KR100816461B1 - Real-time deblocking filter and Method using the same - Google Patents

Abstract

The present invention relates to a deblocking filter and a deblocking method using the same, and more particularly, to a deblocking filter and a deblocking method using a modified hardware structure to reduce decoding time. The deblocking filter according to the present invention is a device for removing a blocking effect generated by decoding an image frame encoded in macroblock units. The deblocking filter temporarily removes data input from an external source or data generated while removing a blocking effect. An input buffer for storing; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And data storage means for storing the data temporarily stored in the output buffer in macroblock units.

Deblocking, Filter, Boundary Area, Edge, Pipeline Structure, Real Time

Description

Real-time deblocking filter and method using the same

The following drawings attached to this specification are illustrative of preferred embodiments of the present invention, and together with the detailed description of the invention to serve to further understand the technical spirit of the present invention, the present invention is a matter described in such drawings It should not be construed as limited to

1 is a block diagram schematically showing the configuration of a decoder using the H.264 method according to the prior art.

2 is a timing diagram of data decoded and deblocked in the H.264 method according to the related art.

3 is a diagram showing the configuration of a frame of image data according to a preferred embodiment of the present invention.

FIG. 4 is a diagram illustrating the (1,1) block of FIG. 3 in a 4: 2: 0 format of the YUV method.

5 is a block diagram schematically illustrating a configuration of a deblocking filter according to a preferred embodiment of the present invention.

FIG. 6 is a diagram illustrating an arrangement state of parameters of part VI of FIG. 4.

7A to 7F are diagrams showing the structure of data stored in the data storage means in order.

8 is a timing diagram of data decoded and deblocked in an H.264 scheme according to an embodiment of the present invention.

<Description of main reference numerals in the drawings>

1 ... inverse converter 2 ... motion compensation processor

3 ... deblocking filter 4 ... memory

10 ... Input buffer 21, 22 ... Selection circuit

30 ... edge filter 40 ... data alignment circuit

50 ... Output buffer 60 ... DMAC

70.Data storage means

The present invention relates to a deblocking filter and a deblocking method using the same, and more particularly, to a deblocking filter and a deblocking method using a modified hardware structure to reduce decoding time.

When an image or sound is converted into digital data, the amount of this data is quite large. In order to reduce the size of image or sound data converted to digital format, various compression techniques (eg, GIF, JPEG image compression technology, MPEG, H.263, H.264 video compression technology) have been researched and developed.

H.264, one of the compression technologies, is a new video compression coding standard, also known as 'ISO / IEC 14496 10 Advanced Video Coding'. Compression standard.

1 is a block diagram of a decoder using the H.264 method according to the prior art. Referring to FIG. 1, a decoder according to the related art includes an inverse transform unit 1 for transforming image data through an inverse discrete cosine transform (IDCT) scheme, a motion compensation processor 2 for estimating and compensating for motion in units of blocks. And a deblocking filter 3 for compensating block boundary regions appearing between adjacent blocks in accordance with compression, and a memory 4 for storing decoded image data.

Looking at the decoding process using the H.264 method, the data of the frame unit input for decoding processing is preferentially inversely transformed. The inverse transformed data is reconstructed into data (hereinafter referred to as first data) in which motion compensation is completed and motion compensation is added. Next, the first data is stored in the memory 4, and the deblocking filter 3 reads the first data from the memory 4 and performs the deblocking process. When the deblocking is completed, the deblocking filter 3 writes the deblocked data (second data) to the memory 4 again. The above process is repeated to decode the image data consisting of a plurality of frames.

In the above-described conventional decoding method, since a read / write must be repeated for one frame, the bandwidth for the memory access increases during the deblocking process. In addition, after decoding of one frame is completed, deblocking of the next frame should be performed (see FIG. 2). Therefore, a problem arises in that the decoding time of the image data becomes long.

The present invention was devised to solve the above problems, and by processing the image data in units of macro blocks, the bandwidth and decoding processing time for memory access due to decoding can be significantly reduced. It is an object of the present invention to provide a blocking filter and a deblocking method using the same.

