WO2013000304A1

WO2013000304A1 - Loop filtering encoding and decoding methods and devices

Info

Publication number: WO2013000304A1
Application number: PCT/CN2012/073111
Authority: WO
Inventors: 李明; 吴平; 张雯
Original assignee: 中兴通讯股份有限公司
Priority date: 2011-06-28
Filing date: 2012-03-27
Publication date: 2013-01-03
Also published as: CN102857746A; CN102857746B

Abstract

Disclosed is a loop filtering encoding method. The method comprises: setting a loop filtering mode for indicating an adopted manner for loop filtering during encoding; generating a parameter set according to the loop filtering mode; and performing encoding according to the generated parameter set. Correspondingly, a loop filtering decoding method, encoder, decoder and an electronic apparatus are further disclosed. The present invention implements the optimization and distribution of the complexity among images during the encoding and decoding, thereby enhancing the compromised performance between the encoder and decoder complexity and the encoding efficiency.

Description

Loop filter codec method and device

The present invention relates to video compression coding techniques, and in particular, to a loop filtering codec method and apparatus. Background technique

Currently, a typical application of loop filtering is an adaptive loop filter (ALF). The existing loop filtering method will be described by taking ALF as an example. In video compression coding, the encoder and decoder filter the decoded restored image by using ALF to further reduce the quantization noise of the restored image and improve the restored image quality. Since the decoded restored image will be used as a reference image for the motion compensation prediction process, ALF can also help to improve the efficiency of motion prediction and improve coding efficiency.

The encoding method of the ALF related information in the prior art may mainly include the following steps: Step 1: In the sequence parameter set (SPS, Sequence Parameter Set), whether the current video sequence can use the ALF identification information and the necessary parameter information. If the ALF identification information is allowed, go to step 2; otherwise, the ALF related information is encoded. Step 2: In the header information of the slice, encode whether the current fragment uses the identification information of the ALF. If the current fragment uses the identification information of the ALF, step 3 is performed; otherwise, the ALF related information encoding process of the fragment ends. Step 3: further encode other related information of the ALF, such as a filter shape, a filter coefficient, a coding unit (CU, Coding Unit), ALF control information, and the like.

The actual test shows that the use of ALF can effectively improve the coding efficiency. However, the filtering method using the SPS overall ALF identification has a large computational and storage complexity. In particular, for decoders using ALF, ALF brings an average of 70% complexity. In some typical configurations, the average decoding complexity of ALF can be as high as 90%. Although the encoder uses slice layer adaptive ALF The control method still cannot effectively achieve a compromise between complexity and coding efficiency, and cannot fully exert the contribution of ALF to coding efficiency. Specifically, the existing loop filtering method mainly has the following first one, and adopts the method of using the SSPPSS whole body AALLFF mark identification method, and although the tube can be used as the code encoder The optimization of the AALLFF module block provides a greater flexibility for the larger, but it does not provide an efficient and efficient way to provide the decoupling code for complex complexity and cocoa. Stretching and contracting, can not be flexible and flexible control of the complex complexity and the efficiency of coding and coding efficiency. . In addition, for the encoding and decoding decoded code for calculating the complex complexity of the low and low calculations, the codec is often used to close the whole body. The method of closing AALLFF"" is to edit and decode the complex complexity of the decoding code, so as to be able to fully utilize the AALLFF to encode the code efficiency rate. Contribute to contribute. .

The second and second, by the method of encoding the code "" by using the method of identifying the whole sequence of the parameter sequence set parameter set, whether or not to use the mark of AALLFF"" Information message, when it is necessary to control the complex complexity of the control, only need to be on some certain image images, or or image image groups, or some time The picture image in the inter-segment segment makes it necessary to use the code AALLFF in the phase-related information of the other relevant picture image. The identification information of the identification information is such that, in the code stream, the information stored in the code stream is redundant and unnecessary. .

The third third, AALLFF's contribution to the coding efficiency efficiency of the coding is mainly reflected in the reduction of the low image image quantized noise noise and the improvement of the high movement movement compensation compensation The pre-predictive reference reference image has two or two aspects of image quality, and the latter often tends to play a leading role. . As far as the process of motion compensation compensation is concerned, the current process efficiency rate of the single-element cocoa reflection image image pair is processed. The impact of the sound. . However, the existing method has adopted the whole sequence label identification with the sequence parameter parameters, and has not been able to fully subdivide the AALLFF used with the use of cocoa and the existing language. The ambiguity between the grammar method and the unit cell includes the method of deriving the guide method, which saves the redundant information of the letter information that is not necessary. . Invent the contents of the invention

In view of this, the main main objective of the present invention is to provide a method for encoding and decoding a code method for a loop loop filter filter and a device arrangement. , in order to solve the problem, the existing loop circuit filtering filter wave method can not effectively and effectively between the complex complexity and the code efficiency rate Take a questionable question that has been compromised. .

In order to achieve the above stated objectives, the technical solution of the present invention is realized in this way:

The present invention provides a method for a loop loop filter filtering code encoding code method, and the method includes:

The setting is used to indicate that the loop circuit filter filter wave is in the process of coding and code passing process, and the ring loop filter filter mode is used in the process mode; In the above solution, the method is:

Determining two or more of the loop filtering modes that can be used in the encoding process; for each of the loop filtering modes, generating one or more of the loop filtering identification information including the loop filtering mode, respectively Parameter set

The loop filtering mode used in the current encoding process is determined and encoded according to the parameter set corresponding to the loop filtering mode used in the current encoding process.

In the above solution, the two or more loop filtering modes include any two or more of the following loop filtering modes:

Represents a zeroth loop filtering mode that allows loop filtering for each slice;

Representing a first loop filtering mode that allows only intra-coded slices to use loop filtering; represents a second loop filtering mode that allows loop filtering only for slices in the image used for inter prediction reference;

Representing a third loop filtering mode that allows loop filtering in the slice when the sliced picture number (POC) value is an integer multiple of the loop filter refresh period;

Indicates that the fourth loop filtering mode that uses loop filtering is not allowed for all slices.

