421 may classify the remaining information into two groups by using the K-means clustering method and determine the representative value of each group and the group identification information for identifying the groups.

Q 422 may then encode the representative value of each group and group identification information identifying the groups without DCT transformation. In this case, the Q 422 may quantize the representative values and then encode the representative values.

6 is an example of generating a representative value and group identification information according to an embodiment of the present invention.

When the residual information 610 is composed of ten '0's and six' -14's as shown in FIG. 6, the residual information classifying unit 210 uses the K-means clustering method to display '0' and ' 7 'may be classified into each of the first group and the second group.

Next, the representative value determiner 220 may determine the representative value 620 of each group. In the sixth example, since the values of the first group and the second group are unified to 0 and -14, respectively, the representative value determiner 220 sets 0, which is a common value of information included in the first groups, to the first group. The representative value m1 may be determined, and the representative value may be determined as −14, which is a common value of information included in the second group, and the representative value m2 of the second group. When there is a difference in the values classified into each group, the representative value determiner 220 may determine another value in consideration of the average value or the standard deviation as the representative value of each group.

Next, the group identification information generation unit 230 may generate group identification information 630 indicating which group of each component of the remaining information 610 corresponds to a group of the first group and the second group. In this case, the group identification information may be generated so that each group can be identified with the smallest number because only the components of the remaining information 610 correspond to which group of the first group and the second group. For example, if the number of groups is two, two groups can be identified by only 0 and 1, and thus group identification information can be generated using a binary map as shown in FIG.

In addition, when the residual information is to be determined, the representative value of the first group is input to the component corresponding to the first group in the group identification information 630, and the representative of the second group is input to the component corresponding to the second group. By entering a value, residual information can be determined. For example, in FIG. 6, since the representative value of the second group is -14 and the value of the component corresponding to the second group in the group identification information 630 is 1, the configuration indicated by 1 in the group identification information 630 is shown. By entering -14 in the element, residual information 610 can be recovered. In this case, since the representative value of the first group is 0 and the value of the component corresponding to the first group in the group identification information 630 is 0, no separate correction may be performed.

When the residual information 610 is encoded as it is, since -14 is included in six, the number of bits of the encoded information increases. On the other hand, in the second encoding method according to the present invention, as shown in FIG. 6, six 1s that can be represented by one bit and -14 are only ones included in the representative value can reduce the number of bits of the encoded information.

7 is a diagram illustrating a structure of an image decoding apparatus according to an embodiment of the present invention.

Referring to FIG. 7, the image decoding apparatus 700 includes a decoding method selector 710, a first decoder 720, a second decoder 730, and an image generator 740.

The decoding method selector 710 may select a method of decoding the encoded information received from the image encoding apparatus 100 according to the encoding identification information received from the encoding method selector 130.

In detail, when the encoding method selection unit 710 receives the encoding identification information corresponding to the first encoding method, the decoding method selection unit 710 transmits the encoding information received from the image encoding apparatus 100 to the first decoding unit 720. When encoding identification information corresponding to the second encoding method is received, the encoding information received from the image encoding apparatus 100 may be transmitted to the second decoder 730.

The first decoder 720 may determine the residual information from the encoding information received from the image encoding apparatus 100 using the first decoding method corresponding to the first encoding method. In detail, the first decoder 720 may decode the encoded information into the frequency domain and perform inverse DCT transformation on the information decoded into the frequency domain to determine the residual information input to the first encoder 140.

The second decoder 730 may determine the residual information from the encoding information received from the image encoding apparatus 100 using the second decoding method corresponding to the second encoding method. In detail, the second decoder 730 decodes the encoding information to restore the representative value and the group identification information of each group, and the residual inputted to the second encoder 150 using the reconstructed group identification information and the representative value. Information can be determined.

In this case, a detailed configuration of the second decoder 730 will be described in detail with reference to FIG. 8.

