KR20140129425A - Method for intra-prediction, and apparatus thereof - Google Patents

Abstract

The present invention relates to an intra-prediction apparatus and an intra-prediction method comprising: a step of determining the number (N) of filtering taps applied to each area by dividing an image to be predicted into multiple areas; a step of generating a prediction pixel by performing interpolation to the divided area using N-reference pixels containing two reference pixels adjacent to each other according to an intra-prediction direction; and a step of generating a residual signal using the generated prediction pixel.

Description

[0001] The present invention relates to intra prediction methods and apparatuses,

The present invention relates to an image encoding / decoding method and apparatus, and more particularly, to a method of performing intra prediction using interpolation filtering in encoding and decoding.

Generally, in video coding, intra prediction and inter prediction are used to generate a residual signal. The reason why the residual signal is obtained is that when the data is coded with the residual signal, the amount of data is small, so that the data compression rate is high, and the better the prediction, the smaller the value of the residual signal is.

The intraprediction method predicts the data of the current block by using the pixels around the current block. The difference between the actual value and the predicted value is called a residual signal block. In the case of HEVC, the intra prediction method is increased to 35 prediction modes as shown in FIG. 1 in nine prediction modes used in the existing H.264 / AVC, and is further segmented and predicted (the planar prediction mode and the DC prediction mode 1).

In the case of the inter prediction method, the current block is compared with the blocks in the neighboring pictures to find the closest block. At this time, the position information (Vx, Vy) of the found block is referred to as a motion vector. The difference between the intra-block pixel values of the current block and the prediction block predicted by the motion vector is called a residual-signal block (motion-compensated residual block).

In this way, intra prediction and inter prediction are further subdivided so that the amount of data of the residual signal is reduced, and a video coding and decoding method with a small amount of computation is required without degrading the codec performance using an efficient transform.

An embodiment of the present invention provides a video encoding and decoding method having good performance with a small amount of computation in a transcoding process of a video codec and an apparatus therefor.

It is to be understood, however, that the technical scope of the present invention is not limited to the above-described technical problems, and other technical problems may be present.

According to an exemplary embodiment of the present invention, there is provided an intra prediction method for dividing an image to be predicted into a plurality of regions and generating a plurality of filtering tap numbers N ); Generating prediction pixels by performing interpolation for each of the divided regions using N reference pixels including two adjacent pixels according to an intra prediction direction; And generating a residual signal using the generated predictive pixel.

Also, an intra prediction apparatus according to an embodiment of the present invention includes: a tap number determination unit for dividing an image to be predicted into a plurality of regions and determining a number of filtering taps N to be applied to each of the plurality of regions; A prediction pixel generation unit for generating prediction pixels by performing interpolation for each of the divided regions using N reference pixels including two adjacent pixel candidates according to an intra prediction direction; And a residual signal generator for generating a residual signal using the generated predictive pixel.

Meanwhile, the intraprediction method may be embodied as a computer-readable recording medium on which a program for execution in a computer is recorded.

According to the present invention, when a prediction for generating a residual signal is performed according to each prediction direction (mode), an optimum filtering method is determined to obtain a prediction value, thereby minimizing the size of the residual signal, Can be improved.

Also, when interpolation is performed in interpolation in various directions applied in the intra prediction method, when the interpolation is performed, the number of filtering taps used for interpolation is varied according to the change of a specific region in an image or a pixel change, Can be improved.

1 is a diagram showing examples of intra prediction modes.
FIG. 2 is a view for explaining an example of image encoding units.
3 is a diagram showing an example of an interpolation method performed for intra prediction.
4 is a diagram for explaining a first embodiment of an intra prediction method according to the present invention.
4 is a diagram for explaining a first embodiment of an intra prediction method according to the present invention.
5 is a diagram for explaining a second embodiment of the intra prediction method according to the present invention.
FIG. 6 is a diagram for explaining a third embodiment of the intra prediction method according to the present invention.
FIG. 7 is a diagram for explaining a fourth embodiment of the intra prediction method according to the present invention.
Figs. 8 and 9 are diagrams for explaining the operation of sync filtering. Fig.
10 to 12 are block diagrams illustrating embodiments of an arrangement for performing intraprediction by adjusting the number of filtering tabs.
13 and 14 are block diagrams illustrating embodiments of an arrangement for performing intraprediction by adjusting the number of filtering taps according to the conversion of pixel values.
15 is a block diagram showing a configuration of an encoding apparatus according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. It should be understood, however, that the present invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In the drawings, the same reference numbers are used throughout the specification to refer to the same or like parts.