In the above solution, the method is:

Determining the loop filtering mode used by the current encoding process;

Generating a parameter set containing loop filter identification information having a determined loop filtering mode; encoding according to the generated parameter set.

In the above solution, the loop filtering mode is any one of the following loop filtering modes: indicating a zeroth loop filtering mode that allows loop filtering for each slice;

Representing a first loop filtering mode that allows only intra-coded slices to use loop filtering; represents a second loop filtering mode that allows loop filtering only for slices in an image used for inter prediction reference; Representing a third loop filtering mode that allows loop filtering in the slice when the POC value of the slice is an integer multiple of the loop filter refresh period;

In the above solution, the determining a loop filtering mode for indicating a manner in which the loop filtering is used in the encoding process is: according to one or more of the following factors: specific application requirements, available computing resources of the encoding process And a storage resource, a video source characteristic, a source coding rate, and a receiving end processing capability, and determining a loop filtering mode for indicating a manner in which the loop filtering is used in the encoding process.

In the above solution, after the generating the parameter set including the loop filtering identification information of the loop filtering mode, the method further includes: allocating a parameter set for identifying the parameter set to the generated parameter set. Index number.

In the above solution, the encoding according to the generated parameter set includes: writing information of the parameter set into a code stream, and writing the parameter set index number into the current coding unit in a slice layer encoding process. The slice header information in the slice, and the filter used for encoding is determined, and the parameters of the filter are written to the slice header information.

In the above solution, the determining the filter used for encoding includes:

All candidate filters are grouped according to the complexity and denoising performance of each candidate filter to obtain one or more candidate filter banks;

Calculating an image importance parameter of the current coding unit;

A filter used for encoding is selected for the current coding unit based on the calculated image importance parameter and the resulting one or more candidate filter sets.

In the above solution, the method further includes: determining, by the risk, whether the loop filtering mode determined by the current coding unit is the same as the loop filtering mode determined when the previous coding unit performs coding; if not, adjusting The currently used parameter set; if it is, the currently used parameter set is not adjusted.

In the above solution, the adjusting the currently used parameter set includes: verifying the current corresponding Whether the parameter set of the loop filtering mode is determined when the code unit performs coding, and if so, directly adjusts the currently used parameter set to a parameter set corresponding to the loop filtering mode determined when the current coding unit performs coding; if not, generates The current coding unit performs a parameter set corresponding to the loop filtering mode determined by the coding unit, and adjusts the currently used parameter set to a parameter set corresponding to the loop filtering mode determined when the current coding unit performs coding.

In the above solution, before the risk filtering mode determined by the current coding unit is encoded, whether the loop filtering mode determined when the previous coding unit performs coding is the same, the method further includes: The current resource information in the current encoding process is obtained, and the available resource information in the current encoding process is obtained. According to the available resource information, the loop filtering mode when the current coding unit performs encoding is determined.

In the above solution, the method further includes: determining loop filtering used by each image or each video segment in the current encoding process according to the obtained source characteristic parameter and the encoder control auxiliary parameter in the last encoding process. Mode: encoding each image or each video segment according to a parameter set corresponding to a loop filtering mode used by each image or each video segment in the current encoding process.

In the above solution, the method further includes: reallocating a parameter set index sequence number for a parameter set corresponding to a loop filtering mode used by each image or each video segment in the current encoding process;

And respectively encoding, according to the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process, the image or each video segment is: according to the reassignment parameter set index sequence number Each of the obtained parameter sets respectively encodes each of the images or video segments.

In the above solution, the parameter set index number is reassigned to the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process, including:

And sorting the loop filtering modes according to how many times the loop filtering mode of each image or each video segment is used in the current encoding process, according to how many times of use; According to the sorting order of the loop filtering modes, the parameter sets corresponding to the loop filtering modes are sequentially assigned index numbers of d to large parameter sets.

The present invention also provides an encoder, the encoder comprising: a determining module, a generating module, and an encoding module;

a determining module, configured to set a loop filtering mode for indicating a manner in which loop filtering is used in the encoding process;

a generating module, configured to generate a parameter set according to the loop filtering mode set by the determining module; and an encoding module, configured to perform encoding according to the parameter set generated by the generating module.

In the above solution, the determining module is further configured to determine two or more types of the loop filtering modes that can be used in the encoding process;

a generating module, configured to generate, for each loop filtering mode determined by the determining module, one or more parameter sets respectively including loop filtering identification information of the loop filtering mode; and an encoding module, further used for The loop filtering mode used in the current encoding process is determined and encoded according to the parameter set corresponding to the loop filtering mode used in the current encoding process.

In the above solution, the determining module is further configured to determine a loop filtering mode used by the current encoding process;

a generating module, configured to generate a parameter set including loop filtering identifier information of a loop filtering mode determined by the determining module;

The encoding module is further configured to perform encoding according to the parameter set generated by the generating module.

In the above solution, the determining module is further configured to determine, according to one or more of the following factors, a loop filtering mode for indicating a manner in which the loop filtering is used in the encoding process: specific application requirements, an encoding process Available computing resources and storage resources, video source characteristics, source code rate, and receiver processing capabilities.

In the above solution, the generating module is further configured to allocate, to the generated parameter set, a parameter set index sequence number for identifying the parameter set; The encoding module is further configured to write the information of the parameter set into the code stream, and write the parameter set index number into the slice layer header information in the current coding unit in the slice layer coding process, and determine the code. A filter is used to write parameters of the filter to the slice header information.

In the above solution, the encoding module is further configured to determine a filter used for encoding by using the following method:

Calculating an image importance parameter of the current coding unit;

In the above solution, the encoder further includes a risk module and an adjustment module, wherein the adjustment module is configured to adjust a currently used parameter set;

And a verification module, configured to verify whether the loop filtering mode determined when the current coding unit performs coding is the same as the loop filtering mode determined when the previous coding unit performs coding, and starts the adjustment module without being identical.

