JP5063678B2 - Method of assigning priority for adjusting bit rate of bit stream, method of adjusting bit rate of bit stream, video decoding method, and apparatus using the method - Google Patents

Description

  The present invention relates to a video coding technique, and more particularly, to a method for adjusting a bit rate of a bitstream composed of a plurality of quality layers.

  As information communication technology including the Internet develops, not only text and voice but also image communication is increasing. Existing character-centric communication methods cannot satisfy the diverse needs of consumers, and as a result, multimedia services that can accommodate various forms of information such as characters, video, and music are increasing. Since the amount of multimedia data is enormous, it requires a large-capacity storage medium and requires a wide bandwidth during transmission. Therefore, it is essential to use a compression coding technique to transmit multimedia data including characters, video, and audio.

  The basic principle of compressing data is a process of removing data redundancy elements. Spatial overlap where the same color or object is repeated in the image, temporal overlap where adjacent frames are almost unchanged in the video frame or the same sound repeats in the audio, or human visual and perceptual ability The data can be compressed by removing perceptual duplication that takes into account the insensitivity to high frequencies. In a general video coding method, temporal overlap is removed by temporal filtering based on motion compensation, and spatial overlap is removed by spatial transform.

  In order to transmit multimedia generated after removing data duplication, a transmission medium is required, but the performance varies depending on the transmission medium. Currently used transmission media have various transmission rates such as an ultra-high-speed communication network capable of transmitting several tens of megabits of data per second to a mobile communication network having a transmission rate of 384 kbits per second. In such an environment, it is possible to transmit multimedia at a transmission rate suitable for supporting transmission media of various speeds or according to the transmission environment, that is, a scalable video coding method is multi-purpose. It can be said that it is more suitable for the media environment.

  In scalable video coding, a part of the bit stream is cut out according to peripheral conditions such as a transmission bit rate, a transmission error rate, and system support with respect to an already compressed bit stream (bit-stream), and a video resolution and a frame rate are obtained. , And SNR (Signal-to-Noise Ratio) and the like, that is, a coding scheme that supports various scalability.

  Currently, JVT (Joint Video Team), a joint working group of MPEG (Moving Picture Experts Group) and ITU (International Telecommunication Union), is H.264. Based on H.264, standardization work relating to scalable video coding (hereinafter referred to as SVC) is in progress. In the SVC standard, FGS (Fine Granularity Scalability) technology is adopted to support SNR scalability.

  FIG. 1 shows an example of a scalable video coding scheme using a multi-layer structure. In the above method, the first layer is set to QCIF (Quarter Common Intermediate Format) and 15 Hz (frame rate), the second layer is set to CIF (Common Intermediate Format) and 30 Hz, and the third layer is set to SD (Standard Definition). ), Set to 60 Hz.

  In order to encode a multi-layer video frame having such various resolutions and / or frame rates, the relationship between layers can be used. For example, a certain region 12 of the video frames of the first enhancement layer is Encoding is efficiently performed by prediction from the corresponding region 13 among the video frames in the base layer. Similarly, region 11 of the second enhancement layer video frame is efficiently encoded by prediction from region 12 of the first enhancement layer.

  FIG. 2 illustrates the concept of scalable video coding inter prediction and intra-based prediction techniques. The block 24 belonging to the frame 21 having the current hierarchy can be predicted with reference to the block 25 belonging to another frame 22 existing in the same hierarchy. This is called inter prediction. The inter prediction includes a motion estimation process for searching for a motion vector for indicating a corresponding block.

  On the other hand, the block 24 may be predicted with reference to the block 26 belonging to the frame 23 of the lower layer (base layer) existing in the same temporal position or picture order count (POC) as the frame 21. it can. This is called intra-base prediction. In intra-based prediction, the motion estimation process is not necessary.

  FIG. 3 is a diagram illustrating an example in which the FGS technique is applied to a residual picture according to the prediction of FIG. The residual picture 30 can be expressed by a plurality of quality layers in order to support SNR scalability. Such a quality hierarchy is necessary for variously expressing video quality, and is distinguished from the resolution and / or frame rate hierarchy of FIG.

  The plurality of quality layers may include one discrete layer (discrete layer: 31) and at least one or more FGS layers 32, 33, and 34. The video quality measured by the video decoder is that when only the discrete layer 31 is received, when the discrete layer 31 and the first FGS layer 32 are received, the discrete layer 31, the first FGS layer 32, and the second FGS layer 33 are When received, and increases in order when all the layers 31, 32, 33, 34 are received.

  FIG. 4 is a diagram illustrating a process of expressing one picture or slice by one discrete layer and two FGS layers.