Throughout this specification, when a part is referred to as being "connected" to another part, it is not limited to a case where it is "directly connected" but also includes the case where it is "electrically connected" do.

Throughout this specification, when a member is " on " another member, it includes not only when the member is in contact with the other member, but also when there is another member between the two members.

Throughout this specification, when an element is referred to as "including " an element, it is understood that the element may include other elements as well, without departing from the other elements unless specifically stated otherwise. The terms "about "," substantially ", etc. used to the extent that they are used throughout the specification are intended to be taken to mean the approximation of the manufacturing and material tolerances inherent in the stated sense, Accurate or absolute numbers are used to help prevent unauthorized exploitation by unauthorized intruders of the referenced disclosure. The word " step (or step) "or" step "used to the extent that it is used throughout the specification does not mean" step for.

Throughout this specification, the term " combination thereof " included in the expression of the machine form means one or more combinations or combinations selected from the group consisting of the constituents described in the expression of the machine form, And the like.

As an example of a method of encoding an actual image and its depth information map, the Moving Picture Experts Group (MPEG) and the Video Coding Experts Group (VCEG) having the highest coding efficiency among the video coding standards developed so far jointly standardize Encoding can be performed using HEVC (High Efficiency Video Coding).

The present invention relates to an intra prediction method of an image encoding / decoding method and apparatus, and more particularly, to an intraprediction method of an image encoding / decoding method and apparatus, in which intra prediction is performed according to each direction, The optimal filtering method is determined to obtain the predicted value, thereby minimizing the size of the residual signal and improving the compression performance.

Specifically, in prediction in various directions applied in the intra prediction method, when prediction for generation of a residual signal is performed according to each prediction direction (Prediction mode), in order to obtain an accurate prediction value, a DCT- (Sync) function based filtering is applied to obtain accurate prediction values, thereby improving the compression performance.

FIG. 2 is a view for explaining an example of image encoding units.

In HEVC, CTB (Coding Tree Block) is used as a unit of video encoding. In this case, CTB is defined as various square shapes. The CTB is called the coding unit CU (coding unit).

Referring to FIG. 2, a coding unit (CU) is a quad tree, and when a maximum coding unit LCU (Largest Coding Unit) having a size of 64 × 64 is used, the depth The coding unit searches for an optimal prediction unit recursively up to a coding unit (CU) of 8 × 8 size.

A prediction unit for performing prediction is defined as a PU (Prediction Unit). Each coding unit (CU) is predicted by a unit divided into a plurality of blocks, and is divided into a square and a rectangle to perform prediction.

According to the HEVC standard for coding / decoding moving picture data, a method for reducing the spatial correlation of residual signals in a block after performing inter / intra prediction and increasing the compression ratio of energy, (Transform Unit) is used. The converted signal according to the conversion unit is finally generated as a bitstream through entropy coding.

On the other hand, intra prediction considers 34 directional directions in total, and is applied to a square PU having a maximum prediction unit of 64x64 from a minimum prediction unit of 4x4. The predicted value of the block boundary used for each prediction generates 1/32 linear interpolation (interpolation) predicted pixel in comparison with the distance of two pixels for prediction according to each direction. Since it considers only the values of two pixels, There is a problem in that the change of various pixels, which is the edge of the pixel, is not considered at all.

Particularly, in the case of a change in the image or a pixel positioned at the boundary of the object, the prediction method using the two pixels has a problem that the efficiency of the conventional method is further reduced.

FIG. 3 shows an example of an interpolation method performed for intra prediction, and shows a method of predicting a pixels for prediction modes 2 to 34 having directionality among 34 prediction methods.

Referring to FIG. 3, a direction is positioned between a left reconstructed pixel and an upper reconstructed pixel according to each direction, and a value of a position applied to the direction is used to generate a predictive pixel using values of two pixels .

