KR20110066523A - Apparatus and method for scalable video decoding - Google Patents

Abstract

PURPOSE: A scalable video decoding device and method thereof are provide to process each class about one part among a decoding process for real time decode by utilizing function blocks. CONSTITUTION: An inverse quantization and an inverse transform result are outputted from a layer selecting unit(330). A residual signal up-sampling unit(350) performs a residual signal up-sampling. A deblocking unit(370) deblocks a restored image. A texture up-sampling unit(380) texture-up-samples the deblocked recovered image. A restructuring unit(360) processes a process of a reconfiguring the recovered image about one class.

Description

Apparatus and method for scalable video decoding {APPARATUS AND METHOD FOR SCALABLE VIDEO DECODING}

The present invention relates to a scalable video decoding apparatus and method, and more particularly, to a scalable video decoding apparatus and method for allowing the streams for each layer can be processed simultaneously.

The present invention is derived from a study conducted as part of the IT growth engine technology development project of the Ministry of Knowledge Economy [Task control number: 2007-S-026-03, task name: MPCore platform-based multi-format multimedia SoC].

With the development of broadcasting network and high speed of wireless communication network and Internet leased line, the network environment is more diversified than before.

On the other hand, various multimedia services including video are being activated as a new paradigm. Unlike the time when only compression coding technology was developed due to the emergence of convergence convergence services, the generation, transmission and consumption environment of digital multimedia has recently changed rapidly.

At present, there are mixed transmission networks with different transmission capacities and error rates, and converged communication and broadcasting services such as DMB and IPTV services as well as existing broadcasting and communication services have been realized.

Moreover, as mobile communication and broadcasting provide mobility, various types of products with various screen sizes and computing powers have also been released.

A video coding technology widely used in multimedia services is a single layer video coding scheme that compresses and transmits at a fixed video format and bit rate. This is because it is difficult to encode in advance to suit all conditions such as transmission conditions, network characteristics, user preferences, and various performances of terminals in the recently changed network environment, thereby satisfying user's desire for seamless quality multimedia. Can't meet.

Accordingly, a transcoding method and a transrating method for adapting the format and the bit rate of the video to the conditions of each network and the performance of the terminal viewing the video have been proposed. However, these methods are not widely used in service environments requiring real-time adaptation due to the narrow range of video format and bit rate change and the high complexity.

A video encoding method that supports scalability uses a base layer and multiple enhancement layers to change spatial, temporal, or image quality changes in one compressed bitstream. Because it can support resolution or various image quality, it is one of the effective solutions for adapting multimedia transmission and consumption environment.

The existing coding standards for scalability in video encoding have been attempted based on hierarchical coding up to MPEG-2, H.263 and MPEG-4 part 2 Visual. However, they have a weak coding efficiency compared to the conventional compression coding method of encoding one layer, and have a weakness in that they cannot comprehensively support various types of scalability.

The scalable video coding included in the MPEG-4 part 10 AVC / H.264 standard compensates for this problem and provides various scalability through combination of various layers of time, space, and picture quality with good coding efficiency.

Scalable video decoders have a significantly higher complexity compared to conventional non-scalable video decoders capable of decoding only a single layer. In particular, since the decoding information of several lower layers must be used to decode the top layer in a bit stream composed of a large number of layers, the amount of computation is greatly increased.

Therefore, for some of the decoding processes for real time decoding, each layer needs to be processed simultaneously.

Accordingly, an aspect of the present invention is to provide a scalable video decoding apparatus and method for allowing each layer to be simultaneously processed for a part of a decoding process for real-time decoding by utilizing functional blocks constituting a non-scalable video decoder.

According to a first aspect of the present invention, there is provided a means for solving the above problems, comprising: a parallel processor for parallel processing of entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation for each layer of a bit stream; A hierarchy selector for sequentially selecting and outputting an output result of the parallel processor for each layer; A motion compensator for performing motion compensation using an entropy decoding result output from the layer selector; A residual signal upsampling unit configured to perform residual signal upsampling using the inverse quantization and inverse transformation results output from the layer selector; A deblocking unit to deblock the reconstructed image; A texture up sampling unit configured to texture up sample the deblocked reconstructed image output from the deblocking unit; And a reconstruction unit configured to simultaneously perform reconstruction of a reconstructed image on at least one layer by selectively using outputs of the layer selector, the motion compensator, the residual signal upsampler, and the texture upsampler. have.