First, the original picture (or slice) 41 is quantized by the first quantization parameter (QP ₁ ) (S1). The quantized picture 42 forms a discrete hierarchy. The quantized picture 42 is dequantized (S2), and the dequantized picture 43 is provided to a subtractor (44). The subtracter 44 subtracts the provided picture 43 from the original picture (S3). The subtracted result is again quantized by the second quantization parameter (QP ₂ ) (S4). The quantized result 45 forms a first FGS hierarchy.

The quantized result 45 is dequantized (S5), and the dequantized result 46 is provided to the adder 47. The picture 46 and the inversely quantized picture 43 provided in the previous period are added by an adder (adder: 47) (S6) and then provided to a subtractor 48. The subtracter 48 subtracts the added result from the original picture 41 (S7). The subtracted result is also quantized by a third quantization parameter (QP ₃ ) (S8). The quantized result 49 forms a second FGS hierarchy.

  A plurality of quality hierarchies as shown in FIG. 3 can be formed by such a process.

  5 and 6 are diagrams illustrating a method of truncating the quality hierarchy used in the current SVC standard. As shown in FIG. 5, the current picture 30 is predicted from another reference picture 35 by inter prediction or intra-base prediction and displayed as a residual picture. However, the current picture 30 displayed as a residual picture is not only composed of a plurality of quality layers 31, 32, 33, 34, but the reference picture 35 is similarly composed of a plurality of quality layers 36, 37, 38, 39. It can consist of

  In this case, according to the current SVC standard, the bitstream extractor cuts out a part of the quality layer as shown in FIG. 6 in order to adjust the SNR of the bitstream after the video encoding. That is, the bitstream extractor sequentially truncates from the highest quality layer 34 of the current picture 30 existing in a high resolution and / or frame rate layer (hereinafter simply referred to as “layer” to be distinguished from “quality layer”). Then, after the quality hierarchy of the higher hierarchy picture 30 is completely cut out, the quality hierarchy of the lower hierarchy reference picture 35 is cut out from the top.

  Such a clipping method is optimal for restoration of a picture (reference picture) in a lower hierarchy (example: QCIF), but in order to restore a picture (current picture) in a higher hierarchy (example: CIF). It's not optimal. The quality hierarchy of some lower hierarchy pictures may be less important than the quality hierarchy of higher hierarchy pictures due to the higher hierarchy picture quality. Therefore, it is necessary to realize efficient SNR scalability by cutting out the quality layer by another method depending on whether the main purpose is a high layer picture or a low layer picture on the video encoder side.

  The problem to be solved by the present invention is to provide an SNR adjustment method and apparatus for a bit stream that focuses on a high hierarchy.

  Another technical problem to be solved by the present invention is to provide a method and apparatus for adaptively adjusting the SNR depending on whether a high layer picture is a main point or a low layer picture is a main point.

  The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

A method of assigning priority for adjusting a bit rate of a bitstream according to an embodiment of the present invention includes a quality hierarchy in which a reference picture encoding unit represents a plurality of quantized pictures each having a different quality with respect to a reference picture. A first quality layer including a step of repeatedly performing inverse quantization and quantization, and a current picture encoding unit having a plurality of different qualities with respect to a current picture encoded with reference to the reference picture Forming a second quality layer including a quality layer representing a quantized picture by repeating inverse quantization and quantization, and a quality level allocator comprising the first quality layer and the second quality layer It comprises the steps of assigning a priority to each quality layer including a respective, small influence on the picture quality reduction of the current picture be removed Low priority is assigned to quality hierarchy.

A method of adjusting a bit rate of a bitstream according to an embodiment of the present invention includes a step of receiving a video bitstream input by a bitstream input unit, and a target bitrate setting unit setting a target bitrate for the video bitstream And a bitstream parser includes a first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture of the current picture from the video bitstream, and the current picture On the other hand, obtaining a second quality layer including a quality layer representing a plurality of quantized pictures, each of which has different qualities, and a bitstream cutting unit according to the target bit rate, the first quality layer and the second quality among the quality layers comprising each hierarchy, the current peak when it is removed Comprising the steps of priority indicating the smallness of the effects on tea quality reduction to remove the low quality layer, a.

In the video coding method according to the embodiment of the present invention, the bitstream input unit receives the video bitstream and the bitstream parser has different qualities from the video bitstream to the reference picture of the current picture. Reference relationship between a first quality layer including a quality layer representing a plurality of quantized pictures and a second quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to the current picture and the current picture A dependency ID indicating the current picture of the current picture when the dependency ID setting unit is removed from each of the quality layers including each of the first quality layer and the second quality layer according to a target bit rate. by priority indicating a small influence on the picture quality decreases lower quality layer is removed, the first If the dependency ID among the quality layers there is no quality layer to instruct, said step depends ID is set to indicate the highest quality layer of the remaining quality layers of the first quality layers, and the current picture decoded Restoring a current picture according to a reference relationship indicated by the dependency ID.