In the above solution, the adjusting module is further configured to: if the risk parameter corresponds to a parameter set of the loop filtering mode determined by the current coding unit, if yes, directly adjust the currently used parameter set to the current coding unit. And a parameter set corresponding to the loop filtering mode determined by the encoding; if not, generating, by the generating module, a parameter set corresponding to the loop filtering mode determined by the current encoding unit, and adjusting the currently used parameter set to the current encoding The parameter set corresponding to the loop filtering mode determined when the unit performs encoding.

In the above solution, the determining module is further configured to determine, according to the obtained source characteristic parameter in the last encoding process, and the encoder control auxiliary parameter, a ring used by each image or each video segment in the current encoding process. Road filtering mode

The encoding module is further configured to determine, according to the determining module, a current encoding process Each image or each video segment is encoded by a parameter set corresponding to a loop filtering mode used by each image or each video segment.

The invention also provides a loop filter decoding method, the method comprising:

Decoding the loop filtering identification information of the loop filtering mode in the parameter set related code stream; decoding the loop filtering identification information of the loop filtering mode in the slice layer;

The loop filtering mode is used to indicate the manner in which loop filtering is used in the encoding process. In the foregoing solution, the decoding, by the loop filter identification mode, in the loop filter mode of the parameter set related code stream, includes: reading bits in the parameter set related code stream, and obtaining the loop filter identification information used to identify the loop filter identification information The values of the syntax units are assigned, and the values of the syntax units are assigned to corresponding variables in the parameter set data structure.

In the foregoing solution, the decoding, by the loop filtering identifier information of the loop filtering mode in the fragmentation layer, includes: obtaining values of the syntax units used to identify the loop filtering identifier information in the parameter set, and Passing the value of the syntax unit to the corresponding variable of the slice layer data structure.

In the above solution, the method further includes: parsing the loop filter related parameters in the slice layer according to the decoded loop filter identification information.

The present invention also provides a decoder, the decoder comprising: a first decoding unit and a second decoding unit, wherein

a first decoding unit, configured to decode loop filter identification information of a loop filtering mode in a parameter set related code stream;

a second decoding unit, configured to decode loop filter identification information of a loop filtering mode in the slice layer;

The loop filtering mode is used to indicate the manner in which loop filtering is used in the encoding process. In the foregoing solution, the first decoding unit is further configured to: read a bit in a parameter set related code stream, and obtain a value of each syntax unit used to identify the loop filter identification information, and The values of the syntax units are assigned to corresponding variables in the parameter set data structure.

In the foregoing solution, the second decoding unit is further configured to obtain a value of each syntax unit used to identify the loop filtering identifier information in the parameter set, and transmit the value of the syntax unit to a point. The corresponding variable of the slice data structure.

In the above solution, the second decoding unit is further configured to parse the loop filter related parameters in the slice layer according to the decoded loop filter identification information.

The present invention also provides an electronic device, the electronic device comprising an encoder and/or a decoder, wherein the encoder comprises a determining module, a generating module and an encoding module, and the determining module is configured to set a loop filtering a loop filtering mode used in the encoding process; the generating module is configured to generate a parameter set according to the loop filtering mode set by the determining module; and the encoding module is configured to perform encoding according to the parameter set generated by the generating module;

The decoder includes a first decoding unit and a second decoding unit, where the first decoding unit is configured to decode the loop filtering identification information of the loop filtering mode in the parameter set related code stream; and the second decoding unit is configured to perform the splitting The loop filtering identification information of the loop filtering mode in the slice layer is decoded; wherein the loop filtering mode is used to indicate the manner in which the loop filtering is used in the encoding process.

The loop filtering codec method and apparatus provided by the present invention generates a parameter set of a loop filtering mode determined by a corresponding encoding process, and performs encoding according to the generated parameter set, so that a preset according to the video application can be One or more loop filtering modes are encoded, which realizes optimal allocation of inter-image complexity in the process of encoding and decoding, and improves the tradeoff between codec complexity and coding efficiency.

In addition, the present invention also uses the existing semantics to apply the implicit derivation of factors such as image importance and processing complexity to the coding and decoding process of the loop filtering, and reduces the overhead of the loop filtering related header information.

In addition, the present invention can dynamically adjust the loop filtering mode during the encoding process, so as to facilitate the complexity of loop filtering between different importance images, and improve the competing performance of the codec computational complexity and coding efficiency. DRAWINGS

1 is a flowchart of an implementation of a loop filter encoding method according to the present invention;

2 is a flowchart of parsing ALF identification information added in a PPS according to Embodiment 1 of the present invention;

FIG. 3 is a flowchart of analyzing ALF identification information in slice layer information according to Embodiment 1 of the present invention; FIG. 4 is a flowchart showing implementation of coding once in a single ALF mode according to Embodiment 2 of the present invention;

FIG. 5 is a flowchart of implementing one-time coding in multiple ALF modes according to Embodiment 3 of the present invention; FIG. 6 is a flowchart of implementing a filter used in determining an encoding process according to Embodiment 4 of the present invention; FIG. 7 is a flowchart of Embodiment 5 of the present invention; A flowchart for implementing the encoding of the video image twice. detailed description

The "compromise between complexity and coding efficiency" and its similar or related expressions involved in the present invention include the following three aspects: First, the improvement of coding efficiency is obtained at the expense of reasonable complexity increase or even decrease. The reduction in complexity is obtained at the expense of negligible loss of coding efficiency or even increased coding efficiency. Second, the encoder or the decoder allocates resources such as calculation and storage reasonably, and increases the coding efficiency under the premise of increasing or even decreasing the complexity; or reducing the encoder under the premise that the loss of coding efficiency is negligible or even improved. Or the overall complexity of the decoder. Third, the encoder or the decoder uses a processing method with different complexity in encoding or decoding different images, different shards, and different basic coding units, and obtains an increase in coding efficiency at the cost of a reasonable overall complexity increase or even decrease. The overall complexity of the encoder or decoder is reduced at the expense of negligible loss of coding efficiency or even increased coding efficiency.

