KR20100138735A - Video encoding and decoding apparatus and method using context information-based adaptive post filter - Google Patents

Abstract

PURPOSE: A video encoding and decoding apparatus and a method using context information based adaptive post filter are provided to improve the efficiency of video encoding. CONSTITUTION: A reconstitution video generator(120) creates the reconstructed image using the differential signal and the restored prediction frame. The reconstitution video generator applies the deblocking filter. A post filter unit(130) decides whether to apply the fixed post filter based on the contextual information of the reconstructed image to the reconstructed image. The post filter unit decides whether to apply the post filter adaptively based on the contextual information of the reconstructed image to the reconstructed image. The post filter unit applies the fixed post filter or the post filter which was adaptively generated to the reconstructed image.

Description

Video Encoding / Decoding Apparatus Using Context Information-Based Adaptive Post Filter and Its Method {Video Encoding and Decoding Apparatus and Method using Context Information-based Adaptive Post Filter}

Embodiments of the present invention relate to a video encoding / decoding apparatus and a method thereof, and more particularly, to video encoding using a context-based adaptive post filter, which can improve the efficiency of video encoding by improving the performance of a post filter. / Decryption apparatus and method thereof.

The present invention is derived from the research conducted as part of the IT source technology development project of the Ministry of Knowledge Economy and the Ministry of Information and Communication Research and Development. [Task management number: 2008-F-011, Project name: Next-generation DTV core source technology development].

Conventional coding standard techniques for video compression include H.261, H.263, MPEG-1, MPEG-2, MPEG-4, and H.264. Conventional coding standard techniques are slightly different, but generally consist of similar structures including motion estimation and compensation, transform coding, and entropy encoding. In particular, H.264 is known to minimize block boundary distortion in the reconstructed image, thereby improving the overall coding efficiency by enabling more accurate prediction in motion estimation and compensation as well as subjective picture quality improvement. Such a deblocking filter shows good performance in a low bitrate image, but has little performance in a high quality image, or rather has a problem of degrading encoding performance.

In addition, the post filter, which is an issue in recent standardization, is a filter that minimizes an error between a reconstructed image and an original image, and when applied like a deblocking filter in a high quality image, it has an effect of increasing encoding efficiency.

However, since the conventional post filter is adaptively generated during the encoding process, all of the coefficients of the generated filter have to be transmitted. Therefore, the conventional post filter has a problem of lowering the compression efficiency.

Embodiments of the present invention provide a video encoding / decoding apparatus and method using an adaptive post filter based on context information, which can improve video encoding efficiency through performance improvement of a post filter.

In addition, embodiments of the present invention, a video encoding apparatus and method using an adaptive post filter based on context information that can increase the coding efficiency by enabling more accurate prediction in the motion estimation and motion compensation process during video encoding To provide.

 An apparatus for encoding a video according to an embodiment of the present invention includes an encoder for encoding a difference signal between an original video to be currently encoded and a predicted video for the original video, a coded differential signal, and a reconstructed differential signal. And a reconstructed image generator to generate a reconstructed image using the predicted image and apply a deblocking filter, and apply a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image. It determines whether or not based on the context information (Context) of the reconstruction image, and includes a post filter unit for applying the fixed post filter or the adaptively generated post filter to the reconstruction image.

In this case, the post filter unit may include a context information analyzer configured to analyze context information of the reconstructed image in image block units, and when the context information belongs to a preset context information set, the fixed post filter in the reconstructed image unit in image blocks. And a post filtering unit for applying the adaptively generated post filter to the reconstructed image in units of image blocks if the context information does not belong to a preset context information set, and to the reconstructed image in units of image blocks. It may include a filter information generator for providing the identification information for the post filter to the encoder.

In this case, the post-filtering unit may apply the fixed post filter or the adaptively generated post filter to the reconstructed image in units of image blocks, overlapping the inside of the image block and the boundary of the image block.

