KR20060084952A - Method and apparatus of video decoding based on interleaved chroma frame buffer - Google Patents

Abstract

The present invention relates to an interleaving based video decoding method of a chroma signal CbCr and an apparatus thereof.

According to the present invention, a chromaticity signal storage means for interleaving and storing the chromatic signal CrCb, decoding means for reading out the chromaticity signal CbCr from the chromaticity signal storage means stored in the interleaving form and performing decoding based on motion compensation, the decoding means And a decoded chromaticity signal storage means for storing the chromaticity signal CbCr in an interleaving form, and storing the chromaticity signal CbCr in an interleaving form in a chromaticity frame buffer and storing the chromaticity signal CbCr in an interleaving form. ) Is sequentially read and the decoding is performed based on motion compensation for the corresponding block, and the decoded chroma signals are stored in a frame buffer in the form of interleaving to separate frame data of the two chroma signals. It can reduce the process time required to save and use in the memory area, and the decoding speed It was to improve performance.

Video Decoder, Motion Compensation, Interleaving

Description

Interleaving based video decoding method of chromaticity signal and device therefor {METHOD AND APPARATUS OF VIDEO DECODING BASED ON INTERLEAVED CHROMA FRAME BUFFER}

1 is a diagram illustrating an example of a method of using a frame buffer of a chroma signal CbCr in a conventional video decoding method.

2 is a block diagram of an H.264 video decoder

3 is a diagram illustrating an example of a method of using a frame buffer of a chroma signal CbCr in a video decoding method according to the present invention.

<Explanation of symbols for the main parts of the drawings>

200: external memory 210: CbCr interleaved frame buffer

220: CbCr block 300: internal memory

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video decoding method and an apparatus thereof. In particular, a video codec stores a chroma signal in a frame buffer based on interleaving and uses the same for a chroma signal. It performs motion compensation and decoding, and stores the decoded chroma signal in the frame buffer in the form of interleaving and outputs it, thereby reducing the computation amount of the decoder and improving the decoding speed and performance. A decoding method and apparatus therefor.

Image compression has a compression standard such as MPEG1, MPEG2, MPEG4, H.263, H.264, and this image compression technique is used in many applications such as a video player, VOD, video telephony, and digital multimedia broadcasting (DMB). In addition, with the development of wireless communication such as 2.5G / 3G, image transmission is commercialized in a wireless mobile base.

Most video compression standards, such as MPEG or H.26x video codecs, adopt compression based on motion estimation and compensation and transformation. In such motion compensation based encoding, motion vector information of each block must be encoded and transmitted, and the compression efficiency may vary greatly depending on how the motion vector is encoded.

In the general process of encoding an image, DCT (Discrete Cosine Transform) processing of a digital image signal is performed, quantization of transform coefficients to perform variable length coding (VLC), and reconstruction of quantized DCT coefficients by inverse quantization and inverse DCT conversion. A video is stored in a memory, a motion vector is calculated using the reconstructed video and the next frame image stored in the memory. The motion vector is variable-length coded, and a bit stream is transmitted along with the encoded video information. The method of decoding an image is performed in the reverse order of the encoding process.

As described above, a method of compressing and encoding an image is performed using spatial redundancy and / or temporal redundancy. In the case of a video codec based on temporal redundancy, temporal redundancy with a previous frame can be efficiently removed by using a motion vector of a macro block. For example, the macro block most similar to the macro block of the current frame is found in the reference frame. The two vectors are based on the similarity between the macro block of the current frame and the macro block of the previous frame. By finding the macroblock where the square of the distances is the minimum, the redundancy between the frames is effectively removed, thereby increasing the compression efficiency.

In accordance with the video compression coding method and hardware and software-based enhancements, a video / telephony system capable of video-calling as well as satellite / terrestrial DMB are being introduced.

Among them, H.264, a high-efficiency compression technology, has improved motion prediction and compensation, block conversion of various block sizes, compared to H.263, which is a video compression standard, in terms of flexibility and video coding efficiency, which can easily adapt to various network environments. In-loop deblocking filter, improved entropy coding, NAL structure for adapting to various networks, multiple sub-block based reference pictures, and so on.

In the video compression standards listed above, a color image signal is represented by one luminance signal component Luma and two chroma signal components Chroma. In the commonly used YCbCr format, Y represents a luminance signal, and Cb and Cr represent a chroma signal.