The basic idea of the present invention is: setting a plurality of loop filtering modes for indicating a loop filtering usage manner according to an application of video communication, and generating a plurality of parameter sets corresponding to the loop filtering modes; The parameter set is encoded in a parameter set corresponding to the loop filtering mode of the current encoding process. In this way, encoder complexity allocation and optimization can be achieved, and the encoding process is different. The complexity of loop filtering between importance images increases the trade-off performance of codec computational complexity and coding efficiency, and reduces information redundancy in the code stream.

The loop filter coding method of the present invention, as shown in FIG. 1, may mainly include the following steps: Step 101: Set a loop filtering mode for indicating a manner in which loop filtering is used in the encoding process;

Step 102: Generate a parameter set according to the loop filtering mode.

Step 103: Perform coding according to the generated parameter set.

The parameter set is a data structure including some image common parameters, and a flag and a parameter set in the parameter set identified by the slice header information in the image decoding process. For example, a sequence parameter set, an image parameter set, or a data structure with similar application properties.

Specifically, the method can be implemented in the following two manners:

First, determining two or more of the loop filtering modes that can be used in the encoding process; for each of the loop filtering modes, respectively generating one of loop filtering identification information including the loop filtering mode Or a plurality of parameter sets; determining a loop filtering mode used in the current encoding process and encoding according to a parameter set corresponding to the loop filtering mode used in the current encoding process.

Second, determining a loop filtering mode used in the current encoding process; generating a parameter set including loop filtering identification information of the determined loop filtering mode; and encoding according to the generated parameter set.

The loop filtering identifier information of the loop filtering mode includes a parameter for indicating a loop filtering identifier manner, or a parameter for indicating a loop filtering identifier manner and an auxiliary parameter of the parameter. The parameter used to indicate the loop filtering identification mode may include a parameter for identifying whether to allow all fragments in the image to use loop filtering, and a parameter for identifying a manner in which all fragments in the image use loop filtering; The auxiliary parameter is specifically an auxiliary parameter for identifying a parameter of a manner in which all slices in the image use loop filtering, and may include a refresh cycle of loop filtering. Period.

The loop filtering mode may be constructed according to other syntax units in a code stream organization such as a slice layer or a coding unit layer. In the encoding process of the slice, loop filtering is allowed only when the construction conditions are true. The loop filtering mode may be one or more of the following loop filtering modes: indicating a zeroth loop filtering mode that allows loop filtering for each slice, indicating that only intra-coded slices are allowed to use the loop The filtered first loop filtering mode, indicating that only the second loop filtering mode of the loop filtering is used for the slice in the image used for the inter prediction reference, indicating the image sequence number (POC, Picture Order Count) in the slice When the value is an integer multiple of the loop filtering refresh period, the third loop filtering mode that uses loop filtering in the slice is allowed to indicate that all slices are not allowed to use the fourth loop filtering mode of loop filtering.

The zero-loop filtering mode: loop filtering is allowed for each slice, and the filter parameters and loops are determined using slice layer adaptive techniques according to the constraints in the profile and level. The filtered image area is suitable for high coding efficiency and the codec allows the use of high complexity algorithms. For example, use a PC for non-real-time video downloads or on-demand, home theater, and more.

First loop filtering mode: Only intra-coded slices allow the use of loop filtering, using slice layer adaptive techniques to determine filter parameters and image regions using loop filtering, depending on the constraints in the grade and level, Applications with low decoding complexity or low coding complexity, such as mobile terminal video playback, camcorders, etc.

Second loop filtering mode: Fragmentation in an image only for inter prediction reference allows the use of loop filtering, using slice layer adaptive techniques to determine filter parameters and using loop filtering, depending on the constraints in the grade and level Image area. This mode is suitable for medium complexity decoding or coding applications.

Third loop filtering mode: Periodic loop filter refresh mode, which allows the ring to be used in the encoding and decoding process of the fragment only when the POC value of the fragment is an integer multiple of the loop filtering refresh period. Path filtering, and using slice layer adaptive techniques to determine filter parameters and image regions using loop filtering, depending on the constraints in the grade and level. This mode can be applied to low, medium and high complexity decoding and coding applications by further setting the loop filter refresh period. The loop filtering identification information of the mode needs to include an auxiliary parameter, which is a loop filtering refresh period.

Wherein, for any one of the two or more loop filtering modes, one or more parameter sets corresponding to the loop filtering mode may be generated. For example, for the third loop filtering mode described above, a plurality of parameter sets may be generated according to different loop filtering refresh periods. For the zeroth loop filtering mode, the first loop filtering mode, and the second loop filtering mode described above, a corresponding parameter set may be generated corresponding to each loop filtering mode.

Wherein, it may be determined according to one or more of the following factors: specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, source coding rate, and receiving end processing capability, and are determined for the indication ring. The path filtering uses a loop filtering mode in the encoding process.

After generating a parameter set including the loop filter identification information of the loop filtering mode, a parameter set index number for identifying the parameter set is also allocated to the generated parameter set.

Specifically, performing coding according to the generated parameter set may include: writing information of the parameter set into a code stream, and writing the parameter set index sequence number into a current coding unit in a slice layer coding process. The slice header information, and a filter used for encoding is determined, and parameters of the filter are written to the slice header information.

Here, the process of determining a filter used for encoding may include: grouping all candidate filters according to complexity and denoising performance of each candidate filter to obtain one or more candidate filter banks; An image importance parameter of the coding unit; a filter used for encoding the current coding unit is selected based on the calculated image importance parameter and the resulting one or more candidate filter banks. In addition, it is also possible to determine the filter used for encoding by using the existing slice layer candidate loop filter selection and control method.