For example, in mode 32, prediction can be performed for a pixel through calculation of weights according to distances between A and B pixels. In this case, since the prediction is performed through two pixels, the sudden change of the image value is not effective at the boundary.

According to the embodiment of the present invention, in the prediction according to various directions applied in the intra prediction method, when a prediction for generation of a residual signal is performed according to each prediction direction (prediction mode) , DCT-IF filtering is applied to obtain accurate prediction values, thereby improving compression performance.

To solve this problem, it is possible to generate a predictive pixel through DCT-based filtering using four neighboring pixels, for example, Q, A, B, and C pixels as shown in FIG.

For example, the DCT-based filtering coefficients can be represented using a type-2 DCT as shown in Equation 1 below.

In this case, the DCT is first performed and a coefficient f (x) according to each position is generated through the inverse transformation of the DCT. 4, f (0) = Q, f (1) = A, f (2) = B, f (3) = C, N = 4.

N: number of samples (if N = 4 becomes 4-point Inverse DCT f (x))

In the case of N = 4 (i.e., 4-point DCT and IDCT are used), x = 1 + 1/2 (= N / 2 + 1/2) in the case of the 30th prediction mode in FIG. 4, An interpolation value of 1 + 1/2 sample position can be generated using Equation (2) of Equation (1).

In the case of the 29th prediction mode in FIG. 4, x = 1 + 13/32 (= N / 2 + 13/32) You can create interpolation values for 32 sample positions.

In the case of the 32nd prediction mode in FIG. 4, x = 1 + 26/32 (= N / 2 + 26/32) You can create an interpolation value for the sample position.

Meanwhile, the number of filter tapes used in the DTC-based interpolation may be set to four as well as various numbers, and the number of Tap may be signaled to a decoder or may include surrounding information (for example, edge information or mode information Etc.).

Although the present invention uses type-2 DCT and inverse DCT (IDCT), it is possible to use other types of DCT and IDCT as well as various types of Discrete Sine Transform (DST) and IDST pairs. Method.

Fig. 5 shows an example of prediction pixel generation in three directional modes for predicting k for the prediction modes related to 34 prediction directions.

5, when a value applied to a direction corresponding to a particular intra prediction mode does not exist in a correct direction, a predictive pixel may be generated through a linear filtering method using values of two pixels on both sides .

For example, the 28th mode of k performs the prediction for the 28th mode through calculation of the weight according to the distance between the pixels of C and D. In the case of the 19th mode, the prediction pixels of k are determined by applying Q and A in the same manner and M and N in the case of the 8th mode.

Since the prediction is performed through only two pixels on both sides, the change of the image value is not efficient at the boundary. In order to solve this problem, each prediction pixel is predicted by DCT-based filtering using four neighboring pixels such as (B, C, D, E) and (L, M, N, O) Pixels can be generated.

FIG. 6 shows an example of a method of generating a predictive pixel using 4-tap DCT-IF filtering.

Referring to FIG. 6, a 4-tap DCT-IF filter is used to predict the mode prediction of the i-th direction and the 32-directional mode prediction using Q, A, B, Mode predictive pixel of C, D, M, N, and O, and generates a 32-mode predictive pixel between C and D using four pixels of L, M, N and O. Here, DCT-IF filter.

Here, the number of Tap can be converted into the size of N-Tap as well as four, and it can be determined by signaling with a decoder or the size of Tap through surrounding information (edge information, mode information, etc.).

FIG. 7 is a diagram for explaining an example of a method of performing intraprediction using a filter having a different number of taps in the horizontal direction and the vertical direction, in which the horizontal direction is 8-Tap and the vertical direction is 4-Tap DCT-IF filtering To generate a predictive pixel.

In the case of FIG. 7, the number of Tap can be determined differently from that of FIG. 6, such as using 8-Tap at the top and 4-Tap at the left according to the characteristics of the image. All of which may be possible.

As shown in FIG. 7, when performing the prediction of the pixel, when the left 4th mode is performed, the predicted pixel is located between M and N, which is input to the predicted pixel of the 4th mode through 4-tap DCT- It is possible to predict the corresponding prediction pixel. In the case where the prediction mode in the i-th pixel is the mode in the 32-th direction, the prediction can be predicted through the value of the prediction pixel through the 8-tap DCT-IF filtering at the top.