In this case, the post-filtering unit may divide the image block into a plurality of partitions based on the context information, and apply the fixed post filter or the adaptively generated post filter to each of the plurality of partitions.

In this case, when the post filter applied to the reconstructed image is the fixed post filter, the identification information of the post filter is a filter index for distinguishing the applied post filter, and the post filter applied to the reconstructed image is adaptive. In the case of the post filter generated by, it is the filter coefficient of the applied post filter.

According to an embodiment of the present invention, a video decoding apparatus includes: a decoder configured to decode an input bit string to restore a differential signal, and a reconstructed image to generate a reconstructed image using the differential signal and the predicted image and apply a deblocking filter. Whether to apply a generator and a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image is determined based on the context information of the reconstructed image. And a post filter for applying the post filter or the adaptively generated post filter to the reconstructed image.

According to an embodiment of the present invention, a video encoding method includes encoding a difference signal between a current video to be encoded and a prediction video of the original video, restoring the encoded difference signal, and restoring the restored difference. Generating a reconstructed image using a signal and the predicted image, applying a deblocking filter, and applying a fixed post filter to the reconstructed image or applying an adaptively generated post filter to the reconstructed image Determining based on the context information of the reconstructed image, and applying the fixed post filter or the adaptively generated post filter to the reconstructed image.

According to an embodiment of the present invention, a video decoding method includes: restoring a differential signal by decoding an input bit stream, generating a reconstructed image by using the difference signal and a predicted image, and applying a deblocking filter; It is determined whether the applied post filter is applied to the reconstructed image or an adaptively generated post filter is applied to the reconstructed image based on the context information of the reconstructed image, and the fixed post filter or the And applying an adaptively generated post filter to the reconstructed image.

According to embodiments of the present invention, the efficiency of video encoding may be improved by improving the performance of the post filter.

Further, according to embodiments of the present invention, encoding efficiency can be increased by enabling more accurate prediction in motion estimation and motion compensation during video encoding.

1 illustrates a configuration of a video encoding apparatus according to an embodiment of the present invention.
FIG. 2 shows a configuration example of the post filter unit shown in FIG. 1.
3 is a flowchart illustrating a post filtering method based on context information according to an exemplary embodiment.
4 illustrates a configuration of a video encoding apparatus according to another embodiment of the present invention.
5 shows an application example of a post filter according to an embodiment of the present invention.
6 shows another example of a post filter application according to an embodiment of the present invention.
7 shows an example of transmitting filter information according to an embodiment of the present invention.
8 shows an example of a configuration of a decoding apparatus according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 illustrates a configuration of a video encoding apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the video encoding apparatus 100 includes an encoder 110, a reconstructed image generator 120, and a post filter 130.

The encoder 110 encodes a residual signal between the original image to be encoded currently and the prediction image for the original image. In this case, the encoder 110 may encode the differential signal through a discrete cosine transform (DCT), quantization, entropy encoding, or the like.

The reconstructed image generator 120 reconstructs the encoded differential signal and generates a reconstructed image using the reconstructed differential signal and the predicted image. In this case, the reconstructed image generator 120 may generate the reconstructed image through inverse quantization, inverse discrete cosine transform, and deblocking filter.

The post filter unit 130 applies the post filter in units of image blocks of the reconstructed image according to the context information of the reconstructed image. In this case, the post filter unit 130 may be configured of one or a plurality of processors.

For example, the post filter unit 130 determines whether to apply a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image. The generated post filter is applied to the reconstructed image. In this case, the post filter unit 130 determines whether to apply a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image based on the context information of the reconstructed image. do.

In addition, the post filter unit 130 may determine whether to use a plurality of fixed post filters or adaptively generate and use post filters according to the context information. The fixed post filter is a filter obtained statistically based on a number of images, and since there may be a plurality of filters, an appropriate filter may be selected according to context information. In addition, a filter may be adaptively generated and applied to an image block having context information different from general context information. In this case, there may be various methods for adaptively generating a filter. For example, there may be a method of generating a filter having the smallest cost necessary to minimize an error between the reconstructed image and the original image. In addition, the size of the post filter may be determined as NⅹN without being fixed as one, and may be variously determined according to the context information of the reconstructed image.