Here, the parameters of the filter are written into the slice header information, which will be used to represent the points. Whether there is identification information of the loop filter parameter, loop filter related filter parameters, and

CU layer ALF controls identification information and the like. Among them, the loop filter correlation filter parameter related code stream may also appear in the parameter set (such as PPS or other parameter set containing image/slice common information). Here, the slice layer code stream is taken as an example for description.

The loop filter related filter parameters may include a filter adaptation method, a filter coefficient, a filter type information, a filter coefficient predictive coding method, and the like, and the CU layer ALF control identification information may include whether to use the filter CU layer control. Identification information, filter CU layer control method, and identification information of whether or not the relevant CU uses a filter to perform a filtering operation.

The method may further include: determining, by the risk, whether the loop filtering mode determined by the current coding unit is the same as the loop filtering mode determined when the previous coding unit performs coding; if not, adjusting the current use The parameter set; if it is, the currently used parameter set is not adjusted.

Specifically, verifying whether a parameter set corresponding to the loop filtering mode determined by the current coding unit is encoded exists, and if yes, directly adjusting the currently used parameter set to a parameter corresponding to the loop filtering mode determined by the current coding unit for encoding If not, generating a parameter set corresponding to the loop filtering mode determined by the current coding unit for encoding, and adjusting the currently used parameter set to a parameter set corresponding to the loop filtering mode determined by the current coding unit for encoding.

Here, the loop filtering mode determined when the current coding unit performs coding is still dynamically monitored for the change of available resources in the current encoding process before the loop filtering mode determined when the previous coding unit performs encoding is the same. Obtaining the available resource information in the current encoding process; determining, according to the available resource information, a loop filtering mode when the current coding unit performs encoding.

The method may further include: determining, according to the obtained source characteristic parameter in the last encoding process, and the encoder control auxiliary parameter, a loop filtering mode used by each image or each video segment in the current encoding process; According to each image or each video segment in the current encoding process Each of the images or video segments is encoded separately using a parameter set corresponding to the loop filtering mode.

Here, the parameter set index number may be re-allocated for the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process; and each parameter set obtained by re-allocating the parameter set index number is respectively Each of the images or video segments is encoded.

Specifically, according to the number of times the loop filtering mode of each image or each video segment is used in the current encoding process, the loop filtering modes are sorted according to the number of times of use; The sorting order of the loop filtering mode is to sequentially assign a parameter number of d to a large parameter set index number for the parameter set corresponding to each loop filtering mode.

Correspondingly, the present invention further provides an encoder, where the encoder includes: a determining module, a generating module, and an encoding module; wherein, the determining module is configured to set a ring for indicating a manner in which the loop filtering is used in the encoding process. a filtering module, configured to generate a parameter set according to the loop filtering mode set by the determining module, and an encoding module, configured to perform encoding according to the parameter set generated by the generating module.

Specifically, the determining module is further configured to determine two or more types of the loop filtering modes that can be used in the encoding process, and the generating module is further configured to use the loop filtering mode determined by the determining module. Generating, respectively, one or more parameter sets including loop filtering identification information of the loop filtering mode; the encoding module is further configured to determine a loop filtering mode used in the current encoding process and according to the current encoding process The coding is performed using a parameter set corresponding to the loop filtering mode.

Specifically, the determining module is further configured to determine a loop filtering mode used by the current encoding process, and the generating module is further configured to generate a parameter that includes loop filtering identifier information of the loop filtering mode determined by the determining module. The encoding module is further configured to perform encoding according to the parameter set generated by the generating module.

The determining module is further configured to determine according to one or more of the following factors: The loop filtering mode indicating the manner in which the loop filtering is used in the encoding process: specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, source coding rate, and receiving end processing capability.

The generating module is further configured to allocate a parameter set index number for identifying the parameter set to the generated parameter set; the encoding module is further configured to write the information of the parameter set into the code stream, Writing the parameter set index number into the slice header information in the current coding unit in the slice layer encoding process, and determining a filter used for encoding, and writing parameters of the filter to the slice layer Header information.

The coding module is further configured to determine, by using the following method, a filter used for coding: grouping all candidate filters according to complexity and denoising performance of each candidate filter to obtain one or more candidate filter banks. Calculating an image importance parameter of the current coding unit; selecting a filter used for encoding for the current coding unit according to the calculated image importance parameter and the obtained one or more candidate filter banks.

The encoder further includes a verification module and an adjustment module, where the adjustment module is configured to adjust a currently used parameter set; and the verification module is configured to verify a loop filtering mode determined when the current coding unit performs coding, and Whether the loop filtering modes determined by one coding unit when encoding are the same, and when they are not the same, the adjustment module is started.

Specifically, the adjusting module is further configured to: if the risk parameter corresponds to a parameter set of the loop filtering mode determined by the current coding unit, if yes, directly adjusting the currently used parameter set to the current coding unit for encoding And determining, by the generating module, a parameter set corresponding to the loop filtering mode determined by the current coding unit, and adjusting the currently used parameter set to the current coding unit The parameter set corresponding to the loop filtering mode determined at the time of encoding.

The determining module is further configured to determine, according to the obtained source characteristic parameter and the encoder control auxiliary parameter in the last encoding process, each image or each video segment in the current encoding process a looping filter mode; the encoding module is further configured to: according to the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process determined by the determining module, respectively Each image or each video segment is encoded.

The present invention further provides a loop filter decoding method corresponding to the above coding method, the method comprising: decoding loop filter identification information of a loop filter mode in a parameter set related code stream; and performing a loop filter mode in a slice layer The loop filtering identification information is decoded; wherein the loop filtering mode is used to indicate the manner in which the loop filtering is used in the encoding process.

Specifically, the decoding, in the loop filter mode, the loop filter identification information in the parameter set related code stream, includes: reading bits in the parameter set related code stream, and obtaining each of the loop filter identification information used to identify the loop filter identification information The values of the syntax units are assigned, and the values of the syntax units are assigned to corresponding variables in the parameter set data structure.