In order to achieve the above object, a deblocking filter according to an aspect of the present invention is an apparatus for removing a blocking effect generated by decoding an image frame encoded by a macroblock unit, and receives data or a blocking effect from an external source. An input buffer for temporarily storing data generated in the process of removing the data; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And data storage means for storing the data temporarily stored in the output buffer in macroblock units.

The input buffer may include: a first input buffer configured to temporarily store boundary area data in a vertical direction among macroblocks that have already been deblocked and which are currently deblocked; A second input buffer configured to temporarily store boundary area data in a horizontal direction in contact with a macroblock currently being deblocked; A third input buffer configured to temporarily store data generated when an edge filter deblocks the data of the macroblock unit into a column unit or a row unit composed of a plurality of subblocks; A fourth input buffer which temporarily stores data in which the deblocking process of the vertical direction is completed during the deblocking process; And a fifth input buffer for temporarily storing original data of the macroblock currently being deblocked.

Data stored in the input buffer is preferably stored in the YUV (luminance, color) method.

It may further include a selection circuit for selectively transmitting the data output from the input buffer to the edge filter in accordance with a predetermined rule.

The selection circuit may include a first selection circuit which receives data output from the first input buffer, the second input buffer, and the third input buffer; And a second selection circuit which receives data output from the fourth input buffer and the fifth input buffer.

The data stored in the output buffer is preferably stored in the data storage means through a direct memory access controller (DMAC).

A deblocking filtering method according to another aspect of the present invention is a method for removing a blocking effect generated by decoding an image frame encoded in macroblock units, the method comprising: (1) removing data and blocking effects of a predetermined region input from an outside; Temporarily storing data of a predetermined region generated in the removing process in a plurality of input buffers; (2) filtering the boundary regions in the vertical direction and the horizontal direction using data output from the plurality of input buffers; (3) selectively converting the matrix alignment of the filtered data through step (2); (4) temporarily storing data generated in the step (3) in an output buffer; And (5) combining the data output from the output buffer in units of macro blocks and storing the data in the data storage means.

The plurality of input buffers include a first input buffer, a second input buffer, a third input buffer, a fourth input buffer, and a fifth input buffer, wherein the step (2) includes a macroblock in which the deblocking process is already completed. Among the first input buffer, the boundary area data in contact with the macroblock currently being deblocked in the vertical direction and the border area data in contact with the macroblock currently being deblocked in the horizontal direction are generated during the second input buffer and the deblocking process. A third input buffer for temporarily storing data and data for which deblocking processing in the vertical direction is completed in the deblocking process are stored in the fourth input buffer and original data of the macroblock being deblocked in the fifth input buffer.

Data stored in the fifth input buffer may be stored in a YUV (luminance, color) method.

Preferably, the method may further include sequentially selecting and transmitting the multiple data output from the input buffer according to a predetermined rule.

The data stored in the output buffer is preferably stored in the data storage means through a direct memory access method (DMA).

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

3 is a diagram showing the configuration of a frame of image data according to a preferred embodiment of the present invention, and FIG. 4 is a diagram showing the (1,1) block of FIG. Referring to FIG. 3 or 4, the image data is composed of a plurality of frames, wherein each frame is defined as a plurality of macro blocks having a fixed size. That is, one frame consists of a combination of n x m macroblocks.

In this embodiment, the frame of image data is preferentially arranged in a row from the left block to the right block ((0,0), (1,0), (2,0), ... (n, 0)) in order. Deblocking is performed. Next, deblocking of blocks ((0, 1), (1, 1), (2, 1), ... (n, 1)) constituting the lower row of the deblocking row is performed. The deblocking of the image data frame is completed by repeating the above process.