The parallel processing unit includes a bit stream supply unit for dividing the bit stream into a stream for each layer; And a plurality of layer processing units for performing entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation on each of the layer-specific streams in parallel.

The layer selector may include an intrapicture prediction result selector that selects and outputs an intrapicture prediction result of a lower layer of a current reconstruction layer among intrapicture prediction results output in parallel from the parallel processor; A first inverse quantization and inverse transform result selector configured to select and output inverse quantization and inverse transform results of the lower layer among inverse quantization and inverse transform results output from the parallel processor; A second inverse quantization and inverse transform selection unit for selecting and outputting inverse quantization and inverse transformation results of the current restoration layer among inverse quantization and inverse transformation results output from the parallel processor; And an entropy decoding result selector which selects a motion vector of the current reconstruction layer or the lower layer among the entropy decoding results output in parallel from the parallel processor and outputs the motion vector to the motion compensation unit.

The deblocking unit may deblock the restored image of the lower layer output from the reconstruction unit and provide the first deblocking filter to the texture up sampling unit; And a second deblocking filter for deblocking a reconstructed image of the current reconstruction layer output from the reconstruction unit and providing the deblocking image to a frame memory.

When the inter-layer texture prediction is performed, the reconstruction unit reconstructs outputs of the first inverse quantization and inverse transform result selection unit and the intra picture prediction result selection unit to generate a reconstruction image of the lower layer, and reconstructs the reconstruction image of the lower layer. After deblocking and texture upsampling through a first deblocking filter and the texture upsampling unit, a reconstructed image of the current reconstruction layer may be generated by adding a result of inverse quantization and inverse transformation of the current reconstruction layer.

The reconstructor may reconstruct the output of the residual signal upsampler, the output of the second inverse quantization and inverse transform result selector, and the output of the motion compensator to generate a reconstructed image of the current reconstruction layer when predicting the residual signal between layers. can do.

The reconstructor may generate a reconstructed image of the current reconstruction layer by adding the output of the second inverse quantization and inverse transform result selector to a prediction image obtained by scaling the motion vector of the lower layer, when the inter-layer motion prediction is performed. can do.

According to the second aspect of the present invention, as a means for solving the above problems, the scalable video decoding method for performing inter-layer texture prediction includes entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation of a bit stream. Parallel processing for each; Generating a reconstructed image of the lower layer by selecting and reconstructing a result of inverse quantization and inverse transformation of the lower layer with respect to a current reconstruction layer and an intra picture prediction result; Obtaining a texture upsampling result by performing deblocking and texture upsampling on the reconstructed image of the lower layer; And reconstructing the reconstructed image of the current reconstruction layer by using the texture up-sampling result and the inverse quantization and inverse transformation result of the current reconstruction layer.

According to a third aspect of the present invention, as a means for solving the above problems, the scalable video decoding method for performing inter-layer residual signal prediction includes a bit stream for entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation. Parallel processing for each layer of the; Selecting a result of inverse quantization and inverse transformation of a lower layer with respect to a current reconstruction layer to perform residual signal upsampling; Selecting a result of inverse quantization and inverse transformation of a current restoration layer; And reconstructing a reconstructed image of the current reconstructed layer by using the result of performing the residual signal upsampling and the inverse quantization and inverse transform result of the current reconstructed layer.

According to the fourth aspect of the present invention, as a means for solving the above problems, the scalable video decoding method for performing inter-layer motion prediction includes entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation of a bit stream. Parallel processing for each; Selecting a motion vector included in the entropy decoding result of the lower layer with respect to the current reconstruction layer and performing motion compensation to obtain a predicted image of the current reconstruction layer; Selecting a result of inverse quantization and inverse transformation of a current restoration layer; And reconstructing the reconstructed image of the current reconstructed layer by using the predicted image of the current reconstructed layer and the result of inverse quantization and inverse transformation of the current reconstructed layer.