In addition, the above-described filter-related information includes all possible methods without signaling through neighboring information, or a method capable of decoding filtering pixels differently in number of Tap through signaling.

Since the generated interpolation coefficient value is composed of decimal points, it is possible to obtain a bilaterally symmetric interpolation filtering coefficient value normalized to an integer value as follows.

Table 1 below shows an example of the interpolation coefficient values according to the intra prediction method of 4-point (4-tap) filtering.

For example, when the 32nd prediction mode is determined for the predictive pixel of FIG. 4A, the value of the position can be calculated as shown in the following Equation 2 by applying the filtering coefficient of the table above.

In this way, when the prediction pixel is generated by the coefficients of the DCT-based interpolation for each position of each mode, it is possible to generate a more accurate prediction pixel than to generate the prediction pixel for each position only considering the existing two surrounding pixels to be.

Here, the DCT-based interpolation filter represents one example and includes all the methods that can selectively apply various filtering.

A moving picture encoding / decoding method in which a compression ratio is increased by generating a residual signal considering a change in a surrounding image through filtering using a signal used for each prediction, Device.

In the above example, the interpolation filter coefficient is derived from the case of N = 4 and used as a prediction sample. However, N = 2n, n = 1, 2, 3, 4, ... The interpolation filter can be used in all cases.

In this example, type-2 DCT and IDCT are used, but other types of DCT and IDCT can be used. Also, when using different types of DST (Discrete Sine Transform) and IDST pairs, Performance can be improved.

On the other hand, when a prediction pixel is fetched from the top, it may be effective to use a small number of filtering tapes when there is an edge between two pixels located at the top (for example, when the value change is large).

Conversely, it may be effective to use a large number of filtering taps to have a large effect on the surrounding pixels if there is no edge (e.g., a small change in value) Can be applied equally to all of the predictions.

That is, by applying a method of increasing or decreasing the number of Tap according to a change in surrounding pixels (for example, edge information or size between pixels), accurate prediction pixels can be generated to achieve the entire coding / decoding performance .

According to another embodiment of the present invention, in prediction according to various directions applied in an intra prediction method, when a prediction for generation of a residual signal is performed according to each prediction direction, In order to obtain accurate prediction values through interpolation by applying Sync filtering, the compression performance can be improved.

For example, in the intraprediction method as described with reference to FIGS. 4 to 7, prediction pixels can be generated by performing interpolation using a Sync filter in place of or in addition to the DCT-based filter .

Figs. 8 and 9 are diagrams for explaining the operation of sync filtering. Fig.

FIG. 8 shows a normalized sync function, and the coefficients of each 4-tap can be obtained through a Sync function as shown in FIG.

Through interpolation, each prediction pixel can be obtained according to the method described with reference to FIGS. 4 to 7. FIG. In addition, it can include all methods that can be applied not only to 4-tap but also to N-tap.

All the images that are viewed through the video equipment are analog signals converted into digital signals through sampling - restoration. Therefore, if the digital signal x [k] exists infinitely, it can be restored to the ideal analog signal by the following equation (3).

In the above equation (3), t represents a time, Ts represents a sampling period, and x [kTs] represents a digital signal. When the sampling period Ts is defined as 1, the graph can be as shown in FIG.

The analog signal x (t) can be restored by substituting the value of the digital signal into Equation 3 above.

For example, a predicted pixel between integer pixels can be obtained using the restored analog signal x (t). The filter coefficient can be obtained according to the following Equation (4) according to the number of the reference integer pixels and the position of the prediction pixel.

In Equation (4),? Is the predicted pixel position, and N is the number of integer pixels to be referred to.

For example, if the number of reference integer pixels is 8 (N = 8), the filter coefficient can be obtained according to the predicted pixel position as shown in the following table.

Since the above filter coefficient is high in calculation complexity with a decimal point operation, the integer filter coefficient can be obtained as shown in Table 3 by scaling.

In a similar way, all filter coefficients can be determined according to the values of α and N.