FIG. 2 shows a configuration example of the post filter unit shown in FIG. 1.

2, the post filter unit 130 may include a context information analyzer 210, a post filtering unit 230, and a filter information generator 250.

In analyzing the context information of the reconstructed image, the context information analyzer 210 analyzes the context information in units of image blocks of the reconstructed image. Here, the context information may be all features that may be provided for each image block, such as macroblock type, SKIP, quantization coefficient, coded block pattern (CBP) information, DCT coefficient, and information on neighboring blocks.

The post filtering unit 230 applies the fixed post filter to the reconstructed image in units of image blocks when the context information belongs to a preset context information set. In this case, the post filtering performing unit 230 applies the adaptively generated post filter to the reconstructed image in units of image blocks if the context information does not belong to a preset context information set. In this case, the preset context information set means, for example, n context information corresponding to each of n (n = 1, 2, 3.) fixed post filters. Accordingly, the post filtering performing unit 230 may perform post filtering using a corresponding fixed post filter when the context information of the image block is any one of n preset context information.

Meanwhile, the post-filtering unit 230 applies the fixed post filter or the adaptively generated post filter to the reconstructed image in units of image blocks. As shown in FIG. 5, the internal and image blocks of the image block are shown. It can be applied to the boundary of.

Also, as illustrated in FIG. 6, the post filtering unit 230 divides the image block into a plurality of partitions based on the context information, and the fixed post filter or adaptively applies to each of the plurality of partitions. You can apply the created post filter.

The filter information generation unit 250 provides the encoding unit 110 with identification information on the post filter applied to the reconstructed image. Here, the identification information on the post filter is a filter index for distinguishing the applied post filter when the post filter applied to the reconstructed image is a fixed post filter. The identification information of the post filter is a filter coefficient of the applied post filter when the post filter applied to the reconstructed image is an adaptively generated post filter. In this case, the encoder 110 may distinguish and encode the identification information of the post filter from the differential image, and transmit the identification information of the encoded post filter to the decoding apparatus. Therefore, according to an embodiment of the present invention, when the fixed post filter is applied to the reconstructed image, the encoding apparatus 100 transmits only the filter index to the decoding apparatus. In addition, according to an embodiment of the present invention, when the adaptively generated post filter is applied to the reconstructed image, the encoding apparatus 100 may transmit the filter coefficients to the decoding apparatus. In this case, the encoding apparatus 100 may transmit the filter coefficient information to the decoding apparatus by using the symmetric characteristic of the filter coefficients.

3 is a flowchart illustrating a post filtering method based on context information according to an exemplary embodiment. The method shown in FIG. 3 may be performed by the post filter unit 130 of FIG. 1.

Referring to FIG. 3, in step 310, the post filter unit 130 analyzes context information of a reconstructed image. In this case, the filter unit 130 may analyze the context information of the reconstructed image in units of image blocks.

In operation 320, the post filter unit 130 determines whether the context information belongs to a preset context information set.

If the context information belongs to the preset context information set, the post filter unit 130 applies the fixed post filter to the reconstructed image in step 330. That is, the post filter unit 130 may select one post corresponding to the context information among the plurality of post filters, and apply the selected post filter to the reconstructed image.

If the context information does not belong to the preset context information set, the post filter unit 130 may apply the adaptively generated post filter to the reconstructed image (340).

In operation 350, the post filter unit 130 generates filter information. In this case, the generated filter information may be provided to the encoder 110.

4 illustrates a configuration of a video encoding apparatus according to another embodiment of the present invention.

Referring to FIG. 4, the video encoding apparatus 400 includes an image predictor 410, a difference signal generator 420, an encoder 430, a reconstructed image generator 440, and a post filter 450. do.