As described above, the apparatus and method for scalable video decoding of the present invention extends some functional blocks constituting the existing non-scalable video decoder and adds functional blocks necessary to provide scalability, thereby decoding and decoding for real-time decoding. For some of the rules, each layer can be processed simultaneously. Accordingly, the present invention provides an effect that can easily improve the performance of the scalable video decoding apparatus.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, in describing in detail the operating principle of the preferred embodiment of the present invention, if it is determined that the detailed description of the related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

In order to clearly illustrate the present invention, parts not related to the description are omitted, and like parts are denoted by similar reference numerals throughout the specification.

In addition, when a part is said to "include" a certain component, this means that it may further include other components, except to exclude other components unless otherwise stated.

Prior to describing the present invention, the configuration and operation of a general non-scalable video decoder will be described first to help understanding of the present invention.

1 is a diagram illustrating a configuration of a general non-scalable video decoder.

Referring to FIG. 1, a general non-scalable video decoder 100 receives a bit stream and receives an intra prediction mode, a transform coefficient, a motion vector, etc. required for decoding. An entropy decoder 110 that outputs a grammar syntax value, an intra picture prediction unit 120 that receives an intra prediction mode, and outputs an image predicted on the screen, and receives an inverse quantization and inverse transformation by receiving a transform coefficient. an inverse quantization and inverse transform unit 130 that outputs a residual signal through a transform, a motion compensator 140 that receives a motion vector, and outputs an image predicted between screens, an intra picture prediction unit 120, or A reconstruction unit 150 for generating a reconstructed image by adding the residual image and the image predicted by the motion compensator 140, a deblocking filter 160 for removing block distortion of the reconstructed image, - comprises a frame memory 180 for storing a scalable decoding process and the decoding control unit 170 to control other functional blocks, and the reference and non-reference frames.

The configuration and connection of the functional blocks as shown in FIG. 1 are not the only way to implement the non-scalable video decoder 100. In some cases, some functions may be implemented in separate functional blocks.

The non-scalable video decoder 100 of FIG. 1 may operate in a pipelined manner. In general, each functional block ends processing of one macroblock and then starts processing of the next macroblock.

FIG. 2 is a diagram illustrating one side of a pipeline operation of the non-scalable video decoder of FIG. 1.

In FIG. 2, ED, ITIQ, IP, MC, REC, and DF are respectively entropy decoder 110, inverse quantization and inverse transform unit 130, intra picture prediction unit 120, and motion compensation unit 140. , An operation of one macroblock of each of the reconstruction unit 150 and the deblocking filter 160 is shown. Vertical lines are shown to distinguish pipeline stages, and functional blocks in the same pipeline stage are executed simultaneously.

2, the pipeline operation of the non-scalable video decoder will be described.

First, the entropy decoder 110 performs entropy decoding on one macroblock (S200), and the inverse quantization and inverse transform unit 130 performs an inverse quantization and inverse transformation process (S210), according to the macroblock type. Intra picture prediction is performed through the picture predictor 120 (S220), or motion compensation is performed through the motion compensator 140 (S220).

Then, the reconstructor 150 generates a reconstructed image by using the operation result of S220 or S220, and the deblocking filter 160 performs a deblocking filter on the final image and finally stores the reconstructed image in the frame memory 180.

In Figure 2, horizontally continuous pipeline stages perform the same operation for consecutive macroblocks. For example, assuming that entropy decoding is performed on the N-th macroblock in S200, entropy decoding may be performed on the N + 1 th macroblock in S201 and the N + 2 th macroblock in S202.

3 is a diagram illustrating a configuration of a scalable video decoder according to an embodiment of the present invention.

Referring to FIG. 3, the scalable video decoder 300 of the present invention includes a bit stream supply unit 310, a plurality of layer processors 321 ˜ 32n, a layer selector 330, a motion compensator 340, and a residual signal. The up sampling unit 350, the reconstruction unit 360, the deblocking unit 370, the texture up sampling unit 380, the decoding control unit 390, and the frame memory 400 are included.

Each of the plurality of hierarchical processing units 321 to 32n again includes an entropy decoder 411 to 41n, an intra picture prediction unit 421 to 42n, and an inverse quantization and inverse transform unit 431 to 43n. 370 includes a first order deblocking filter 371 and a second order deblocking filter 372, and the layer selector 330 includes an intra picture prediction result selector 331, a first inverse quantization, and an inverse transform result. And a selector 332, a second inverse quantization and inverse transform result selector 333, and an entropy decoding result selector 334. For reference, the number of the plurality of layer processors 321 ˜ 32n may be determined according to the number of layers that the scalable video decoder of the present invention intends to support.