According to an embodiment of the present invention, when selecting interpolation, the number of taps is selected in order to improve performance by varying the number of filtering taps used for horizontal and vertical interpolation through edge information of an image. The number of taps can be predicted in decryption by signaling or by using surrounding information.

As described above, in the method of finding a motion vector (MV), interpolation can greatly affect performance. In particular, prediction through accurate interpolation affects performance that continues with subsequent transformations, quantization, scanning, and entropy encoding / decoding. In other words, accurate interpolation through interpolation has a great effect on the performance improvement of the entire codec.

The embodiment of the present invention is an interpolation method that improves the performance of the entire encoding / decoding by changing the number of filtering tap according to the information according to the edge direction of the image for an effective method than the conventional interpolation.

Hereinafter, embodiments for performing intra prediction by adjusting the number of filtering tabs will be described in detail with reference to FIGS. 10 to 12. FIG.

FIG. 10 is a diagram for explaining an example of a method for determining the tap number of filtering. FIG. 10 shows a method for analyzing an edge of an image and determining the number of taps in the horizontal and vertical directions.

10, the edges of the image are analyzed and detected. After the tap size (number of tap) of the filter is determined according to the detected edge, the number of tapes in the vertical direction and the horizontal direction Interpolation may be performed using a filter of < RTI ID = 0.0 >

For example, as shown in FIG. 7, interpolation using a 4-tap filter may be performed in the horizontal direction, and interpolation using an 8-tap filter may be performed in the vertical direction.

Here, the number of filtering taps may be transmitted from the encoder to the decoder in order to decode them, or may include all the methods that can be distinguished without signaling using certain characteristics.

FIG. 11 shows an example of 2x extended interpolation using a 4-tap filter and an 8-tap filter. An example of a case where a picture used as a reference is a half-pixel prediction of a frame or a picture is shown.

Interpolation is performed with a tap size determined by the method as described with reference to FIG. 10, according to the edge of Horizontal and Vertical according to the edge.

Step 1 can be predicted by 4-tap filtering for horizontal interpolation, and pixel prediction can be performed using 8-tap for vertical interpolation.

The size of the tap of the present invention is not limited to the above-mentioned number but includes all the methods of constituting N-tap, and tap coefficients can be applied regardless of the size of the width and the length. This applies differently depending on the edge characteristics of the image.

In addition, through the signaling, it is possible to determine the size of the tap according to the size and direction of the tap, or the information about the edge direction and the change of the image (MV change, pixel change, etc.) And a predictable method.

FIG. 12 illustrates an example in which various interpolation methods are applicable. In FIG. 12, various interpolation methods are illustrated to illustrate the four possible interpolation methods.

Referring to FIG. Through the total step 4, four interpolation methods can be applied to both the horizontal and vertical directions.

As described above, the present invention refers to an interpolation method in which not only one interpolation technique but also N interpolation methods can be adaptively changed according to the enhancement of characteristics and performance of an image with N-Tap number, This is a method that includes both possible methods and methods that can be decoded through signaling in some cases.

On the other hand, when there is an edge or a large image is changed, it is effective to use the number of tapes for filtering, which is an interpolation method that utilizes edges, and to use the number of filtering tap when there is little edge change and few pixels. .

For example, if the edges are distributed horizontally, it is effective to increase the number of tapes of the horizontal filtering and reduce the number of tapes of the vertical filtering. If the edges are distributed vertically, increase the tap number of the vertical filtering, It is effective to reduce the number of tapes.

Alternatively, you can use both methods and include a method of selecting a high-performance method through signaling.

In the case of predicting and finding the MV in the inter prediction, the double interpolation means a method of predicting at a 1/2 pixel position and the 4 times interpolation means a method of predicting a 1/4 pixel position.

According to another embodiment of the present invention, when interpolation is performed, performance can be improved by changing the number of filtering tap used for interpolation for each region through a change of a specific region or a pixel change in an image, In this case, the number of taps can be predicted in decryption by using signaling or by using surrounding information.

In order to be more effective than the conventional interpolation method, the number of filtering taps may be varied according to the change of the pixel value in the area or block unit, thereby improving the performance of the entire coding / decoding.