In the implementation of the video decoder, each Y, Cb, Cr frame data is usually stored in a separate memory area. In the process of the decoder, the signal components are processed independently of each other in the process of intra prediction, variable length decoding (VLD), and transform, and the separated frame buffer structure is independent of each other. Although it may be considered suitable for processing, in some cases, the decoding performance is deteriorated due to the frame buffer having such a separate structure.

Meanwhile, as described above, most video codecs use a motion compensation technique. To decode the current block, a prediction value of the current block is obtained by reading pixel values at a position indicated by a motion vector in a previously decoded reference frame. . In general, in the case of using the separated chroma signal frame buffer as described above in the implementation of the video decoder, a separate memory read time is required for Cb and Cr, and the decoded block is transmitted to the memory. In the case of using a separate chroma signal frame buffer, write time is similarly required.

Figure 1 shows the structure of a conventional chroma signal frame buffer and a method of using the same.

As shown in FIG. 1, the structure of the frame buffer is stored in the Cb and Cr frame data in separate regions on the external memory 100, that is, the regions separated by the Cb frame buffer 101 and the Cr frame buffer 102, respectively. In this case, separate memory accesses are required to obtain Cb pixel data and Cr pixel data. That is, the Cb block is read from the Cb frame buffer 101 at the position indicated by the chroma vector and stored in the internal memory 110, and the same chroma vector is used in the Cr frame buffer 102. Since a necessary block (Cr block) is read and stored in the internal memory 110, it takes a relatively long time for memory access.

Therefore, due to these factors, it takes a long time to decode a chromaticity signal or the like based on motion compensation in the existing video codec, which causes a problem in that decoding speed and performance are deteriorated.

SUMMARY OF THE INVENTION An object of the present invention is to provide an interleaving based video decoding method and apparatus for chromaticity signal that can improve decoding performance by shortening memory access time when compensating for motion of a chroma signal in a video codec.

An object of the present invention is to compensate for motion by storing chromaticity signals in one frame buffer based on interleaving technique and reading chromaticity signal blocks stored in interleaving format without using chromaticity signal frame buffer of structure separated from video codec. And interleaving-based video decoding method and apparatus for chromaticity signal to reduce memory access time and improve decoding speed and performance by performing decoding.

An object of the present invention is to store the CbCr chromaticity signal in the form of interleaving in one frame buffer without storing the frame data of the two chromaticity signal components of Cb and Cr when decoding the frame data in a separate memory area. The present invention provides an interleaving based video decoding method and apparatus for chromaticity signals to improve motion compensation speed and decoding performance of a video codec using direct memory access (DMA) based memory access.

An object of the present invention is to store the frame data of the two chroma signal components of Cb, Cr when using the inter-leaved frame buffer when storing the frame data, and to use the motion compensation and decoded chroma signal components The present invention provides an interleaving-based video decoding method and apparatus therefor for reducing chromaticity by improving the decoding performance of a decoder by storing and outputting the result in a memory in an interleaving form.

In order to achieve the above object, an interleaving based video decoding method of a chroma signal according to the present invention includes a motion codec based on motion compensation and transformation.

Storing the chroma signal CbCr in the chroma frame buffer in an interleaved form without separately storing the chroma signals CbCr; Sequentially reading the chroma signals CbCr stored in the interleaving form and performing decoding based on motion compensation for the corresponding block; Storing and decoding the decoded chroma signal in a frame buffer in an interleaving form; Characterized in that comprises a.

In the video decoding method of the present invention, the chroma signal CbCr stored in the interleaving mode is transmitted between memories based on DMA.

In the video decoding method of the present invention, the chroma signal CbCr stored in the interleaving mode is characterized in that Cb and Cr are alternately stored in a memory.

Also, in the video decoding method of the present invention, the chroma signals CbCr stored in the interleaving form are alternately stored in the memory of Cb and Cr, so that memory access for motion compensation and storage of the decoded chroma signals is performed on the CbCr frame. Starting from one location, it can be performed by continuous reading and writing.

In addition, the interleaving-based video decoding apparatus of the chroma signal according to the present invention for achieving the above object, in the video codec based on motion compensation and conversion,

Chromaticity signal storage means for storing the chromaticity signal CrCb in an interleaved form; Decoding means for reading the chroma signal CbCr from the chroma signal storage means stored in the interleaving form and performing decoding based on motion compensation; And decoded chromaticity signal storage means for storing the decoded chromaticity signal CbCr in an interleaving form.