A priority disgusting assignment apparatus for adjusting a bit rate of a bitstream according to an embodiment of the present invention includes a first quality layer including a quality layer representing a plurality of quantized pictures each having a different quality with respect to a reference picture. A reference picture encoding unit configured by repeatedly performing inverse quantization and quantization, and a quality hierarchy representing a plurality of quantized pictures having different qualities with respect to a current picture encoded with reference to the reference picture. A current picture encoding unit configured by repeatedly performing inverse quantization and quantization on a second quality layer including the same, and a quality level assignment for assigning a priority to each quality layer including each of the first quality layer and the second quality layer The lower priority is assigned to a quality hierarchy that has a small effect on the reduction in image quality of the current picture even if it is removed. The temple.

An apparatus for adjusting a bit rate of a bit stream according to an embodiment of the present invention includes a bit stream input unit that receives an input of a video bit stream, a target bit rate setting unit that sets a target bit rate for the video bit stream, and the video A first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture of the current picture from the bitstream, and a plurality of quantized units having different qualities with respect to the current picture. A bitstream parser that obtains a second quality layer including a quality layer representing a picture, and a case in which each of the quality layers including each of the first quality layer and the second quality layer is removed according to the target bit rate Has a priority indicating a small impact on the picture quality reduction of the current picture Including a bit stream cutting unit for removing the quality layer have a.

A video decoder according to an embodiment of the present invention represents an entropy decoding unit that receives an input of a video bitstream and a plurality of quantized pictures having different qualities from a reference picture of a current picture from the video bitstream. Reference relationship of pictures in a first quality layer including a quality layer, a second quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to the current picture, and the second quality layer A bit stream parser that obtains a dependency ID, and a reduction in image quality of the current picture when each of the quality layers including each of the first quality layer and the second quality layer is removed according to a target bit rate. that priority indicating a small influence on the low quality layer is removed by, before one of the first quality layer When there is no quality hierarchy indicated by the dependency ID, the dependency ID indicates that the dependency ID is set to indicate the highest quality hierarchy among the remaining quality hierarchies of the first quality hierarchy. A current picture decoding unit that restores the current picture according to a reference relationship.

  According to the present invention, there is an effect that the bit rate can be adjusted adaptively with the main purpose being the image quality of a high-level picture in the bitstream.

  Specific matters of other embodiments are included in the detailed description and the drawings.

  Advantages and features of the present invention, and methods for achieving them will be clarified by referring to embodiments described later in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, and may be embodied in various forms different from each other. This embodiment is provided merely for the complete disclosure of the present invention and for the purpose of fully informing the person skilled in the art to which the present invention pertains the scope of the invention. Is defined only by the scope of the claims. Throughout the specification, the same reference numerals refer to the same components.

  FIG. 7 is a diagram showing a configuration of a conventional SVC system, and FIG. 8 is a diagram showing a configuration of an SVC system according to an embodiment of the present invention.

  Referring to FIG. 7, the video encoder 61 generates a multi-layer based scalable bit stream, for example, a CIF bit stream. Then, the bitstream extractor 62 can transmit the generated CIF bitstream to the video decoder 1 (63) as it is, and cuts a part of the higher layer to extract a QCIF bitstream having a lower resolution, This can be transmitted to the video decoder 2 (64). At this time, in these two cases, a part of the quality layers is cut and the resolution is the same, but only the SNR can be changed.

  On the other hand, referring to FIG. 8, a quality level (priority) is assigned to the CIF bit stream generated by the video encoder 50 by the quality level assigner 140. That is, a priority ID is assigned to each NAL (Network Abstraction Layer) unit constituting the CIF bit stream. At this time, in the priority ID assignment, the priority ID assignment method considering a plurality of hierarchies proposed in the present invention is followed.

  When transmitting the bit stream to the video decoder 2 (300b), the bit stream extractor 200 cuts out the upper layer and transmits the bit stream optimized for QCIF as the lower layer. At this time, if the SNR adjustment is necessary and the quality hierarchy is cut, the method is the same as the conventional method.

  However, when transmitting the bit stream to the video decoder 1 (300a), the bit stream extractor 200 transmits a CIF bit stream including all the layers as they are. At this time, if the SNR adjustment is required and the quality hierarchy is cut, the quality hierarchy having a lower priority ID is cut according to the priority ID assigned by the quality level assigner 140.