Also, in the present embodiment, each macro block is sampled in the 4: 2: 0 format of the YUV method. The macroblock sampled in the 4: 2: 0 format of the YUV method is subjected to deblocking in the order of the luminance component (Y), the red color component (U), and the blue color component (V). Each component is first deblocked for the boundary region in the column direction, and then deblocking for the boundary region in the row direction is performed. In addition, the deblocking process is performed based on a sub-block (4 × 4 pixels, each region enclosed by a thick solid line in FIG. 4) as a basic unit. For example, in the case of the deblocking of the luminance component (Y), the order proceeds from the leftmost a region to the column direction in the order of b, c, d regions. Next, deblocking of regions f, g, and h starts from region e, which is a boundary region of the uppermost subblocks in the row direction.

Although in the embodiment of the present invention, the macro block is sampled in the 4: 2: 0 format of the YUV method, the present invention is not limited thereto, and for example, may be sampled in the RGB method. In addition, the sampling format can be variously modified to 4: 4: 2 and 4: 4: 4 formats.

5 is a block diagram showing the configuration of a deblocking filter according to a preferred embodiment of the present invention. 5, a deblocking filter according to an embodiment of the present invention includes a plurality of input buffers 10, an edge filter 30 for filtering a block boundary region using data output from the input buffer 10, In the data alignment circuit 40 for arranging the format of the data output from the edge filter 30, the output buffer 50 for temporarily storing the data output from the data alignment circuit 40, and the output buffer 50. And a data storage means 70 for combining and storing the temporarily stored data.

The input buffer 10 includes a first input buffer 11, a second input buffer 12, a third input buffer 13, a fourth input buffer 14, and a fifth input buffer 15.

The first input buffer 11 is directly in contact with the block boundary region on one side located in the reverse direction of the deblocking direction, for example, the top, bottom, left, right, and four sides of the macroblock ((1, 1) in FIG. 3) being deblocked. Among the four neighboring blocks adjacent to each other, the boundary area with the block located above is temporarily stored.

The second input buffer 12 is not opposite to the block located on the one side, and the block boundary area of the side located in the reverse direction of the deblocking direction, for example, the macroblock in the deblocking process (see FIG. 1 and 1) temporarily store the boundary area (b) data with the block located on the left side.

The third input buffer 13 generates data generated in the process of processing the boundary values of the respective macroblocks divided into a plurality of subblocks, for example, from a region to a subblock of the luminance component Y, b, c, and d. Temporarily stores data generated in the process of filtering sequentially up to regions e, f, g, and h.

The fourth input buffer 14 receives and temporarily stores the filtered data in the column direction of the macroblock currently undergoing the deblocking process, for example, the filtered data from the a region to the b, c, and d regions.

The fifth input buffer 15 stores the original data of the macroblock currently undergoing the deblocking process in the 4: 2: 0 format of the YUV method.

Preferably, the selection circuits 21 and 22 may be further provided at the rear end of the input buffer 10. The selection circuits 21 and 22 receive data from the plurality of input buffers 11, 12, 13, 14, and 15, and selectively select the data required for deblocking by a control signal applied from the outside. To 30).

In the embodiment of the present invention, two multiplexers 21 and 22 are adopted as the selection circuit. Then, the output terminals of the first input buffer 11, the second input buffer 12, and the third input buffer 13 are connected to the first multiplexer 21, and the fourth input buffer is connected to the second multiplexer 22. 14, and the output terminal of the fifth input buffer (15).

Although the embodiment of the present invention implements a circuit having two multiplexers 21 and 22 as the selection circuit, the present invention is not limited thereto, and the selection circuit selectively selects data necessary for deblocking an edge filter ( A circuit capable of transmitting to 30 is sufficient.

The edge filter 30 deblocks the block boundary region using the data input from the input buffer 10. Specifically, it is determined that the data is located in the edge region with reference to the parameter value of the data, and selectively smoothes the pixels located in the edge region. In addition, the edge filter 30 transmits the filtered data to the third input buffer 13 and the data alignment circuit 40.

The data alignment circuit 40 converts the matrix alignment state of the data output from the edge filter 30 from row to column or from column to row (see FIG. 6).

In addition, the data alignment circuit 40 does not selectively convert the alignment state with respect to the data of a specific area (e.g., (a) area data and (e) area data of FIG. 4).

The output buffer 50 temporarily receives the filtered data of the macroblock currently being filtered from the data alignment circuit 40.