In addition, in the video decoding apparatus of the present invention, the motion compensation and decoded chroma signal storage means is a frame buffer provided in an external memory, and the decoding means accesses the frame buffer by DMA, and the chroma signal and motion decode for motion compensation are decoded. Characterized in that the transmission of the chroma signal.

Further, in the video decoding apparatus of the present invention, the chroma signal CbCr stored in the interleaving form is characterized in that Cb and Cr are alternately stored in a memory.

Hereinafter, an embodiment of a video decoding method and apparatus thereof according to the present invention will be described with reference to the accompanying drawings.

In general, a processor such as a general-purpose RISC or DSP commonly includes an internal memory having a high access speed and a small size. However, since the frame data is large in a video decoder, it is not stored in the internal memory. It is difficult and is usually stored in a frame buffer on slow external memory.

SDRAM, which is often used as an external memory in embedded systems, takes a long time to access a specific location and read the first data, while the speed of continuous data read in adjacent areas after reading the data once is first Compared to the very fast feature.

Accordingly, in the present invention, two chroma signals are stored using an interleaved frame buffer on a pixel-by-pixel basis without storing the chroma signals Cb and Cr separately based on a separate frame buffer structure, thereby ensuring a faster access speed.

In the present invention, an external memory access is performed based on DMA, and an H.264 decoder implemented with a DSP stores two chroma signals of Cb and Cr using an interleaved frame buffer in units of pixels. The above problems are solved by the video decoding method and apparatus according to an embodiment of the present invention.

In the embedded environment, it usually takes a long time to access the frame buffer stored in the external memory, so the speed is increased by using DMA. However, in the case of the conventional chroma signal frame buffer, separate DMAs must be driven for Cb and Cr, which consumes a lot of CPU cycles due to the overhead of variable setting or interrupts. In particular, in the case of the H.264 decoder, which can reduce the size of the chromaticity signal block to 2x2, the overhead caused by driving DMA separately in Cb and Cr is greater than the amount of data actually read from the memory. In addition to the motion compensation step, when the decoded block is transferred to the external memory using DMA, CPU cycle consumption is high as in the case of using the separated chroma signal frame buffer.

In the existing video codec, the luminance signal data is stored in the Y buffer among the data constituting each frame, and the chroma signal data is stored in the Cb and Cr buffers. In the case of the most popular YCbCr 4: 2: 0 format, the frame sizes of Cb and Cr are 1/2 each of the length and width of the frame size of Y. That is, the 2 x 2 pixel region of the Y component corresponds to the 1 x 1 pixel region of the chroma signal components Cb and Cr. In this case, the chroma signal motion vector has a size corresponding to 1/2 of the luminance signal motion vector in the horizontal and vertical directions, respectively, and this vector is commonly used for Cb and Cr. That is, the positions accessed by the Cb and Cr frame buffers are relatively the same at the beginning of each frame buffer.

Therefore, based on this, even if the two chroma signals are stored in the frame buffer by the interleaving technique, there is no obstacle in accessing the block and using it for motion compensation. When used as a memory, the frame buffer access for motion compensation starts at one location on the CbCr frame and can improve its access speed through continuous fast reads.

First, referring to Figure 2 as an embodiment of the decoder to which the present invention is applied, the configuration and operation of the H.264 decoder will be described.

As shown in FIG. 1, an entropy decoder 1 performs entropy decoding of input video data based on a NAL unit, and rearranges the corresponding video data by a reorder 2. The data recovered through the inverse quantizer (Q ⁻¹ ) 3 and the inverse DCT (T ⁻¹ ) 4 is input to the adder 5.

The output of the adder 5 is stored and output in the frame memory 7 reconstructed through the filter 6, and the frame image of the reference frame memory 8 of the previous frame is moved in the motion compensator 9 In this case, the intra frame predictor 10 performs motion prediction on the output of the adder 5.

The switching unit 11 supplies the current frame image by switching the output of the motion compensator 9 and the output of the intra frame predictor 10 to the adder 5 in response to an intra or inter mode. To be decoded.

A software solution can be used as a solution for video decoding based on H.264 in a mobile terminal device or a hand-held device. In this case, the H.264 decoder can be implemented in a digital signal processor (DSP) based system, which is shown in FIG.