  FIG. 9 is a diagram illustrating an example of cutting a quality hierarchy according to an embodiment of the present invention. According to this, the quality level assigner 140 assigns priority IDs according to the following order, and the bitstream extractor 200 realizes SNR scalability by cutting out from the quality hierarchy with the lower priority ID. .

  The quality level assigner 140 first grasps the reference relationship in the input bitstream. Such a reference relationship is used for prediction, and examples of prediction techniques include inter prediction and intra-base prediction. At this time, a picture to be referred to in the prediction technique is called a current picture 30, and a picture to be referred to is called a reference picture 35. 9 illustrates an example in which the number of quality layers of the current picture 30 and the number of quality layers of the reference picture 35 are the same. However, the number of quality layers between the current picture 30 and the reference picture 35 may be different from each other. Of course.

  Specifically, the process of assigning priority IDs is as follows. The first candidate from which the highest quality layer 34 of the current picture 30 is removed is compared with the second candidate from which the highest quality layer 39 of the reference picture 35 is removed, and the one that is advantageous in terms of image quality is selected. The first candidate is a case where the image of the layer to which the current picture belongs is restored from the three quality layers 31, 32, 33 of the current picture 30 and the four quality layers 36, 37, 38, 39 of the reference picture 35. The second candidate means a case where an image of a layer to which the current picture belongs is restored from the four quality layers 31, 32, 33, and 34 of the current picture 30 and the three quality layers 36, 37, and 38 of the reference picture 35.

  The specific process of restoring the picture is to restore the reference picture 35 from the quality hierarchy that forms the reference picture 35 first, and then restore the residual signal of the current picture 30 from the quality hierarchy that forms the current picture 30, This is performed in the process of adding the restored reference picture 35 and the restored residual signal.

  After obtaining the first candidate and the second candidate in this way, the costs of the two candidates are compared. In the method for determining the cost, a rate-distortion function is usually used. The following equation (1) shows a process for obtaining a cost by a rate distortion function.

    C = E + λ × B (1)

  Here, C is the cost, E is the degree of distortion from the original signal (for example, it can be calculated by MSE (Mean Square Error)), B is the bit amount required when the corresponding data is compressed, and λ is the Lagrange. A multiplier (Lagrangian multiplier) is shown. The Lagrange multiplier is a coefficient that can adjust the reflection ratio of E and B. Therefore, the cost C decreases as the difference E from the original signal E and the required bit amount B become smaller. Therefore, the lower cost C indicates that more efficient encoding has been performed.

  Thus, if the case where the cost C is low is selected from the first candidate and the second candidate, the priority ID is assigned accordingly. For example, if the first candidate is selected, it means that even if the quality layer 34 of the current picture 30 is removed, it has the smallest effect on the overall video quality, so the quality layer 34 has the lowest priority ID. Some 0 is set.

  Next, priority IDs must be set for the remaining quality layers 31, 32, 33 of the current picture 30 and the remaining quality layers 36, 37, 38, 39 of the reference picture 35. However, the next process is the same as the process of comparing the first candidate and the second candidate. That is, of the remaining quality layers of the current picture 30, the first candidate from which the highest quality layer 33 is removed is compared with the second candidate from which the highest quality layer 39 of the reference picture 35 is removed, and the one with low cost C is selected. To do.

  As described above, the candidate obtained by removing the highest quality layer from the remaining quality layers to which the priority ID is not assigned in the current picture 30 and the highest quality layer from among the remaining quality layers to which the priority ID is not assigned by the reference picture 35. By repeating the process of selecting one of the candidates that have been removed, priority IDs can be assigned to all quality layers of the current picture 30 and the reference picture 35.

  The quality level assigner 140 records the priority ID in the header (NAL header) of the NAL unit corresponding to each quality layer.

  FIG. 10 is a diagram illustrating a bitstream 80 to which a priority ID is assigned according to an embodiment of the present invention. The quality hierarchy related to the current picture is recorded in the plurality of NAL units 81, 82, 83, and 84, and the quality hierarchy related to the reference picture is recorded in the plurality of NAL units 86, 87, 88, and 89. One NAL unit is composed of a NAL header and a NAL data field. Among these, the NAL header is a part for displaying additional information for the NAL data and includes a priority ID. In the NAL data field, encoded data corresponding to each quality layer is recorded.

  In FIG. 10, the priority ID set by the quality level assigner 140 is displayed in the NAL header. The bitstream extractor adjusts the SNR of the bitstream with reference to the priority ID. The bitstream extractor 200 performs image quality reduction by removing NAL units by cutting NAL units in the order of low priority ID to high priority ID (81, 82, 86, 83, 87, 84, 88, 89 order). Minimize the decrease.