That is, in the present invention, by extending some functional blocks constituting a general non-scalable video decoder and adding functional blocks necessary to provide scalability, each layer simultaneously processes some of the decoding processes for real-time decoding. It can be seen that.

Hereinafter, the function of each component will be described.

The bit stream supply unit 310 outputs a bit stream compressed into a base layer and at least one enhancement layer according to a scalable coding scheme, and divides the bit stream into a stream for each layer.

Each layer processor 321 to 32n includes an entropy decoder 411 to 41n, an intra picture predictor 421 to 42n, and an inverse quantization and inverse transform unit 431 to 43n to decode entropy for each stream for each layer. Intra picture prediction, and inverse quantization and inverse transformation are performed in parallel. That is, each of the entropy decoders 411 to 41n is provided with a stream for each layer and parses it, and thus, syntax syntaxes such as an intra prediction mode, a transform coefficient, and a motion vector required for decoding are provided. Print the values hierarchically. Each of the intra picture prediction units 421 to 42n receives an intra prediction mode from the corresponding entropy decoders 411 to 41n, performs intra picture prediction on the intra prediction mode, and acquires and outputs an intra prediction image for each layer. Each inverse quantization and inverse transform unit 431 to 43n receives a transform coefficient from the entropy decoders 411 to 41n, performs inverse quantization and inverse transformation on the same, and acquires and outputs a residual signal for each layer.

The intra picture prediction result selector 331 receives the outputs of all the intra picture predictors 421 to 42n, and under the control of the decoding controller 390, a layer to which the scalable video decoder is currently reconstructed (hereinafter, referred to as a current reconstruction layer). Selects only the intra-picture prediction result corresponding to the lower layer (for example, the N-th layer) of the T1 and provides it to the reconstruction unit 360.

The first inverse quantization and inverse transform result selector 332 receives the outputs of both the inverse quantization and inverse transform units 431 to 43n and corresponds to a lower layer (N-1 th layer) under the control of the decoding controller 390. Only inverse quantization and inverse transformation results are selected and provided to the reconstruction unit 360.

The second inverse quantization and inverse transform result selector 333 receives the outputs of both the inverse quantization and inverse transform units 431 to 43n and inversely corresponds to the current reconstruction layer (Nth layer) under the control of the decoding controller 390. Only the quantization and inverse transform results are selected and provided to the reconstruction unit 360.

The motion compensator 340 generates the inter prediction image by performing motion compensation on the motion vector output from the entropy decoding result selector 334 so as to support the inter-layer motion prediction operation. That is, the motion compensator 340 receives and scales a motion vector of a lower layer (N-1 th layer) to generate and output an inter prediction image when predicting inter-layer motion. However, when the inter-layer motion prediction operation is not needed (that is, when only the motion compensation operation of a specific layer is required), the motion compensation operation may be performed by receiving the motion vector of the corresponding layer.

The residual signal upsampling unit 350 performs residual signal upsampling on the output of the first inverse quantization and inverse transform result selector 332 to support the inter-layer residual signal prediction operation. That is, the residual signal upsampling is performed using the inverse quantization and inverse transform results corresponding to the lower layer (N-1 th layer).

The reconstructor 360 may include an intra-picture prediction result selector 331, first and second inverse quantization and inverse transform result selectors 332 and 333, and primary deblocking according to the type and layer of the macroblock input thereto. The output of the filter 371, the texture up sampling unit 380, and the motion compensator 340 is selectively used to simultaneously process the reconstruction of one or more layers.

More specifically, during inter-layer texture prediction, the reconstruction unit 360 reconstructs the outputs of the intra picture prediction result selection unit 331 and the first inverse quantization and inverse transformation result selection unit 332 to reconstruct the image of the lower layer. Is generated first, and then a texture up-sampling result corresponding thereto is obtained through the first deblocking filter 371 and the texture up-sampling unit 380. Then, the output of the second inverse quantization and inverse transform result selector 333 is additionally input, and the texture up-sampling result is added to finally generate a reconstructed image of the current reconstruction layer, and the second deblocking filter 372 is added. After that, it is stored in the frame memory 400.