The data storage unit 70 receives data in block units temporarily stored in the output buffer 50 and sequentially stores the data in frame units.

Preferably, the deblocking filter of the present invention stores the data output from the output buffer 50 to the data storage means 70, reads the data stored in the data storage means 70 to the first input buffer (11) DMAC 60 may be further provided for temporary storage.

Hereinafter, a structure of data stored in the data storage means 70 will be described with reference to FIGS. 7A to 7F. In the figure, S ₁ represents a macroblock that is not deblocked, S ₂ represents a macroblock that is currently deblocking, and S ₃ represents a macroblock that has already been deblocked and stored.

First, as the deblocking in the row direction proceeds from the (0,0) block to the (4,0) block, data from the (0,0) block to the (4,0) block is sequentially stored. At this time, since the right boundary of the (0,0) block is not deblocked, the four rightmost subblocks are not stored. As the deblocking of the (1,0) block proceeds, the boundary area where the (0,0) block contacts the (1,0) block is updated (Fig. 7B), and the data of the (1,0) block is stored. do.

By repetition of the above process (deblocking process of (1,0) block), deblocking is performed to block (3,0) and data is stored. In the rightmost macro block (4,0), since there is no macro block in contact with the right boundary area, the deblocking process for the right area is not necessary. Therefore, the rightmost macro block is stored as it is without deblocking the right boundary area (FIG. 7C).

Next, deblocking is performed from the next row (0,1) block to the (4,1) block, so that data from the (0,1) block to the (4,1) block is stored in the data storage means 70. Are stored sequentially.

At this time, as in the blocks (0,0) to (4,0), the right boundary of the block (0,1) is not deblocked, and thus the four rightmost subblocks are not stored (Fig. 7D). ). Also, as the deblocking of the (1,1) block proceeds, the boundary region where the (1,1) block contacts the (1,0) block is updated (FIG. 7E). By repetition of the above process (deblocking process of (1,1) block), deblocking up to (3,1) proceeds and is stored up to the rightmost (4,1) block (FIG. 7F). As a result, the deblocking of the image frame proceeds through the deblocking repetition of the macroblock unit as described above, and the storage is completed in the data storage means 70.

Hereinafter, the deblocking method according to an embodiment of the present invention will be described by explaining the operation of the deblocking filter according to the embodiment of the present invention with reference to the above-described components.

Frames consisting of a combination of a plurality of macro blocks are preferentially arranged from the top left (0,0) block to the (1,0), (2,0), ... (n, 0) blocks in the row direction. Proceed with deblocking. Next, deblocking of (0,1), (1,1), (2,1), ... (n, 1) blocks forming the lower row of the deblocking row is performed. The deblocking of the image data frame is completed by repeating the above process. In the process of deblocking, the filtered data of the boundary area is stored in the data storage means 70 in units of macro blocks.

The deblocking process will be described in detail by exemplifying the (1,1) block. First, data for the rightmost four subblocks are temporarily stored in the second input buffer 12 in the deblocking process of the block processed in the previous step (eg, (0,1) block). In addition, the data of the lowest four subblocks of the block ((1,0)) located above the block (1,1) currently processed among the already processed blocks is read from the data storage means 70. It is temporarily stored in the first input buffer 11. In the fifth input buffer 15, original data for the (1, 1) block is temporarily stored in the YUV method.

The edge filter 30 deblocks the boundary area (a) using data temporarily stored in the second input buffer 12 and the fifth input buffer 15, and deblocking of the area (a) is performed. Completed data (eg, the rightmost four subblocks of the (0,1) block and the leftmost four subblocks of the (1,1) block) are temporarily stored in the output buffer 50 and the third input buffer 13, respectively. Save it. When the deblocking of the boundary region for region (a) is completed, the edge filter 30 uses the data temporarily stored in the third input buffer 13 and the fifth input buffer 15, and then the boundary region for the region (b). Proceed with deblocking. In addition, when the edge region 30 deblocking for the region (b) is completed, for example, the leftmost four subblocks of the (1,1) block are arranged in the data through the data alignment circuit 40. 6 is temporarily stored in the fourth input buffer 14, and the second four sub blocks from the left of the (1, 1) block are temporarily stored in the third input buffer 13. The edge filter 30 sequentially deblocks the regions (c) and (d) in the same manner as described above. When the deblocking up to the area (d), that is, the deblocking on the vertical boundary area is completed, the data whose alignment has been converted by the data alignment circuit 40 is temporarily stored in the fourth input buffer 14. The data alignment circuit 40 temporarily stores the rightmost four subblocks in the second input buffer 12 among the deblocked data for the vertical boundary region.