  Of course, since this is optimized for the video quality of the upper layer (when transmitted to the video decoder 1 (300a) in FIG. 8), it is optimized for the video quality of the lower layer (video decoder 2 (FIG. 8)). In the case of transmission to 300b), it is possible to use a technique of sequentially cutting from the highest quality layer of the upper layer, as in the past, regardless of the priority ID.

  However, as proposed in the present invention, the quality layer of the base layer (the layer to which the reference picture belongs) may be cut out earlier than the quality layer of the current layer (the layer to which the current picture belongs). In such a case, the quality hierarchy of the basic hierarchy indicated by the dependency ID of any quality hierarchy in the current hierarchy may not exist. The dependency ID indicates a dependency relationship between data that must be decoded and referred to in order to decode certain data. Accordingly, in the video decoding process, if there is no base layer quality layer referenced by the dependency ID, a method may be used in which the dependency ID is regarded as referring to the highest quality layer among the remaining quality layers. .

  Referring to FIG. 11, the current picture 30 is obtained by cutting out the highest quality layer 34 in the quality layer and the highest quality layer 39 in the quality layer of the reference picture 35 by the bitstream extractor 200. However, according to the present invention, since the quality hierarchy of the lower hierarchy may be cut first before the quality hierarchy of the upper hierarchy is cut out, the quality hierarchy 39 in which the dependency ID of the quality hierarchy 33 of the current hierarchy is gone is indicated. Sometimes. In this case, the dependency ID of the quality layer 33 must be corrected at the video decoder end so as to indicate the highest quality layer 38 among the remaining quality layers 36, 37, and 38.

  12 to 14 are block diagrams of the apparatus configuration according to an embodiment of the present invention. Of these, FIG. 12 is a block diagram showing the configuration of the priority assignment apparatus 100 according to an embodiment of the present invention. The priority assignment device 100 is a device that assigns priority according to quality level in order to adjust the bit rate of the bitstream.

  The priority allocation apparatus 100 may include a current picture encoding unit 110, a reference picture encoding unit 120, a quality level allocator 140, and an entropy encoding unit 150.

  The reference picture encoding unit 120 configures a quality layer (referred to as a first quality layer) related to the reference picture. For this, the reference picture encoding unit 120 may include a prediction unit 121, a conversion unit 122, a quantization unit 123, and a quality layer generation unit 124.

  The prediction unit 121 obtains a residual signal by subtracting an image predicted by a predetermined prediction method in the current macroblock. As the prediction method, as shown in FIG. 2, there are inter prediction and intra-base prediction. Inter prediction includes a motion estimation process that seeks a motion vector to represent relative motion between a current picture and a picture having the same resolution and other temporal position as the current picture.

  On the other hand, the current picture exists at the same temporal position as the current picture, and can be predicted with reference to a picture in a lower layer (base layer) having a resolution different from that of the current picture. This is called intra-base prediction. Of course, the motion estimation process is not necessary for intra-base prediction.

  The transform unit 122 transforms the obtained residual signal using a spatial transform technique such as DCT or wavelet transform to generate transform coefficients. In such a spatial conversion method, DCT (Discrete Cosine Transform), wavelet transform (wavelet transform), or the like is used. As a result of spatial transformation, a transformation coefficient is obtained. When DCT is used as a spatial transformation method, a DCT coefficient is obtained, and when wavelet transformation is used, a wavelet coefficient is obtained.

  The quantization unit 123 quantizes the transform coefficient obtained by the transform unit 122 to generate a quantized coefficient. Quantization means an operation in which the transform coefficient expressed by an arbitrary real value is divided into fixed intervals and indicated by a discrete value (discrete value). Such quantization methods include scalar quantization and vector quantization. Of these, simple scalar quantization methods divide transform coefficients by quantization parameters, and then round off to the nearest whole number. It is a process.

  The quality hierarchy generation unit 124 generates a plurality of quality hierarchies through the process described with reference to FIG. The plurality of quality layers may include one discrete layer and at least one FGS layer.

  Meanwhile, like the reference picture encoding unit 120, the current picture encoding unit 110 includes a prediction unit 111, a conversion unit 112, a quantization unit 113, and a quality layer generation unit 114. The operation of each component is the reference picture encoding 120. Same as in. However, the picture used for prediction of the current picture by the prediction unit 111 uses the reference picture input to the reference picture encoding unit 120. The prediction unit 111 performs inter prediction or intra-base prediction using the input reference picture to generate a residual signal.