The image generator 740 generates a predictive image by predicting a value of a next frame of a previously reconstructed image, and reconstructs the predicted image and the residual information reconstructed by the first decoder 720 or the second decoder 730. Can be combined to restore the original image.

In this case, the image generator 740 may generate the predicted image by predicting the value of the next frame of the image in the same configuration as the image predictor 110. In detail, the image generator 740 selects an optimal mode from various blocks of the intra mode or the inter mode, and generates a predictive image predicting the next frame of the image using the selected mode. can do.

In this case, the remaining information is a difference between the next frame of the image predicted by the image predictor 110 and the next frame of the actual image. That is, since the predicted image generated by predicting the same configuration as the image predictor 110 and the next frame of the real image and other portions are included in the residual information, when the residual information is combined with the predicted image, the next frame of the actual image is obtained. Can be restored

For example, the decoding method selection unit 710 receives encoding identification information corresponding to the second encoding method, the second encoding method encodes the residual signal into two groups, and corresponds to the first group in the group identification information. If the value of the component is 1, the image generator 740 may reconstruct the original image using Equation 4.

At this time,

Is a representative value of the decoded first group,

May be a representative value of the decoded second group. In addition, E _i may be a predicted image generated by the image predictor 110, and P _i may be a value of each component included in the group identification information.

8 is a diagram illustrating a detailed structure of a second decoder according to an embodiment of the present invention.

Referring to FIG. 8, the second decoder 730 includes a representative value decoder 810, a group identification information decoder 820, and a residual information determiner 830.

The representative value decoder 810 may decode the representative value of each group by decoding the encoded information of the representative value among the encoding information received from the image encoding apparatus 100.

In this case, the representative value decoding unit 810 may decode the representative value of each group by performing inverse quantization after decoding information encoding the representative value by using entropy decoding. For example, the representative value decoder 810 may inverse quantize the representative value decoded using Equation 5.

At this time,

Is a representative value of each of the reconstructed groups, Z _i may be a representative value of each group decoded. In this case, Z _i may be equal to a representative value quantized by the encoder 240 of the image encoding apparatus 100. In addition, Q _step is a value used in H.264 / AVC as shown in Table 1, ε may be a constant for adjusting the Q _step .

The group identification information decoding unit 820 may decode the representative value of each group by decoding the information encoding the group identification information among the encoding information received from the image encoding apparatus 100. In this case, the group identification information decoder 820 may decode the group identification information encoded by using entropy decoding.

The residual information determiner 830 may determine the residual information using the representative value decoded by the representative value decoder 810 and the group identification information decoded by the group identification information decoder 820. Specifically, the residual information determiner 830 inputs a representative value of the first group to a component corresponding to the first group among the components of the group identification information, and inputs a representative value of the second group to a component corresponding to the second group. Residual information can be determined by entering a representative value.

9 is a diagram illustrating a video encoding method according to an embodiment of the present invention.

In operation S910, the encoding cost determiner 120 encodes the first encoding cost, which is a cost when the original image is encoded by the first encoding method using the bit rate distortion optimization method, and the second encoding of the original image. The second encoding cost, which is a cost in the case of encoding, can be determined.

In operation S920, the encoding method selector 130 may determine whether the first encoding cost determined in operation S910 is greater than the second encoding cost. In this case, when the first encoding cost is greater than the second encoding cost, the encoding method selecting unit 130 transmits the remaining information to the second encoding unit 150 to correspond to the second encoding method (S930) to ( S960) can be executed. In addition, when the first encoding cost is not greater than the second encoding cost, the encoding method selecting unit 130 transmits the remaining information to the first encoding unit 140 to correspond to the first encoding method (S970) to step. (S980) can be executed.

In this case, the encoding method selection unit 130 generates encoding identification information indicating the encoding method selected from the first encoding method or the second encoding method, and the first encoding unit 140 or the second encoding unit 150 displays the remaining information. Before encoding, the encoding identification information may be transmitted to the image decoding apparatus in macroblock units.