Decoding the loop filter identification information of the loop filtering mode in the slice layer, including: obtaining values of the syntax units used to identify the loop filter identification information in the parameter set, and using the syntax unit The value is passed to the corresponding variable of the slice layer data structure.

The method further includes: parsing the loop filter related parameters in the slice layer according to the decoded loop filter identification information.

Correspondingly, the present invention further provides a decoder, the decoder comprising: a first decoding unit and a second decoding unit, where the first decoding unit is configured to loop a loop filtering mode in a parameter set related code stream The second filtering unit is configured to decode the loop filtering identification information of the loop filtering mode in the slice layer, where the loop filtering mode is used to indicate that the loop filtering is in the encoding process. How to use it.

Specifically, the first decoding unit is further configured to: read a bit in a parameter set related code stream, obtain a value of each syntax unit used to identify the loop filter identification information, and use the syntax unit The value is assigned to the corresponding variable in the parameter set data structure.

Specifically, the second decoding unit is further configured to obtain the parameter set for identifying the The loop filter identifies the value of each syntax unit of the information, and passes the value of the syntax unit to the corresponding variable of the slice layer data structure.

The second decoding unit is further configured to parse the loop filter related parameters in the slice layer according to the decoded loop filter identification information.

The method of each embodiment will be described below by taking the ALF of the loop filtering application as an example. Other loop filtering applications (e.g., Sample Adaptive Offset Compensation (SAO)) can also achieve a trade-off between complexity and coding efficiency using the method of the present invention and corresponding embodiments.

Embodiment 1

In this embodiment, the method for identifying and configuring the loop filter mode in the parameter set and the slice header information, and the method for analyzing the loop filter identification information corresponding to the loop filter mode in the code stream by the decoder are described in detail. .

In this embodiment, the ALF identification information corresponding to the ALF mode is identified by using a method of variable length coding in a picture parameter set (PPS, Picture Parameter Set) as an example to illustrate the organization and identification of the loop filter mode in the parameter set. method. Similarly, loop filtering modes can be combined and identified in the data structure of other parameter sets.

The ALF identification information includes a parameter for indicating an ALF identification mode, or a parameter for indicating an ALF identification mode and an auxiliary parameter of the parameter. For example, the parameters used to indicate the ALF identification mode may include adaptive_loop_filter_enable_flag and adaptive_loop_filter_pattern_idx in Table 1, and the auxiliary parameters of the parameter may be adaptive_loop in Table 1. Filter_period. Only the auxiliary parameters need to be included in the ALF identification information of ALF mode 3 shown in Table 2.

Specifically, the manner of identifying the ALF identification information in the PPS in this embodiment is as shown in Table 1. Pic_parameter— set— rbsp( ) { Descriptor pic_parameter— set— id ue(v) seq_parameter— set— id ue(v) entropy— coding— mode—flag u(l)

Num—ref—idx— 10— default— active— minusl ue(v) num ref idx 11 default active minusl ue(v) pic_init_qp_minus26 se(v) constrained—intra_pred—flag u(l)

Adaptive— loop— filter— enable—flag u(l)

If (adaptive- loop-filter-enable-flag) {

Adaptive— loop— filter_pattern— idx ue(v) if (adaptive— loop— filter_pattern— idx == 3) {

Adaptive— loop— filter_period ue(v)

}

Rbsp— trailing— bits( )

}

Table 1

In Table 1, the added ALF identification information is shown in lines 9-15, and the "Descriptor" column gives a binary representation of the values of each syntax unit.

In Table 1, pic_parameter_set_id, seq_parameter_set_id, entropy_coding_mode_flag, num_ref_idx-10_ default-active- minusl, num-ref-idx-11-default-active- Miner1, pic_init_qp_minus26, constrained_intra_pred_flag, adaptive_loop_filter-enable_flag, adaptive_loop_filter_pattern_idx, adaptive_loop_filter_period, respectively, ll denotes the name of the syntax unit that needs to be written to the stream.

Specifically, the adaptive_loop_filter_enable_flag is used to identify whether all fragments in the image are allowed to use ALF. When the value of adaptive_loop_filter-enable_flag is 0, ALF cannot be used for all fragments in the image. In this case, it can be considered to correspond to a special ALF mode, which means that all fragments in the image are not available. ALF ; Adaptive—loop—filter—enable—When the value of flag is 1, all fragments in the image can use ALF. Adaptive_loop—filter_pattern—idx is used to identify the ALF mode used by all fragments in the image. The value of adaptive_loop_filter_pattern—idx can be the index number of each ALF mode (such as 0, 1, 2 in Table 2). 3) Specifically, the correspondence between the index number of each ALF mode and the identifier mode of the slice layer ALF information is as shown in Table 2.

Table 2

The POC value of the image in which the fragment is located can be directly calculated according to the relevant parameters in the parameter set and the fragment header information. Here, the relevant parameters in the slice layer information may include: the value of the semantic unit such as pic_order_cnt_type and log2_max_pic_order-cnt-lsb-minus4 in the sequence parameter set, and the semantics in the slice header information Unit pic—order—cnt—lsb.

Adaptive_loop—filter_period is used to identify the refresh rate of ALF, that is, in the order of playback. Every (adaptive-loop-filter_period - 1) image allows ALF to be used for the next image. When the value of adaptive_loop_filter_pattern_idx is 3, the value of adaptive_loop_filter_period needs to be set in PPS. The refresh period of the ALF may be pre-configured according to specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, source coding rate, and receiving end processing capability. For example, for a mobile video application, when the terminal device detects that the battery is fully charged (such as 80% power remaining), it can directly inform the encoder to use a refresh cycle with a smaller value (such as 4, 8, etc.) to obtain a higher compression code. Efficiency; on the other hand, when the terminal device detects that the battery is low (such as 20% power remaining), it can directly inform the encoder to use a larger refresh period (such as 16, 32, etc.) to minimize the power consumption of the encoder, and Guarantee a certain coding efficiency.