  Eventually, the current picture encoding unit 110 configures a quality layer (referred to as a second quality layer) related to the current picture (more precisely, related to the residual signal of the current picture). The input reference picture may have a resolution different from that of the current picture (in the case of intra-based prediction), or may have a different temporal level (in the case of inter prediction).

  The quality level assigner 140 assigns a priority ID to each of the first quality hierarchy and the second quality hierarchy. The priority is assigned in such a manner that a low priority is assigned to a quality layer that has a small influence on the reduction in image quality of the current picture, and a high priority is assigned to a quality layer that has a large influence (see FIG. 9).

  As a criterion for determining the decrease in image quality, a cost function such as Equation (1) can be used. The cost function can be expressed by the difference from the original image and the addition of the bit amount required for encoding.

  The entropy encoding unit 150 generates a bitstream by entropy encoding the priority ID determined by the quality level allocator 140, the first quality layer related to the reference picture, and the second quality layer related to the current picture. Entropy coding is a lossless coding technique using statistical characteristics of data, and includes arithmetic coding, variable length coding, and the like.

  FIG. 13 is a block diagram illustrating a configuration of an apparatus for adjusting a bit rate of a bitstream, that is, a bitstream extractor 200 according to an embodiment of the present invention.

  The bitstream extractor 200 may include a bitstream input unit 210, a bitstream parser 220, a bitstream cutting unit 230, a target bit rate setting unit 240, and a bitstream transmission unit 250.

  The bit stream input unit 210 receives an input of the video bit stream, and the bit stream transmission unit 250 transmits the video bit stream whose bit rate is changed. The bit stream input unit 210 corresponds to a reception unit in the network interface, and the bit stream transmission unit 250 corresponds to a transmission unit in the network interface.

  The target bit rate setting unit 240 sets a target bit rate for the video bit stream. Such a target bit rate can be determined by comprehensively considering the bit rate of the bit stream currently transmitted, the network status, or the performance of the receiving end (video decoder) device.

  The bitstream parser 220 reads priority IDs of a first quality layer related to a reference picture and a second quality layer related to the current picture in the video bitstream. Such priority ID is assigned by the quality level assigner 140.

  The bitstream cutting unit 230 cuts from the quality layer with the lower priority among the first quality layer and the second quality layer according to the target bit rate. The cutting process is repeated until the target bit rate is reached.

  FIG. 14 is a block diagram illustrating a configuration of a video decoder 300 according to an embodiment of the present invention.

  The video decoder 300 includes an entropy decoding unit 310, a bit stream parser 320, a current picture decoding unit 330, a reference picture decoding unit 340, and a dependency ID setting unit 350.

  The entropy decoding unit 310 receives an input video bitstream and performs lossless decoding thereof. The lossless decoding is performed in reverse of the lossless encoding of the entropy encoding unit 150 in FIG.

  The bitstream parser 320 includes encoded data of a reference picture (first quality layer), encoded data of a current picture (second quality layer), a dependency ID of a first quality layer related to a reference picture, and the current The dependency ID of the second quality hierarchy related to the picture is read. The dependency ID indicates information regarding which quality layer of the reference picture is required to restore any quality layer of the current picture, that is, a dependency relationship.

  However, as described with reference to FIG. 11, according to the present invention, the quality hierarchy in the lower hierarchy may be cut first before all the quality hierarchies in the upper hierarchy are cut out. A quality hierarchy with no ID may be indicated.

  In this case, the dependency ID setting unit 350 sets the dependency ID to indicate the highest quality layer among the remaining quality layers.

  The reference picture decoding unit 340 decodes the encoded data of the reference picture. For this, the reference picture decoding unit 340 may include an inverse quantization unit 341, an inverse transform unit 342, and an inverse prediction unit 343.

  The inverse quantization unit 341 inversely quantizes the encoded data of the reference picture.

  The inverse transform unit 342 performs inverse transform on the inverse quantization result. Such reverse conversion is executed in reverse of the conversion process executed by the conversion unit 122 of FIG.

  The inverse prediction unit 343 restores the reference picture by adding the restored residual signal provided from the inverse transform unit 342 to the prediction signal. At this time, the prediction signal is obtained by inter prediction or intra-base prediction in the same manner as at the video encoder end.

  The current picture decoding unit 330 decodes the encoded data of the current picture according to the dependency ID. For this, the current picture decoding unit 330 may include an inverse quantization unit 331, an inverse transform unit 332, and an inverse prediction unit 333. The operation of each component of the current picture decoding unit 330 is the same as that of the reference picture decoding unit 340. However, the inverse prediction unit 333 uses the restored reference picture as a prediction signal, and restores the current picture from the restored residual signal of the current picture provided from the inverse transform unit 332 (the residual signal and the prediction signal are to add). At this time, the dependency ID read by the bitstream parser 320 or the modified dependency ID is used. The dependency ID indicates the quality layer (first quality layer) related to the reference picture necessary for restoring the quality layer (second quality layer) of the current picture.