In operation S930, the second encoder 150 may classify the remaining information into a plurality of groups.

In operation S940, the second encoder 150 may determine a representative value that may represent values of residual information included in the classified groups in operation S930. For example, the second encoder 150 may determine an average value of the residual information included in the classified group as the representative value.

In operation S950, the second encoder 150 may determine group identification information for identifying groups classified from the remaining information. In this case, the group identification information may be information obtained by changing values of components included in the residual information into integers or binary numbers that can identify a group corresponding to each component.

In operation S960, the second encoding unit 150 encodes the representative value of each group determined in operation S940 and the group identification information generated in operation S950, thereby encoding residual information with fewer bits than the first encoding method. Can be encoded.

In operation S970, the first encoder 140 may perform DCT transformation on the residual information to convert the residual information into the frequency domain. In detail, the first encoder 140 may express the residual information having high repeatability spatially by DCT, and may express it as several coefficients in the frequency domain.

In operation S980, the first encoder 140 may encode residual information converted into the frequency domain in operation S970. In detail, the first encoder 140 may encode residual information transformed into coefficients according to a DCT transform by using quantization and entropy encoding.

10 is a diagram illustrating an image decoding method according to an embodiment of the present invention.

In operation S1010, the decoding method selector 710 may receive encoding identification information from the encoding method selector 130 of the image encoding apparatus 100.

In this case, the encoding identification information includes the first encoding method in which the encoding information received by the image decoding apparatus 700 performs the DCT transformation according to the identification result, and the second encoding method in which the DCT transformation is not performed. It may be information indicating what information is encoded by using an encoding method.

In operation S1020, the decoding method selector 710 may determine whether the encoding identification information received in operation S1010 is encoding identification information corresponding to the second encoding method.

In this case, when the encoding identification information received in operation S1010 is encoding identification information corresponding to the second encoding method, the decoding method selection unit 710 receives the encoding information received from the image encoding apparatus 100. In operation 730, steps S1030 to S1050 corresponding to the second encoding method may be performed. In this case, when the encoding identification information received in operation S1010 is not encoding identification information corresponding to the second encoding method, the decoding method selection unit 710 removes the encoding information received from the image encoding apparatus 100. The method may transmit the data to the first decoder 720 to perform operations S1070 to S1080 corresponding to the first encoding method.

In operation S1030, the second decoder 730 may decode information obtained by encoding a representative value from among encoding information received from the image encoding apparatus 100.

In this case, the representative value decoding unit 810 may decode the representative value of each group by performing inverse quantization after decoding information encoding the representative value by using entropy decoding.

In operation S1040, the second decoder 730 may decode the encoded information of the group identification information among the encoding information received from the image encoding apparatus 100. In this case, the group identification information decoder 820 may decode the information encoding the group identification information by using entropy decoding.

In operation S 1050, the second decoder 730 may determine the residual information by using the representative value decoded in operation S 1030 and the group identification information decoded in operation S 1040. Specifically, the second decoder 730 inputs a representative value of the first group to a component corresponding to the first group among the components of the group identification information, and inputs a representative value of the second group to a component corresponding to the second group. Residual information can be determined by entering a representative value.

In operation S1060, the image generator 740 may generate a prediction image by predicting a value of a next frame of the original image, and combine the prediction image with the residual information reconstructed in operation S1050, or operation S1080. The original image of the frame can be generated. In operation S1070, the first decoder 720 may decode the encoding information received from the image encoding apparatus 100 into a frequency domain.

In operation S1080, the first decoder 720 may perform inverse DCT transformation on the information decoded in the frequency domain in operation S1070 to determine the residual information input to the first encoder 140. In this case, the first decoder 720 may transfer the reconstructed residual information to the image generator 740 to generate an original image using the residual information and the predicted image (S1060).