In the inter-layer residual signal prediction, the reconstruction unit 360 obtains the residual signal up-sampling result of the lower layer by using the output of the residual signal up-sampling unit 350, and selects the second inverse quantization and inverse transform result selection unit 333. ) To obtain the residual signal of the current reconstruction layer. After adding the residual signal of the current reconstruction layer to the residual signal upsampling result for the lower layer, the output image of the motion compensation unit 340 is added again to generate a reconstructed image of the current reconstruction layer, and the second deblocking filter ( 372 is stored in the frame memory 400.

In inter-layer motion prediction, the reconstructor 360 reconstructs the current image obtained by the second inverse quantization and inverse transform result selector 333 in the prediction image obtained by the motion compensator 340 scaling the motion vector of the lower layer. The residual signal of the layer is added to generate a reconstructed image of the current reconstruction layer.

The deblocking unit 370 includes a first deblocking filter 371 and a second deblocking filter 372, and the first deblocking filter 371 is activated in the inter-layer texture prediction mode to restore the lower layer. The image is deblocked and transferred to the texture up sampling unit 380. The second deblocking filter 372 receives the reconstructed image of the current reconstruction layer transmitted from the reconstruction unit 360, deblocks the received reconstruction image, and stores the deblocking image in the frame memory 400.

The texture up sampling unit 380 is activated in the inter-layer texture prediction mode, and when the deblocked reconstructed image is transmitted from the first deblocking filter 371, the texture up sampling unit 380 provides the texture up sampling to the reconstructing unit 360.

Further, in the above embodiment, it is most preferable that the streams of the base layer and one or more enhancement layers are processed in parallel through a plurality of layer processing units. However, when parallel processing using the plurality of layer processing units is impossible due to various external factors, The stream of the base layer and one or more enhancement layers may be serially processed using only one layer processor. That is, one layer processor may sequentially process the streams of the base layer and the one or more enhancement layers in the order of the layers.

FIG. 4 is a diagram for describing an inter-layer texture prediction method (that is, an I_BL type macroblock prediction method) using the scalable video decoder of FIG. 3.

In FIG. 4, for convenience of description, two layer processing units 321 and 322 are provided, and only the case of obtaining a reconstructed image of an enhancement layer by receiving a bit stream of two layers including a base layer and an enhancement layer is input. do.

As shown in FIG. 4, ED1, ITIQ1, and IP1 of the base layer include a first entropy decoder 411, a first inverse quantization and inverse transform unit 431, and a first intra provided in the first layer processing unit 321 of FIG. 3. A second entropy decoder 412, a second inverse quantization unit and an inverse transform unit 432 are provided in the second layer processor 322 for ED2 and ITIQ2 of the enhancement layer. Means the operation for one macroblock. Similarly, DF1, DF2, TUP, and REC denote operations of the primary deblocking filter 371, the secondary deblocking filter 372, the texture upsampling unit 380, and the reconstruction unit 360, respectively.

Subsequently, an inter-layer texture prediction method will be described with reference to FIG. 4.

First, when a bit stream composed of a base layer and an enhancement layer is input, the first entropy decoder 411 performs entropy decoding of the base layer (S400), the first inverse quantization and inverse transform unit 431, and the first intra picture. The prediction unit 421 performs inverse quantization / inverse transformation and intra picture prediction (S410 and S420).

Then, the reconstructor 360 generates a reconstructed image of the base layer by using the outputs of the first inverse quantization and inverse transform unit 431 and the first intra-picture predictor 421 (S430), and the first deblocking filter ( Deblocking the reconstructed image of the base layer through 371) and then up-sampling the texture through the texture upsampling unit 380 (S440 and S450), and inputs the result back into the reconstruction unit 360.

Meanwhile, the second entropy decoder 412 receives the bit stream of the enhancement layer, performs entropy decoding on the enhancement layer (S470), and performs inverse quantization / inverse transformation through the second inverse quantization and inverse transform unit 432. In operation S480, the result is transmitted to the reconstruction unit 360.

The reconstructor 360 receives the texture upsampled texture from the texture upsampling unit 380 and the residual signal of the enhancement layer from the second inverse quantization and inverse transform unit 432. A restored image is generated (S460).