  As described above, the picture used in the present invention means one frame. However, in this specification, the picture is H.264. Those skilled in the art will appreciate that the concepts introduced after H.264 may be replaced by slices.

  To date, each of the components shown in FIGS. 12 to 14 is a task, class, subroutine, process, object, execution thread (execution thread), software such as a program, or FPGA (FPGA) It can be implemented by hardware such as a field-programmable gate array (ASIC) or application-specific integrated circuit (ASIC), or can be configured by a combination of the software and hardware. The components can be included in a computer-readable storage medium, or a part of the components can be distributed and distributed over a plurality of computers.

  The embodiments of the present invention have been described above with reference to the accompanying drawings. However, those who have ordinary knowledge in the technical field to which the present invention belongs do not change the technical idea and essential features of the present invention. It can be understood that, in scope, it can be implemented in other specific forms. Therefore, it should be understood that the above embodiment is illustrative in all aspects and not limiting.

It is a figure which shows an example of the scalable video coding system using a multi-hierarchy structure. It is a figure which shows the concept of the inter prediction and intra-base prediction technique of scalable video coding. It is a figure which shows the example which applied the FGS technique to the residual picture by the prediction of FIG. It is a figure which shows the process of expressing one picture or slice by one discrete hierarchy and two FGS hierarchies. It is a figure which shows the system which cuts out the quality hierarchy used by the present SVC standard. It is a figure which shows the system which cuts out the quality hierarchy used by the present SVC standard. It is a figure which shows the structure of the conventional SVC system. It is a figure which shows the structure of the SVC system by one Embodiment of this invention. It is a figure which shows the example which cuts out the quality hierarchy by one Embodiment of this invention. FIG. 3 is a diagram illustrating a bitstream assigned a priority ID according to an embodiment of the present invention. It is a figure which shows the case where the quality hierarchy which dependence ID instruct | indicates does not exist in a reference picture. It is a block diagram which shows the structure of the priority assignment apparatus by one Embodiment of this invention. It is a block diagram which shows the structure of the bit stream extractor by one Embodiment of this invention. It is a block diagram which shows the structure of the video decoder by one Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Priority allocation apparatus 110 Current picture encoding part 111,121 Prediction part 112,122 Conversion part 113,123 Quantization part 114,124 Quality hierarchy generation part 120 Reference picture encoding part 140 Quality level assignor 150 Entropy encoding part 200 Bit stream extractor 210 Bit stream input unit 220, 320 Bit stream parser 230 Bit stream cutting unit 240 Target bit rate setting unit 250 Bit stream transmission unit 300 Video decoder 310 Entropy decoding unit 330 Current picture decoding unit 331, 341 Inverse quantum 332, 342 Inverse transform unit 333, 343 Inverse prediction unit 340 Reference picture decoding unit 350 Dependency ID setting unit

Claims (19)