The above ALF identification information is decoded by using the binary representation method given by "Descriptor" in Table 1, and the binary symbols read from the code stream are converted into the values of the corresponding syntax unit items. Specifically, as shown in FIG. 2, the process of parsing the ALF identification information added by the decoder in the PPS includes the following steps:

S201, the decoder reads the bit in the PPS-related code stream, and uses the decoding method corresponding to u(l) to obtain the value corresponding to the syntax unit adaptive_loop_filter-enable_flag, and the value obtained by ^ ¹ is assigned to Variable in the data structure of the PPS layer adaptive_loop_filter-enable_flag;

5202, the decoder determines whether the value of the variable adaptive_loop_filter_enable_flag is 1, if yes, executing S203; if not, ending the ALF identification information parsing process of the PPS in the code stream;

5203, the decoder reads the bit in the PPS-related code stream, and uses the decoding method corresponding to ue(v) to obtain the value corresponding to the syntax unit adaptive_loop_filter_pattern_idx, and the value obtained by ^ ¹ is assigned to the PPS layer. The variable in the data structure adaptive_loop_filter_pattern_idx;

5204, the decoder determines whether the value of the variable adaptive_loop_filter_pattem_idx is 3, and if yes, executing S205, and continuing to parse the ALF identification information; otherwise, ending the PPS in the code stream ALF identification information parsing process;

S205, the decoder reads the bit in the PPS-related code stream, and uses the decoding method corresponding to ue(v) to obtain the value corresponding to the syntax unit adaptive_loop_filter_period, and assigns the obtained value to the PPS layer data structure. The variable adaptive_loop_filter_period, and ends the ALF identification information parsing process of the PPS in the code stream.

According to the ALF identification information in the PPS and the fragmentation layer ALF information identification manner corresponding to each ALF mode, the fragmentation layer ALF information identification manner is as shown in Table 3.

Alf_param() }

}

If( deblocking— filter— control_present— flag ) {

Disable— deblocking—filter— idc ue(v) if( disable— deblocking— filter— idc != 1 ) {

Slice— alpha— cO— offset— div2 se(v) slice— beta—offset— div2 se(v)

}

If( slice_ type = = B )

Collocated— from— 10 — flag u(l)

} else

If( entropy— coding— mode— flag && slice— type != I)

Cabac init idc ue(v)

}

table 3

In Table 3, the 27th-35th behavior adds the ALF identification information to the slice layer. Alf_param() is the data structure of the filter-related parameters in the slice layer. The "Descriptor" column gives the binary representation of the values of the syntax units in the slice layer.

In the decoding process of the slice header information, the parsing of the ALF identification information in the slice layer information is a decoding method using the binary representation method given by "Descriptor" in Table 3, converting the binary symbols read from the code stream into Corresponds to the value of the syntax unit item. Specifically, as shown in FIG. 3, the process of parsing the ALF identification information in the fragmentation layer information may specifically include the following steps:

S301, the decoder obtains the value of the ALF related syntax element adaptive_loop_filter-enable_flag in the PPS, and passes the value to the corresponding variable adaptive_loop_filter_enable_flag in the slice layer data structure;

S302, the decoder determines whether the value of the variable adaptive_loop_filter_enable_flag is 1, if yes, step S303 is performed; otherwise, the parsing process of the ALF identification information in the slice layer information is ended;

S303, the decoder obtains the ALF related syntax unit in the PPS Adaptive_loop—filter_pattern—value of idx, and pass its value to the corresponding variable adaptive_loop_filter_pattern_idx in the slice layer data structure;

S304, the decoder determines whether the value of the variable adaptive_loop_filter_pattern_idx is 0, if yes, execute S308; if not, execute S305;

S305, the decoder determines whether the value of the variable adaptive_loop_filter_pattern_idx is 1, and whether the current fragment is an intra-coded fragment, and if so, executing S308; if not, executing S306;

5306, the decoder determines whether the value of the variable adaptive_loop_filter_pattern_idx is 2, and whether the value of the corresponding variable of the syntax unit nal_ref_idc in the NAL data unit of the current fragment is not equal to 0, if , then execute S308; if not, execute S307;

5307. The decoder determines whether the value of the variable adaptive_loop_filter_pattern_idx is 3, and whether the POC value of the image of the current fragment is an integer multiple of the value of the variable adaptive_loop_filter_period in the data structure of the PPS layer. If yes, execute S308; if not, end the parsing process of the ALF identification information in the slice layer information;

S308. The decoder parses the relevant parameters of the ALF in the slice layer, and the current process ends.

Specifically, according to the ALF identification information in the fragmentation layer, it is determined that the relevant parameters of the ALF in the fragmentation layer need to be parsed, and the decoder parses the relevant parameters of the ALF in the fragmentation layer. Here, the related parameters of the ALF may include: Whether the identification information of the ALF parameter, the ALF related filter parameter, the ACL layer ALF control identification information, etc. exist in the slice layer code stream, and the information is included in the data structure alf_param().

In the parsing process of the ALF identification information in the slice layer information, the decoder obtains the value of each syntax unit by using a decoding method corresponding to the binary representation method of each syntax unit in the data structure alf_param(). And pass its value to the corresponding variable in the data structure.

Embodiment 2

The encoder of this embodiment encodes an input video source using a single ALF mode to suit The trade-off between the codec rate and the codec's computational complexity and coding efficiency should be used. As shown in Figure 4, the specific process of encoding once using a single ALF mode includes the following steps:

S401. The sending end determines an ALF mode used when the encoder performs encoding.

Specifically, the primary basis for determining the ALF mode by the transmitting end includes, but is not limited to, one or more of the following factors: specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, and source coding rate. The receiving end processing capability, etc., enables the selection of an ALF mode with a compromise between complexity and coding efficiency.