For example, the A-tap filtering is performed on a region of an object through a specific object detection and the B-tap filtering is performed on the remaining region. Interpolation can be performed adaptively with different filtering coefficients.

13 and 14 are block diagrams showing an embodiment of an arrangement for performing intraprediction by adjusting the number of filtering taps according to the conversion of pixel values

FIG. 13 shows an example of a method of differentiating taps of filters used in interpolation for each frame in one frame.

Referring to FIG. 13, the number of taps can be determined differently for each region in one frame, and the interpolation method including all interpolation at the number of taps, rather than using the same tap number in one frame.

The determination of the number of Tap is performed by detecting an object or interpolating the number of taps in one frame as shown in FIG. 10 through a change of a pixel in a block (area) unit and a change of an adjacent block (area) to be.

Here, the number of taps includes both decoding by signaling, and methods capable of performing the same decoding through surrounding pixels and area changes without signaling.

FIG. 14 shows an example of the division of the interpolation tap and the vertical and horizontal tap for each region, and shows an example of a method of further differentiating the vertical and horizontal tap from the case shown in FIG. 13 .

Referring to FIG. 14, by changing the number of taps and the length and width of each region, optimum interpolated pixel values can be obtained in accordance with the characteristics of an image, thereby improving the performance of the entire encoding / decoding.

Here, the tap for Vertical and Horizontal tap and the area is a method including both decryption by signaling and decryption without signaling.

For example, if the area is divided by one frame and there are many edges, it is necessary to determine the number of filtering tapes in order to preserve the edges. .

In detail, in FIG. 13, an area using 8-tap interpolation is interpolated using a large number of filtering tapes with a small number of edges. In the case of 4-Tap, You can interpolate the whole image using fewer filtering tap counts.

In addition, FIG. 14 shows a method of varying the number of tapes according to the direction of the edge, and dividing the number of tapes of Vertical and Horizontal in each region by the method of FIG.

This may be the same as applying the method of varying the number of tapping taps according to Vertical and Horizontal as described with reference to Figs. 10 to 12 above.

15 is a block diagram illustrating a configuration of a video encoding apparatus according to an embodiment of the present invention.

Generally, the encoding apparatus includes an encoding process and a decoding process, and the decoding apparatus has a decoding process. The decoding process of the decoding apparatus is the same as the decoding process of the encoding apparatus. Therefore, the encoding apparatus will be mainly described below.

As shown in FIG. 15, an image encoding apparatus according to an embodiment of the present invention includes an encoding unit and structure, Inter prediction, Intra prediction, Interpolation, Filtering, Transform ) Method and so on.

15, an image encoding apparatus includes an encoding mode determination unit 110, an intra prediction unit 120, a motion compensation unit 130, a motion estimation unit 131, a transcoding / quantization unit 140, A dequantization / conversion decoding unit 160, a deblocking filtering unit 170, a picture storage unit 180, a subtracting unit 190, and an adding unit 200. The dequantization /

The encoding mode determination unit 110 analyzes an input video signal to divide a picture into a predetermined size of an encoding block, and determines a coding mode for the divided predetermined size of the encoding block. The encoding mode includes intraprediction encoding and inter prediction encoding.

The picture is composed of a plurality of slices, and the slice is composed of a plurality of maximum coding units (LCU). The LCU can be divided into a plurality of coding units (CUs), and the encoder can add information indicating whether or not to be divided to a bit stream. The decoder can recognize the position of the LCU by using the address (LcuAddr). The coding unit CU in the case where division is not allowed is regarded as a prediction unit (PU), and the decoder can recognize the position of the PU using the PU index.

The prediction unit PU may be divided into a plurality of partitions. Also, the prediction unit PU may be composed of a plurality of conversion units (TUs).

The encoding mode determination unit 110 sends the image data to the subtraction unit 190 in units of blocks of a predetermined size (for example, in units of PU or TU) according to the determined encoding mode.

The transform coding / quantizing unit 140 transforms the residual block calculated by the subtracting unit 190 from the spatial domain to the frequency domain. For example, two-dimensional discrete cosine transform (DCT) or discrete cosine transform (DST) -based transform is performed on the residual block.