A step in which the reference picture encoding unit configures a first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to the reference picture by repeating inverse quantization and quantization;
A current picture encoding unit performs dequantization and quantization on a second quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a current picture encoded with reference to the reference picture. And a quality level assigner assigns priority to each quality hierarchy including each of the first quality hierarchy and the second quality hierarchy, and
A method of assigning a priority for adjusting a bit rate of a bitstream, wherein a lower priority is assigned to a quality layer that has a small influence on a reduction in image quality of the current picture even if it is removed. The reference picture and current picture are:
The method of claim 1, wherein the bit rate is a frame or a slice.   The method of claim 1, wherein the reference picture and the current picture have different resolutions or different picture order counts. The first quality hierarchy and the second quality hierarchy are:
The method of assigning priority for bit rate adjustment of a bitstream according to claim 1, comprising one discrete layer and at least one FGS layer. The step of configuring the first quality hierarchy and the step of configuring the second quality hierarchy include:
Predicting the reference picture or the current picture to obtain a residual signal;
Transforming the residual signal to generate transform coefficients;
Quantizing the transform coefficient with a first quantization parameter to form a discrete hierarchy;
Subtracting a result obtained by dequantizing the discrete layer with the residual signal, and quantizing the subtracted result with a second quantization parameter to form the one or more FGS layers. The method of assigning priority for bit rate adjustment of a bitstream according to claim 4. The quality hierarchy having a small influence on the image quality reduction is
The method for assigning priority for bit rate adjustment of a bitstream according to claim 1, wherein a cost required for encoding is a quality layer that is smaller than other quality layers. The cost is
7. The method of assigning a priority for adjusting a bit rate of a bit stream according to claim 6, comprising a difference from an original image and addition of a bit amount required for encoding. A bitstream input unit receiving a video bitstream input;
A target bit rate setting unit setting a target bit rate for the video bitstream;
A bitstream parser includes a first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture of the current picture from the video bitstream, and a quality for each of the current picture Obtaining a second quality layer including a quality layer representing a plurality of quantized pictures with different bit rates, and a bitstream cutting unit, according to the target bit rate, for each of the first quality layer and the second quality layer methods of modulating among the quality layers, including the steps of removing the priority is low quality layer indicating a current small influence on the picture quality reduction of the picture when it is removed, the bit rate of the bitstream containing . The reference picture and current picture are:
The method of adjusting a bit rate of a bitstream according to claim 8, which is a frame or a slice.   The method of claim 8, wherein the reference picture and the current picture have different resolutions or different picture order counts. The first quality hierarchy and the second quality hierarchy are:
The method of adjusting a bit rate of a bitstream according to claim 8, comprising one discrete layer and at least one FGS layer. A bitstream input unit receiving a video bitstream input;
A bitstream parser has a quality for each of a first quality layer including a quality layer representing a plurality of quantized pictures having different qualities from a reference picture of the current picture from the video bitstream and the current picture, respectively. Obtaining a dependency ID indicating a reference relationship between a second quality layer including a quality layer representing a plurality of different quantized pictures and a picture;
The dependency ID setting unit indicates a small influence on the image quality reduction of the current picture when the dependency ID setting unit is removed from the quality layers including each of the first quality layer and the second quality layer according to a target bit rate. If the quality hierarchy indicated by the dependency ID does not exist in the first quality hierarchy by removing the quality hierarchy having a low priority , the dependency ID is the highest in the remaining quality hierarchy of the first quality hierarchy. step configured to instruct the quality layer, and the current picture decoding unit, including the steps of restoring a current picture by reference relation in which the dependency ID is indicated, the video decoding method. The reference picture and current picture are:
The video decoding method according to claim 12, wherein the video decoding method is a frame or a slice. The reference picture and the current picture are:
The video decoding method according to claim 12, wherein the resolutions are different from each other or the picture order counts are different. The first quality hierarchy and the second quality hierarchy are:
The video decoding method according to claim 12, comprising one discrete layer and at least one FGS layer. Restoring the current picture comprises:
Restoring the reference picture according to a reference relationship indicated by the dependency ID;
The video decoding method according to claim 12, comprising: restoring the residual signal of the current picture; and adding the restored reference picture and the restored residual signal. A reference picture encoding unit that configures a first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture by repeating inverse quantization and quantization;
A second quality layer including a quality layer representing a plurality of quantized pictures having different qualities is configured by repeating inverse quantization and quantization for the current picture encoded with reference to the reference picture. A current picture encoding unit, and a quality level allocator that assigns a priority to each quality layer including each of the first quality layer and the second quality layer,
An apparatus for assigning priority for adjusting a bit rate of a bitstream, wherein the priority is assigned to a quality layer having a small influence on a reduction in image quality of the current picture even if the bitstream is removed. A bitstream input unit for receiving a video bitstream input;
A target bit rate setting unit for setting a target bit rate for the video bit stream;
A first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture of the current picture from the video bitstream, and a plurality of quanta having different qualities with respect to the current picture, respectively. A bitstream parser that obtains a second quality layer including a quality layer representing a normalized picture, and is removed from each quality layer including each of the first quality layer and the second quality layer according to the target bit rate And a bitstream cutting unit that removes a quality layer having a low priority indicating a small influence on the reduction in image quality of the current picture in the case of adjusting the bit rate of the bitstream. An entropy decoding unit that receives an input of a video bitstream;
A first quality layer including a quality layer representing a plurality of quantized pictures having different qualities with respect to a reference picture of the current picture from the video bitstream, and a plurality of quanta having different qualities with respect to the current picture, respectively. A bitstream parser that obtains a second quality layer that includes a quality layer that represents a converted picture, and a dependency ID that indicates a reference relationship between pictures in the second quality layer;
Of the quality layers including each of the first quality layer and the second quality layer according to the target bit rate, a quality layer having a low priority indicating a small influence on the image quality reduction of the current picture when removed. by but be removed, when the quality layer in which the dependency ID instructs one of the first quality layer is not present, the dependency ID is to direct the highest quality layer of the remaining quality layers of the first quality layer And a current picture decoding unit that restores a current picture according to a reference relationship indicated by the dependency ID.

Images