The image predictor 410 generates a predicted image of the current frame 401 from the reference frame 403 or the reconstructed image. In this case, the current frame 401 corresponds to the original image to be currently encoded, and the reference frame 403 corresponds to the reference image to which post filtering is applied.

As illustrated in FIG. 4, the image predictor 410 may include a motion predictor 411, an intra prediction selector 413, a motion compensator 415, and an intra predictor 417. have. The motion predictor 411 predicts the motion of the current frame 401 from the reference frame 403, and the motion compensator 415 compensates for the motion of the reference frame 403. The intra prediction selecting unit 413 controls the image predicting unit 410 to select intra prediction when generating the prediction image in the intra mode. In addition, the intra predictor 417 generates a predicted image of the current frame 401 from the reconstructed image by intra prediction.

The difference signal generator 420 subtracts the prediction image generated by the image predictor 410 from the current frame 401 to generate the difference signal Dn. That is, when one image block is encoded in the current frame 401, the image predictor 410 generates a predicted image block of the image block of the current frame 401 to be encoded, and generates a differential signal generator ( 420 may generate a difference signal for the image block of the current frame 401.

The encoder 430 encodes a difference signal between the original video to be currently encoded and the predicted video for the original video.

As illustrated in FIG. 4, the encoder 430 may include a discrete cosine transform unit 431 for discrete cosine transforming a differential signal, a quantization unit 433 for performing quantization, and a reordering unit 435 for rearranging quantized data. And an entropy encoder 437 for entropy encoding the reordered data.

The reconstructed image generator 440 reconstructs the encoded difference signal, generates a reconstructed image using the reconstructed differential signal D'n, and the predicted image uFn`, and applies a deblocking filter.

The reconstructed image generator 440 may include an inverse quantization unit 445 for performing inverse quantization, an inverse discrete cosine transform unit 443 for performing inverse discrete cosine transform, an adder 441, and a D. It may be configured to include a blocking filter 447. That is, the reconstructed image generating unit 440 generates a reconstructed image and allows the reconstructed image to be used to generate a predicted image for the next image block or the next frame.

The post filter unit 450 applies the post filter to the reconstructed image based on the context information of the reconstructed image output from the reconstructed image generating unit 440. In this case, as shown in FIG. 4, the post filter unit 450 may include a context information analyzer 451, a post filtering unit 453, and a filter information generator 455. In this case, the functions of the context information analyzer 451, the post filtering unit 453, and the filter information generator 455 are respectively the context information analyzer 210 and the post filtering unit 230 shown in FIG. 2. And the filter information generator 250.

5 shows an application example of a post filter according to an embodiment of the present invention.

Referring to FIG. 5, the post filter unit according to an embodiment of the present disclosure applies a fixed post filter or an adaptively generated post filter to a reconstructed image in units of image blocks. It can be applied to the interior 501, 505 and the boundary 503 of the image blocks 510, 520. In this case, the post filters applied to the interiors 501 and 505 of the image blocks 510 and 520 and the boundary 503 of the image blocks 510 and 520 may be the same or different. In addition, the video block unit is not limited to 4 × 4 blocks, but may be determined as N × N, such as 8 × 8 and 16 × 16. In addition, an offset value to be added may be determined differently according to the position of the pixel, and a range of the offset value to be added may be determined to be -1 to 1 or a larger value.

6 shows another example of a post filter application according to an embodiment of the present invention.

6, a post filter unit according to an embodiment of the present disclosure divides an image block into a plurality of partitions based on context information, and fixes a post filter or an adaptively generated post for each of the plurality of partitions. You can apply filters. That is, one post filter may be applied to one image block 610, but different post filters may be applied to the plurality of partitions 601 and 603, respectively. In this case, partitions may be divided through Edge or similar block information in the reconstructed image. In addition, the number of partitions is not limited to two, and may be divided into several partitions. In addition, according to the partition classification, a post filter of the same size may be applied to each partition, or a post filter of a different size may be applied to each partition.