In addition, the transcoding / quantization unit 140 determines a quantization step size for quantizing the transform coefficient, and quantizes the transform coefficient using the determined quantization step size. The quantization matrix can be determined according to the determined quantization step size and encoding mode.

The quantized two-dimensional transform coefficients are transformed into one-dimensional quantized transform coefficients by one of the predetermined scanning methods. The transformed one-dimensional sequence of quantization transform coefficients is supplied to the entropy encoding unit 150.

The inverse quantization / conversion decoding unit 160 dequantizes the quantization coefficients quantized by the transcoding / quantization unit 140. Further, the inverse quantization coefficient obtained by inverse quantization is inversely transformed. Accordingly, the residual block transformed into the frequency domain can be restored into the residual block in the spatial domain.

The deblocking filtering unit 170 receives the inverse quantized and inverse transformed image data from the inverse quantization / inverse transform coding unit 160 and performs filtering to remove a blocking effect.

The picture storage unit 180 receives the filtered image data from the deblocking filtering unit 170 and restores and restores the image in picture units. The picture may be a frame-based image or a field-based image. The picture storage unit 180 has a buffer (not shown) capable of storing a plurality of pictures. A plurality of pictures stored in the buffer are provided for intra prediction and motion estimation.

The pictures provided for intra prediction or motion estimation are referred to as reference pictures.

The motion estimation unit 131 receives the at least one reference picture stored in the picture storage unit 180 and performs motion estimation to output motion data including an index indicating a motion vector and a reference picture and a block mode do.

In order to optimize the prediction precision, a motion vector is determined with a fractional pixel precision, for example, 1/2 or 1/4 pixel accuracy. Since the motion vector can have a fractional pixel precision, the motion compensation unit 130 applies the interpolation filter for calculating the pixel value of the fractional pixel position to the reference picture so that the pixel value of the fractional pixel position .

The motion compensation unit 130 is configured to perform motion compensation on a block to be coded from a reference picture used for motion estimation among a plurality of reference pictures stored in the picture storage unit 180 according to the motion data input from the motion estimation unit 131 And outputs the extracted prediction block.

The motion compensation unit 130 determines a filter characteristic of the adaptive interpolation filter necessary for motion compensation with a decimal precision. The filter characteristic is, for example, information indicating the filter type of the adaptive interpolation filter and information indicating the size of the adaptive interpolation filter.

The size of the filter is, for example, the number of taps, which is the number of filter coefficients of the adaptive interpolation filter.

Specifically, the motion compensation unit 130 may determine either a separate type or a non-separable type adaptive filter as an adaptive interpolation filter. Then, the number of taps of the determined adaptive interpolation filter and the value of each filter coefficient are determined. The value of the filter coefficient can be determined differently for each position of the fractional pixel relative to the integer pixel. Also, the motion compensation unit 130 may use a plurality of non-adaptive interpolation filters with fixed filter coefficients.

The motion compensation unit 130 can set the characteristics of the interpolation filter in a predetermined processing unit. For example, it can be set in a fractional pixel unit, a coding basic unit (encoding unit), a slice unit, a picture unit, or a sequence unit. In addition, one characteristic may be set for one video data.

Therefore, since the same filter characteristic is used in a predetermined processing unit, the motion compensation unit 130 has a memory that temporarily holds the filter characteristic. This memory maintains filter characteristics, filter coefficients, and the like as needed. For example, the motion compensation unit 130 can determine the filter characteristic for each I picture and determine the filter coefficient for each slice.

The motion compensation unit 130 receives a reference picture from the picture storage unit 180 and applies a filter process using the determined adaptive interpolation filter to generate a prediction reference picture of a decimal precision.

Then, based on the generated reference picture and the motion vector determined by the motion estimation unit 131, motion compensation is performed with a small number of pixels to generate a prediction block.

The subtractor 190 receives the block in the reference picture corresponding to the input block from the motion compensator 130 and performs a difference operation with the input macroblock in the case of performing inter picture prediction coding on the input block to be coded, and outputs a residue signal.

The intraprediction unit 120 performs intraprediction encoding using the reconstructed pixel values in a picture to be predicted. The intra prediction unit receives the current block to be predictively encoded and performs intra prediction by selecting one of a plurality of intra prediction modes preset according to the size of the current block. The intra predictor 120 determines the intra prediction mode of the current block using the previously coded pixels adjacent to the current block, and generates a prediction block corresponding to the determined mode.