Then, the second deblocking filter 372 deblocks the reconstructed image of the enhancement layer generated by the reconstruction unit 360 and stores the deblocked image in the frame memory 400 (S490).

In Figure 4, horizontally continuous pipeline stages perform the same operation for consecutive macroblocks. For example, if entropy decoding is performed on the N-th macroblock of the base layer in step S400, entropy decoding is performed on the N + 1 th macroblock in S401 and the N + 2 th macroblock in S402.

FIG. 5 is a diagram for describing a method for predicting an inter-layer residual signal using the scalable video decoder of FIG. 3.

In FIG. 5, two layer processing units 321 and 322 are provided for convenience of description similarly to FIG. 4, and when a bit stream of two layers including a base layer and an enhancement layer is received and a restored image of the enhancement layer is obtained. It will be described only. In this case, it is assumed that the inter-layer motion prediction operation is deactivated.

As shown in FIG. 5, ED1 and ITIQ1 of the base layer perform operations on one macroblock of the first entropy decoder 411 and the first inverse quantization and inverse transform unit 431 included in the first layer processing unit 321. ED2 and ITIQ2 of the enhancement layer mean an operation of one macroblock of the second entropy decoder 412, the second inverse quantization and inverse transform unit 432 included in the second layer processing unit 322. Similarly, RUP, MC, REC, and DF2 mean operations of the residual signal upsampling unit 350, the motion compensator 340, the reconstructor 360, and the second order deblocking filter 372, respectively.

Subsequently, an inter-layer residual signal prediction method will be described with reference to FIG. 5.

When a bit stream composed of a base layer and an enhancement layer is input, the first entropy decoder 411 performs entropy decoding of the base layer (S500), and the remaining of the base layer is performed through the first inverse quantization and inverse transform unit 431. After acquiring the signal (S510), the residual signal upsampling unit 350 upsamples the residual signal of the base layer obtained through the step S510 (S520).

Meanwhile, the second entropy decoder 412 receives the enhancement layer stream, performs entropy decoding on the enhancement layer (S530), and obtains a residual signal of the enhancement layer through the second inverse quantization and inverse transform unit 432. (S540). The motion compensation unit 340 in which the inter-layer motion prediction operation is deactivated performs motion compensation using the motion vector of the enhancement layer (S550).

Then, the reconstructor 360 adds the residual signal of the upsampled base layer input from the residual signal upsampling unit 350 and the residual signal of the enhancement layer input from the second inverse quantization and inverse transform unit 432. The reconstructed image of the enhancement layer is generated in addition to the inter prediction image input from the motion compensator 340 (S560).

The reconstructed image of the enhancement layer generated through the step S560 is deblocked through the second deblocking filter 372 and then stored in the frame memory 400 (S570).

Also in FIG. 5, horizontally continuous pipeline stages perform the same operation on consecutive macroblocks as in FIG. For example, if entropy decoding is performed on the N-th macroblock of the base layer in step S500, entropy decoding is performed on the N + 1 th macroblock in step S501 and the N + 2 th macroblock in step S502.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and it is common in the art that various substitutions, modifications, and changes can be made without departing from the technical spirit of the present invention. It will be apparent to those skilled in the art.

1 is a diagram illustrating a configuration of a general non-scalable video decoder.

FIG. 2 is a diagram illustrating one side of a pipeline operation of the non-scalable video decoder of FIG. 1.

3 is a diagram illustrating a configuration of a scalable video decoder according to an embodiment of the present invention.

FIG. 4 is a diagram for describing an inter-layer texture prediction method using the scalable video decoder of FIG. 3.

FIG. 5 is a diagram for describing a method for predicting an inter-layer residual signal using the scalable video decoder of FIG. 3.