DSP-based processors are typically designed to use direct memory access (DMA). Therefore, in using this to compensate for the motion of the current macro block, the reference image stored in the external memory 8 is accessed with the motion vector and the reference index of each sub-block, and the DMA data is stored in the reference block buffer in the internal memory. The motion compensation is performed with the data, and the motion compensation and decoded frame data are stored in the external memory 7 and finally output as a reconstructed image. In this case, there is an advantage that an environment in which other tasks can be performed simultaneously while transferring data using DMA is secured.

FIG. 3 schematically illustrates the external memory structure and the concept of data transfer to the internal memory in the video decoder implemented by the DSP based on DMA.

As shown in FIG. 3, the chroma signal frame buffer according to the present invention is allocated to the external memory 200. The structure of the chroma signal frame buffer 210 allocated to the external memory 200 intersects Cb and Cr data by one pixel to form a CbCr frame buffer.

When the CbCr frame buffer 210 is configured in the form of interleaving as described above, the access of the frame buffer for motion compensation reads the corresponding CbCr block 220 and transmits it to the internal memory 300, starting from one position on the CbCr frame. This can be done by successive reads.

Here, the number of pixels actually required is the same in both cases, but the number of accesses to an arbitrary position is halved as compared to the case where it is not interleaved. Therefore, the number of overhead cycles generated by approaching an arbitrary position is reduced, thereby reducing the time required for motion compensation.

In addition, when the interleaved chroma signal frame buffer 210 is used, even when the decoded chroma signal block is transmitted to the frame buffer, the Cb frame buffer and the Cr frame buffer need not be accessed separately, and the CbCr frame does not need to be accessed separately. Since successive writes are performed starting from one position of the buffer, the number of overhead cycles is similarly reduced in this case, thereby improving the decoding output speed.

In both cases of storing motion compensation or a decoded block, if a decoder is implemented using DMA, a large amount of overhead is required for specifying an address for starting DMA or a block size, and a large number of blocks are used. As the number of DMA interrupts increases, it acts as an overhead for the CPU. However, when the interleaved frame buffer is used as in the present invention, the DMA overhead for reading and writing the chroma signal can be greatly reduced.

According to the present invention, a frame buffer for motion compensation is read by interleaving and storing frame data of two chroma signal components of Cb and Cr in a video decoder using a motion compensation technique such as H.263, H.264, MPEG1, MPEG2, MPEG4, etc. The number of cycles required to come and the number of cycles required to store the decoded block in the frame buffer can be reduced, thereby reducing the calculation amount of the overall decoder, thereby improving the decoding speed and performance.

Claims (7)

In video codec based on motion compensation and conversion, Storing the chroma signal CbCr in the chroma frame buffer in an interleaved form without separately storing the chroma signals CbCr; Sequentially reading the chroma signals CbCr stored in the interleaving form and performing decoding based on motion compensation for the corresponding block; Storing and decoding the decoded chroma signal in a frame buffer in an interleaving form; Interleaving based video decoding method of the chroma signal, characterized in that comprises a. The method of claim 1, wherein the chroma signal (CbCr) stored in the interleaving form is transmitted between memories based on DMA. The video decoding method of claim 1, wherein the chroma signals CbCr stored in the interleaving form are alternately stored in the memory, Cb and Cr, respectively. The method of claim 1, wherein the chroma signal CbCr stored in the interleaving form is alternately stored in the memory Cb and Cr, so that the memory access for motion compensation and storage of the decoded chroma signal is performed at one position on the CbCr frame. The interleaving-based video decoding method of the chroma signal, characterized in that can be performed by successive reading and writing. In video codec based on motion compensation and conversion, Chromaticity signal storage means for storing the chromaticity signal CrCb in an interleaved form; Decoding means for reading the chroma signal CbCr from the chroma signal storage means stored in the interleaving form and performing decoding based on motion compensation; And a decoded chromaticity signal storing means for storing the decoded chromaticity signal (CbCr) in an interleaving form. 6. The apparatus of claim 5, wherein the motion compensation and decoded chroma signal storage means is a frame buffer provided in an external memory, and the decoding means accesses the frame buffer by DMA to convert the chroma signal and the decoded chroma signal for motion compensation. An interleaving based video decoding apparatus of a chroma signal, characterized in that for transmitting. The apparatus of claim 5, wherein the chroma signals CbCr stored in the interleaving form are alternately stored in the memory, Cb and Cr, respectively.

Images