The previously encoded region of the current picture is decoded again for use by the intra prediction unit 120 and stored in the picture storage unit 180. [ The intra prediction unit 120 generates a prediction block of a current block using pixels neighboring the current block or non-adjacent but applicable pixels in the previously coded area of the current picture stored in the picture storage unit 180. [

The intra prediction unit 120 may adaptively filter adjacent pixels to predict an intra block. For the same operation in the decoder, it is possible to transmit information indicating whether or not filtering is performed in the encoder. Or the intra-prediction mode of the current block and the size information of the current block.

The prediction type used by the image coding apparatus depends on whether the input block is coded in the intra mode or the inter mode by the coding mode determination unit.

The switching between the intra mode and the inter mode is controlled by the intra / inter selector switch.

The entropy encoding unit 150 entropy-codes the quantization coefficients quantized by the transcoding / quantization unit 140 and the motion information generated by the motion estimation unit 131. [ Also, an intra prediction mode, control data (e.g., quantization step size, etc.), and the like can be coded. Also, the filter coefficient determined by the motion compensation unit 130 is encoded and output as a bit stream.

As described above, the configuration of the image decoding apparatus according to an embodiment of the present invention can be derived from the configuration of the image encoding apparatus shown in FIG. 15, and for example, the inverse of the encoding process as described with reference to FIG. So that the image can be decoded.

The method according to the present invention may be implemented as a program for execution on a computer and stored in a computer-readable recording medium. Examples of the computer-readable recording medium include a ROM, a RAM, a CD- , A floppy disk, an optical data storage device, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet).

The computer readable recording medium may be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner. And, functional programs, codes and code segments for implementing the above method can be easily inferred by programmers of the technical field to which the present invention belongs.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It should be understood that various modifications may be made by those skilled in the art without departing from the spirit and scope of the present invention.

Claims (11)

A method for performing intra prediction,
Dividing an image to be predicted into a plurality of regions and determining a number N of filtering taps applied to each of the plurality of regions;
Generating prediction pixels by performing interpolation for each of the divided regions using N reference pixels including two adjacent pixels according to an intra prediction direction; And
And generating a residual signal using the generated predictive pixel. 2. The filter according to claim 1, wherein the filtering tap number (N)
Wherein the intra prediction method is determined according to at least one of a position of an object in the image, a pixel value change in a specific area, and a pixel value change in a surrounding area. 3. The method of claim 2,
Wherein a different filtering tap number is applied to the first area corresponding to the position of the object and the remaining second area. The method according to claim 1,
Wherein different filtering tap numbers are applied to any one of the plurality of divided regions in the vertical direction and the horizontal direction. The method according to claim 1,
Wherein at least one of the type of filtering and the number of taps is transmitted from an encoding device to a decoding device. The method according to claim 1,
Wherein the number of taps of the filtering is determined to be smaller than the case where edges exist between the adjacent two pixels when an edge exists between the adjacent two pixels. The method according to claim 1,
The number of filtering taps in the horizontal direction is determined to be larger than the number of filtering taps in the vertical direction when edges are present in a horizontal direction more than the vertical direction. An apparatus for performing intra prediction, the apparatus comprising:
A tap number determination unit for dividing an image to be predicted into a plurality of regions and determining a number of filtering taps N to be applied to each of the plurality of regions;
A prediction pixel generation unit for generating prediction pixels by performing interpolation for each of the divided regions using N reference pixels including two adjacent pixel candidates according to an intra prediction direction; And
And a residual signal generator for generating a residual signal using the generated predictive pixel. 9. The method of claim 8, wherein the filtering tap number (N)
Wherein the intra prediction is determined according to at least one of a position of an object in the image, a pixel value change in the specific area, and a pixel value change in the surrounding area. 10. The method of claim 9,
Wherein different filtering tap numbers are applied to the first area corresponding to the position of the object and the remaining second area. 9. The method of claim 8,
Wherein a different filtering tap number is applied to any one of the divided regions in the vertical direction and the horizontal direction.

Images