Claims (10)

A parallel processor for performing entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation on a layer-by-layer basis of the bit stream; A hierarchy selector for sequentially selecting and outputting an output result of the parallel processor for each layer; A motion compensator for performing motion compensation using an entropy decoding result output from the layer selector; A residual signal upsampling unit configured to perform residual signal upsampling using the inverse quantization and inverse transformation results output from the layer selector; A deblocking unit to deblock the reconstructed image; A texture up sampling unit configured to texture up sample the deblocked reconstructed image output from the deblocking unit; And And a reconstruction unit configured to simultaneously perform a process of reconstructing a reconstructed image on at least one layer by selectively using outputs of the layer selector, the motion compensator, the residual signal upsampler, and the texture upsampler. Video decoding device. The method of claim 1, wherein the parallel processing unit A bit stream supply unit dividing the bit stream into a stream for each layer; And And a plurality of layer processors configured to perform entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation on each of the streams of each layer in parallel. The method of claim 1, wherein the layer selector An intra picture prediction result selector configured to select and output an intra picture prediction result of a lower layer from the intra picture prediction results output in parallel from the parallel processor; A first inverse quantization and inverse transform result selector configured to select and output inverse quantization and inverse transform results of the lower layer among inverse quantization and inverse transform results output from the parallel processor; A second inverse quantization and inverse transform selection unit for selecting and outputting inverse quantization and inverse transformation results of the current restoration layer among inverse quantization and inverse transformation results output from the parallel processor; And And an entropy decoding result selector which selects a motion vector of the current reconstruction layer or the lower layer among the entropy decoding results output in parallel from the parallel processor and outputs the motion vector to the motion compensation unit. The method of claim 3, wherein the deblocking unit A first deblocking filter which deblocks the reconstructed image of the lower layer output from the reconstruction unit and provides the block to the texture up sampling unit; And And a second deblocking filter for deblocking a reconstructed image of the current reconstruction layer output from the reconstruction unit and providing the deblocking image to a frame memory. The method of claim 4, wherein the reconstruction unit In the inter-layer texture prediction, a reconstructed image of the lower layer is generated by reconstructing outputs of the first inverse quantization and inverse transform result selection unit and the intra-picture prediction result selecting unit, and the reconstructed image of the lower layer is stored in the first order. And deblocking and texture upsampling through a blocking filter and the texture upsampling unit, and then adds a result of inverse quantization and inverse transformation of the current reconstruction layer to generate a reconstructed image of the current reconstruction layer. The method of claim 4, wherein the reconstruction unit In the prediction of the residual signal between layers, a reconstruction image of the current reconstruction layer is generated by reconstructing the output of the residual signal up-sampler, the output of the second inverse quantization and inverse transform result selector, and the output of the motion compensation unit. Scalable video decoding apparatus. The method of claim 4, wherein the reconstruction unit In the inter-layer motion prediction, the motion compensation unit generates a reconstructed image of the current reconstructed layer by adding the output of the second inverse quantization and inverse transform result selector to the predicted image obtained by scaling the motion vector of the lower layer. Scalable video decoding apparatus. A scalable video decoding method for performing inter-layer texture prediction, Parallel processing entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation for each layer of the bit stream; Generating a reconstructed image of the lower layer by selecting and reconstructing a result of inverse quantization and inverse transformation of the lower layer with respect to a current reconstruction layer and an intra picture prediction result; Obtaining a texture upsampling result by performing deblocking and texture upsampling on the reconstructed image of the lower layer; And And reconstructing a reconstructed image of the current reconstructed layer by using the texture up-sampling result and the inverse quantization and inverse transform result of the current reconstructed layer. A scalable video decoding method for performing inter-layer residual signal prediction, Parallel processing entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation for each layer of the bit stream; Selecting a result of inverse quantization and inverse transformation of a lower layer with respect to a current reconstruction layer to perform residual signal upsampling; Selecting a result of inverse quantization and inverse transformation of a current restoration layer; And And reconstructing a reconstructed image of the current reconstructed layer by using the result of performing the residual signal upsampling and the inverse quantization and inverse transform result of the current reconstructed layer. A scalable video decoding method for performing inter-layer motion prediction, Parallel processing entropy decoding, intra-picture prediction, and inverse quantization and inverse transformation for each layer of the bit stream; Selecting a motion vector included in the entropy decoding result of the lower layer with respect to the current reconstruction layer and performing motion compensation to obtain a predicted image of the current reconstruction layer; Selecting a result of inverse quantization and inverse transformation of a current restoration layer; And And reconstructing a reconstructed image of the current reconstructed layer by using the predicted image of the current reconstructed layer and inverse quantization and inverse transform results of the current reconstructed layer.

Images