Next, the edge filter 30 proceeds to deblock the boundary region for region (e) using data temporarily stored in the first input buffer 11 and the fourth input buffer 14, The output buffer 50 and the third input buffer of the data (for example, the bottom four subblocks of the (1,0) block and the top four subblocks of the (1,1) block) for which the boundary area deblocking has been completed Each of them is temporarily stored in (13). When the deblocking of the boundary region for the region (e) is completed, the edge filter 30 uses the data temporarily stored in the third input buffer 13 and the fourth input buffer 14 to define the boundary region for the region (f). Proceed with deblocking. When the deblocking of the boundary region for the region (f) is completed, for example, the top four subblocks of the (1,1) block are converted into data through the data alignment circuit 40, and the output buffer 50 is changed. Is stored temporarily). Also, the second four subblocks at the top of the (1,1) block are temporarily stored in the third input buffer 13. In the same manner as the boundary area deblocking for the region (f), the boundary area deblocking for the regions (g) and (h) is sequentially performed. When filtering to area (h) (that is, boundary area deblocking for luminance values) is completed, the deblocking data to area (h) is transmitted to the output buffer 50.

Thereafter, the edge filter 30 deblocks the boundary area for the areas (i) to (p) by using the data temporarily stored in the third input buffer 13 and the data temporarily stored in the fourth input buffer 14. To deal with.

Finally, the DMAC 60 stores the data stored in the output buffer 50 (for example, data for which the deblocking of the boundary area of the (1,1) block currently being processed) is completed as the data storage means 70.

In addition, the DMAC 60 stores data for the bottom four subblocks SB, SB, SB, and SB of the (2,0) block located next to the (2,1) block which is then deblocked. The data is read from the data storage means 70 and temporarily stored in the first input buffer 11. Thereafter, the above process is repeated to deblock the plurality of macroblocks.

By deblocking the data in macroblock units and sequentially storing the data by the deblocking method according to an exemplary embodiment of the present invention, a bandwidth for memory access according to decoding can be reduced, and a decoding processing time Can be significantly reduced.

Although the present invention has been described above by means of limited embodiments and drawings, the present invention is not limited thereto and will be described below by the person skilled in the art to which the present invention pertains. Of course, various modifications and variations are possible within the scope of the claims.

According to the real-time deblocking filter and the deblocking method using the same according to the present invention, by processing the image data in units of macro blocks, the bandwidth and decoding processing time for memory access due to decoding are significantly reduced. Can be.

In addition, as the decoding processing time is shortened, the operation clock can be lowered and the power consumption of the decoder can be reduced.

Claims (11)