7 shows an example of transmitting filter information according to an embodiment of the present invention.

Referring to FIG. 7, the encoding apparatus according to an embodiment of the present invention may transmit adaptive filter coefficients or filter index information for a fixed filter to each decoding block, such as 710, for each macroblock, as shown in 710. Filter index information about the specific filter coefficient or the fixed filter may be transmitted to the decoding apparatus. In this case, the size of the post filter is not fixed to one, but is determined as NⅹN, and may be variously determined according to the context information of the reconstructed image.

8 shows an example of a configuration of a decoding apparatus according to an embodiment of the present invention.

Referring to FIG. 8, the decoding apparatus 800 includes a decoder 810, a reconstructed image generator 820, and a post filter 830. It has a configuration corresponding to the encoding apparatus shown in FIG. 1 or 4.

The decoder 810 decodes the input bit string and restores the difference signal. In this case, the decoder 810 performs an entropy decoder 811 for performing entropy decoding, a reordering unit 813 for rearranging the entropy decoded data, an inverse quantization unit 815 for performing inverse quantization, and inverse discrete cosine transform. The inverse discrete cosine transform unit 817 may be configured to include.

The reconstructed image generator 820 generates a reconstructed image by using the difference signal Dn` and the predicted image uFn`. As shown in FIG. 8, the reconstructed image generator 820 may include a motion compensator 821 performing motion compensation from a reference frame 801, an intra predictor 823 performing intra prediction from a reconstructed image, And an adder 825 for generating a reconstructed image using the predicted image and the difference signal, and a deblocking filter unit 827 for removing a blocking error in the reconstructed image.

The post filter unit 830 determines whether to apply a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image based on the context information of the reconstructed image. The fixed post filter or the adaptively generated post filter is applied to the reconstructed image.

As illustrated in FIG. 8, the post filter unit 830 may include a context information analyzer 831, a filter information receiver 833, and a post filtering unit 835.

The context information analyzer 831 analyzes context information of the reconstructed image in units of image blocks. In this case, the context information analyzer 831 may perform the same function as the context information analyzers 210 and 451 of the encoding apparatus.

The filter information receiver 833 receives the filter information included in the input bit string from the decoder 810.

If the context information of the reconstructed image belongs to a preset context information set, the post filtering unit 835 determines the filter information received by the filter information receiver 833 as a filter index, and displays the post filter corresponding to the filter index. Applies to reconstructed image in block unit. In addition, the post filtering unit 835 determines the filter information received by the filter information receiving unit 833 as the filter coefficient information when the context information does not belong to the preset context information set. That is, the post filtering performer 835 adaptively generates a post filter using filter coefficients, and applies the adaptively generated post filter to the reconstructed image in units of image blocks.

On the other hand, the post filtering unit 835 may perform post filtering in a manner similar to the post filtering unit 230 or 450 of the encoding apparatus. Accordingly, in applying the fixed post filter or the adaptively generated post filter to the reconstructed image in units of image blocks, the post filtering unit 835 includes the inside and the image blocks of the image block as shown in FIG. 5. It can be applied to the boundary of.

In addition, as illustrated in FIG. 6, the post filtering unit 835 divides the image block into a plurality of partitions based on the context information, and applies the fixed post filter or adaptively to each of the plurality of partitions. You can apply the created post filter.

Method according to an embodiment of the present invention is implemented in the form of program instructions that can be executed by various computer means may be recorded on a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination.

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

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

Claims (1)

An encoder which encodes a difference signal between an original image to be currently encoded and a prediction image of the original image;
A reconstructed image generation unit reconstructing an encoded differential signal, generating a reconstructed image using the reconstructed differential signal and the prediction image, and applying a deblocking filter; And
Whether to apply a fixed post filter to the reconstructed image or to apply an adaptively generated post filter to the reconstructed image is determined based on the context information of the reconstructed image, and the fixed post filter or And a post filter unit applying the adaptively generated post filter to the reconstructed image.

Images