The methods according to embodiments of the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer readable medium may include a program command, an information file, an information structure, etc. alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of computer software.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

100: image encoding apparatus 110: image prediction unit
120: encoding cost determination unit 130: encoding method selection unit
140: first encoder 150: second encoder
700: video decoding apparatus 710: decoding method selection unit
720: first decoder 730: second decoder
740: the image generating unit

Claims

A residual information classification unit classifying residual data, which is a difference between the original image and the predicted image, into a plurality of groups;
Representative value determination unit for determining a representative value that can represent the value of the remaining information included in the classified groups;
A group identification information determining unit which determines group identification information identifying the groups from the remaining information; And
An encoder for encoding the representative value of the groups and group identification information
Image encoding apparatus comprising a.

The method of claim 1,
The remaining information classification unit,
And the residual information is classified into a plurality of groups using a clustering method or an image segmentation method.

The method of claim 1,
Wherein the group identification information comprises:
And encoding the values of the elements included in the residual information in the form of a map capable of identifying a group corresponding to each element.

The method of claim 1,
The group identification information determiner
Generate indexes related to the format of group identification information,
Wherein the encoding unit comprises:
And the group identification information is encoded using the generated index.

The method of claim 1,
Wherein the encoding unit comprises:
And quantize the representative values of the groups, and encode the quantized representative values and group identification information.

A first encoding cost determiner configured to determine a cost of a first encoding method for performing discrete cosine transform (DCT) transformation;
A second encoding cost determiner configured to determine a cost of a second encoding method that does not perform a discrete cosine transform (DCT);
An encoding method selection unit for selecting one of the first encoding method and the second encoding method by comparing the cost of the first encoding method with the cost of the second encoding method; And
Image encoder to encode the original image using the selected encoding method
Image encoding apparatus comprising a.

The method according to claim 6,
The first encoding cost determiner,
The cost of the first encoding method is determined by using a distortion ratio between the first reconstructed image and the original image decoded by the first encoding method, and the bit amount of the original image encoded by the first encoding method. An image encoding device.

The method according to claim 6,
The image encoder,
When the second encoding method is selected, residual data, which is a difference between the original image and the predicted image, may be classified into a plurality of groups, and a representative value may represent a value of the residual information included in the classified groups. And encoding the group identification information identifying the groups from the residual information.

The method according to claim 6,
The second encoding cost determiner,
The cost of the second encoding method is determined by using a distortion ratio between the second reconstructed image and the original image decoded by the second encoding method, and the bit amount of the original image encoded by the second encoding method. An image encoding device.

10. The method of claim 9,
The second encoding cost determiner,
Encoding with a second encoding method using original information, a second reconstructed image, header information of a mode used to generate a predictive image, representative values of groups classified from residual data, and group identification information identifying the groups And calculating a bit amount of the original image.

10. The method of claim 9,
The encoding method selection unit,
And encoding identification information indicating the encoding method selected from the first encoding method and the second encoding method, to the image decoding apparatus.

A representative value decoder for decoding representative values of groups from encoded information encoded by an encoding method that does not perform a discrete cosine transform (DCT);
A group identification information decoder for decoding group identification information from encoded information encoded by an encoding method that does not perform a discrete cosine transform (DCT);
A residual information determiner configured to determine residual data, which is a difference between the original image and the predicted image, by using the representative values of the reconstructed groups and the group identification information; And
Image generator to generate the original image using the predicted image and the reconstructed residual information
And decodes the decoded image.

The method of claim 12,
The remaining information determination unit,
And residing information is determined by converting identification information corresponding to each group among the group identification information into representative values of the decoded groups.

The method of claim 12,
The encoding information is,
And a representative value of a plurality of groups classifying residual information that is a difference between an original image and a predicted image and group identification information for identifying the groups from the residual information.

The method of claim 12,
Wherein the group identification information comprises:
And the values of the components included in the residual information are information expressed in the form of a map capable of identifying a group corresponding to each component.

The method of claim 12,
The representative value decoding unit,
And decoding the representative value of each group by using the entropy decoding to decode the coded representative value of the encoded information, and to dequantize the decoded information.