delete An apparatus for removing a blocking effect generated by decoding an image frame encoded in macroblock units, An input buffer for temporarily storing data input from the outside or data generated in a process of removing a blocking effect; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And And data storage means for combining data temporarily stored in the output buffer in macro block units. The input buffer, A first input buffer which temporarily stores boundary area data in a vertical direction among macroblocks that have already been deblocked and which are currently deblocked; A second input buffer configured to temporarily store boundary area data in a horizontal direction in contact with a macroblock currently being deblocked; A third input buffer configured to temporarily store data generated when an edge filter deblocks the data of the macroblock unit into a column unit or a row unit composed of a plurality of subblocks; A fourth input buffer which temporarily stores data in which the deblocking process of the vertical direction is completed during the deblocking process; And And a fifth input buffer configured to temporarily store original data of a macroblock currently being deblocked. The method of claim 2, Deblocking filter, characterized in that the data stored in the input buffer is stored in a YUV (luminance, color) method. An apparatus for removing a blocking effect generated by decoding an image frame encoded in macroblock units, An input buffer for temporarily storing data input from the outside or data generated in a process of removing a blocking effect; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And And data storage means for combining data temporarily stored in the output buffer in macro block units. And a selection circuit for selectively transmitting data output from the input buffer to the edge filter according to a predetermined rule. An apparatus for removing a blocking effect generated by decoding an image frame encoded in macroblock units, An input buffer for temporarily storing data input from the outside or data generated in a process of removing a blocking effect; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And And data storage means for combining data temporarily stored in the output buffer in macro block units. The selection circuit A first selection circuit which receives data output from the first input buffer, the second input buffer, and the third input buffer; And And a second selection circuit configured to receive data output from the fourth input buffer and the fifth input buffer. An apparatus for removing a blocking effect generated by decoding an image frame encoded in macroblock units, An input buffer for temporarily storing data input from the outside or data generated in a process of removing a blocking effect; An edge filter which applies the data output from the input buffer to a deblocking algorithm to filter the boundary region of the macro block; A data alignment circuit for selectively converting a matrix alignment state of the data filtered through the edge filter; An output buffer for temporarily storing data output from the data alignment circuit; And And data storage means for combining data temporarily stored in the output buffer in macro block units. The data stored in the output buffer is stored in the data storage means through a direct memory access controller (DMAC). delete A method of removing a blocking effect caused by decoding an image frame encoded in macroblock units, (1) temporarily storing data of a predetermined area input from the outside and data of the predetermined area generated in a process of removing a blocking effect in a plurality of input buffers; (2) filtering the boundary regions in the vertical direction and the horizontal direction using data output from the plurality of input buffers; (3) selectively converting the matrix alignment of the filtered data through step (2); (4) temporarily storing data generated in the step (3) in an output buffer; And (5) combining the data output from the output buffer in units of macroblocks and storing the data in the data storage means; The plurality of input buffers include a first input buffer, a second input buffer, a third input buffer, a fourth input buffer, and a fifth input buffer. The second step is Among the macroblocks that have already been deblocked, the first input buffer receives the boundary area data in contact with the macroblock currently deblocking in the vertical direction, and the second area receives the border area data in contact with the macroblock currently deblocking in the horizontal direction. A third input buffer for temporarily storing data generated during buffering and deblocking processing, and a fourth input buffer for data in which the deblocking processing in the vertical direction is completed during the deblocking processing, and a fifth input of original data of the macroblock being deblocked. Deblocking method characterized in that the storage in the buffer. The method of claim 8, The data stored in the fifth input buffer is stored in a YUV (luminance, color) method. A method of removing a blocking effect caused by decoding an image frame encoded in macroblock units, (1) temporarily storing data of a predetermined area input from the outside and data of the predetermined area generated in a process of removing a blocking effect in a plurality of input buffers; (2) filtering the boundary regions in the vertical direction and the horizontal direction using data output from the plurality of input buffers; (3) selectively converting the matrix alignment of the filtered data through step (2); (4) temporarily storing data generated in the step (3) in an output buffer; And (5) combining the data output from the output buffer in units of macroblocks and storing the data in the data storage means; (1-1) sequentially selecting and transmitting the multiple data output from the input buffer according to a predetermined rule. A method of removing a blocking effect caused by decoding an image frame encoded in macroblock units, (1) temporarily storing data of a predetermined area input from the outside and data of the predetermined area generated in a process of removing a blocking effect in a plurality of input buffers; (2) filtering the boundary regions in the vertical direction and the horizontal direction using data output from the plurality of input buffers; (3) selectively converting the matrix alignment of the filtered data through step (2); (4) temporarily storing data generated in the step (3) in an output buffer; And (5) combining the data output from the output buffer in units of macroblocks and storing the data in the data storage means; The data stored in the output buffer is stored in the data storage means through a direct memory access method (DMA).

Images