The method of claim 12,
The group identification information decoding unit,
And an image obtained by encoding group identification information among encoded information using entropy decoding.

Classifying residual data, which is a difference between an original image and a predicted image, into a plurality of groups;
Determining a representative value that can represent a value of residual information included in the classified groups;
Determining group identification information identifying the groups in residual information; And
Encoding the representative value and group identification information of the groups.
Image encoding method comprising a.

19. The method of claim 18,
Wherein said classifying comprises:
And the residual information is classified into a plurality of groups using a clustering method or an image segmentation method.

19. The method of claim 18,
Wherein the group identification information comprises:
And encoding values of components included in the residual information in a map form for identifying a group corresponding to each component.

19. The method of claim 18,
Determining the group identification information
Generate indexes related to the format of group identification information,
Wherein the encoding comprises:
And the group identification information is encoded using the generated index.

19. The method of claim 18,
Wherein the encoding comprises:
Quantizing representative values of the groups; And
Encoding the quantized representative values and the group identification information
Image encoding method comprising a.

Determining a cost of the first encoding method for performing discrete cosine transform (DCT);
Determining a cost of a second coding method that does not perform discrete cosine transform (DCT);
Selecting one of the first encoding method and the second encoding method by comparing the cost of the first encoding method with the cost of the second encoding method; And
Encoding the original image using the selected encoding method
Image encoding method comprising a.

24. The method of claim 23,
Determining the cost of the first encoding method,
Determining the cost of the first encoding method using the distortion rate between the first reconstructed image generated by decoding the original image encoded by the first encoding method and the original image, and the bit amount of the original image encoded by the first encoding method. An image encoding method.

24. The method of claim 23,
Encoding the original image,
When the second encoding method is selected, residual data, which is a difference between the original image and the predicted image, may be classified into a plurality of groups, and a representative value may represent a value of the residual information included in the classified groups. And encoding group identification information for identifying the groups from the residual information.

24. The method of claim 23,
Determining the cost of the second encoding method,
Determining the cost of the second encoding method using the distortion ratio between the second reconstructed image generated by decoding the original image encoded by the second encoding method and the original image, and the bit amount of the original image encoded by the second encoding method. An image encoding method.

The method of claim 26,
Determining the cost of the second encoding method,
Encoding with a second encoding method using original information, a second reconstructed image, header information of a mode used to generate a predictive image, representative values of groups classified from residual data, and group identification information identifying the groups And calculating a bit amount of the original image.

24. The method of claim 23,
Transmitting encoding identification information indicating the encoding method selected from the first encoding method or the second encoding method.
Image encoding method further comprising.

Decoding representative values of groups from encoded information encoded by an encoding method that does not perform a discrete cosine transform (DCT);
Decoding group identification information from encoded information encoded by an encoding method that does not perform a discrete cosine transform (DCT);
Determining residual data, which is a difference between the original image and the predicted image, using the representative values of the decoded groups and the group identification information; And
Generating the original image using the predicted image and the reconstructed residual information
Image decoding method comprising a.

30. The method of claim 29,
Determining the remaining information,
And residing information is determined by converting identification information corresponding to each group among the group identification information into representative values of the decoded groups.

30. The method of claim 29,
The encoding information is,
And a representative value of a plurality of groups classifying the residual information, which is a difference between the predicted image and the original image, and the group identification information for identifying the groups in the residual information.

30. The method of claim 29,
Wherein the group identification information comprises:
And the values of the components included in the residual information are in a form of a map capable of identifying a group corresponding to each component.

30. The method of claim 29,
Decoding the representative value,
And decoding the representative value of each group by dequantizing the decoded information by using entropy decoding, and inversely quantizing the decoded information.

30. The method of claim 29,
Decoding the group identification information,
And decoding the encoded information of the group identification information among the encoded information by using entropy decoding.

A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 18 to 34.