The specific application requirement is a system performance indicator, a transmission network capability, and a terminal processing capability, which are proposed for ensuring the user experience of the receiving end in the video communication application, and may be mainly included to ensure the quality of the user experience. Set parameter limits, such as end-to-end delay, compression efficiency, etc. The available computing resources and storage resources of the encoding process may include: processor load, available memory storage capacity, power supply system status (such as AC power) Parameters such as power supply, battery power and remaining power, and energy consumption status of the whole machine. These parameters can be obtained through the status monitoring and detection module in the terminal. The video source characteristics are related parameters for reflecting the statistical characteristics of the video source. The method may include: an image activity degree, a coding complexity, a parameter reflecting a motion severity, a parameter reflecting a scene switching, and the like; the source coding code rate is specifically a target code of a source compression coding determined according to a channel state and the like. Rate; the receiving end processing capability includes The estimated information or the estimated information indicating the available computing resources, the power supply mode, and the energy consumption of the whole device at the receiving end, and the estimated information may be estimated according to an application scenario (such as a mobile video application on the mobile terminal).

Alternatively, an ALF mode used by the encoder to perform encoding may be directly configured on the transmitting end according to actual application requirements. At this time, the transmitting end determines that the ALF mode configured by itself is the ALF mode used when the encoder performs encoding.

Here, the ALF mode may be any one shown in Table 2 in Embodiment 1, or In the first embodiment, the value of adaptive_loop_filter-enable_flag is 0, which means that ALF mode is not available for all fragments in the image.

S402, the encoder generates a parameter set corresponding to the ALF mode, where the parameter set includes ALF identification information of the ALF mode, and allocates a parameter set index number for identifying the parameter set for the generated parameter set;

5403. The encoder performs encoding according to the generated parameter set.

Specifically, the encoder writes the information of the parameter set into the code stream, writes the parameter set index number into the slice layer header information in the current coding unit, and determines the filtering used by the coding in the slice layer coding process. And writing parameters of the filter to the slice header information.

Here, the process of determining the filter used for encoding may be implemented by an existing method, such as a slice layer block/sub-region filter selection method based on rate distortion criterion, or may be described in the following fourth embodiment. Method implementation.

Here, the coding unit may be an image of an image currently being subjected to encoding processing or a slice of an image currently being encoded.

It should be noted that the ALF mode mentioned in this embodiment is based on the ALF mode mentioned in the first embodiment. In this embodiment, the code stream obtained by the encoder coding can be decoded by the decoding method provided in Embodiment 1. .

Embodiment 3

The encoder of this embodiment uses a plurality of ALF modes to encode the input video source to accommodate the tradeoff between the available source code rate, the codec's computational complexity, and the coding efficiency.

As shown in FIG. 5, the specific process of performing coding in multiple ALF modes may include the following steps:

S501. The encoder generates a plurality of parameter sets corresponding to each ALF mode, and allocates a parameter set index sequence number for each generated parameter set. Specifically, each of the ALF modes includes five ALF modes, which are respectively four ALF modes shown in Table 2 of the first embodiment, and corresponding to the value of the adaptive_loop_filter_enable_flag in the first embodiment. , means that ALF mode of ALF is not available for all shards in the image.

The ALF mode and the parameter set are - corresponding, or a multi-corresponding relationship, that is, corresponding to an ALF mode, one or more parameter sets may be generated. Specifically, for the ALF mode 3 of Table 2 in the first embodiment, a plurality of parameter sets corresponding to different ALF refresh periods can be generated.

Before the step 501, the foregoing ALF mode may be configured in advance on the sending end, that is, the ALF identification information of each ALF mode is pre-configured to the transmitting end, so that the encoder can generate corresponding multiple according to each ALF mode configured in the sending end. Parameter sets.

S502: The sending end determines, for the encoder, an initially used ALF mode, where the encoder determines an initially used parameter set according to the ALF mode determined by the sending end.

Here, the specific implementation process of the ALF mode used by the transmitting end to determine the initial use of the encoder is similar to the specific implementation process of the S401 in the second embodiment, and details are not described herein.

S503. The encoder encodes the current coding unit according to the currently used parameter set. Specifically, the encoder writes information about the currently used parameter set into the code stream, and the parameter set is used in the slice layer coding process. The parameter set index number is written into the slice header information in the current coding unit, and the filter used for encoding is determined, and the parameters of the filter are written to the slice header information.

For the first coding unit, the encoder writes information of the initially used parameter set to the code stream, and writes the parameter set index number of the parameter set to the first coding unit in the slice layer coding process. In the header information of each slice, a filter used for encoding is determined, and parameters of the filter are written to the slice header information.

5504, the encoder determines whether the input video sequence has been encoded, and if so, the encoder call ends; if not, continues to execute S505.

Here, the input video sequence can be determined by judging whether there is an uncoded coding unit. Whether the column is encoded or not.

Here, the coding unit may be an image. Since the image is divided into one or more slices before encoding, the encoding unit may also be a slice of the image.

5505. The sender dynamically monitors the change of available resources in the encoding process, and obtains available resource information in the encoding process.

5506, the encoder dynamically determines whether to adjust the ALF mode used in the encoding process according to the available resource information obtained by the sending end. If yes, proceed to S507; otherwise, return to S503; where the available resource is in the second embodiment S401. Determine the factors on which the ALF mode is based. Specifically, the encoder can determine the ALF mode under the current available resource condition by using the same method as the second embodiment S401. If the ALF mode currently determined by the encoder is different from the ALF mode currently being used, it is determined that the ALF mode needs to be adjusted. If the ALF mode currently determined by the encoder is the same as the ALF mode currently being used, it is determined that the ALF mode is not required. Adjust the ALF mode.

5507, the encoder verifies whether the currently determined parameter set corresponding to the ALF mode exists, and if yes, continues S509, otherwise, continues to S508;

For example, when the ALF mode is not used in the previous process, there may be no parameter set corresponding to the ALF mode.

Specifically, the encoder verifies whether the currently determined parameter set corresponding to the ALF mode exists, and specifically includes: verifying whether the parameter set includes the ALF identification information of the currently determined ALF mode, and if yes, the parameter set corresponding to the currently determined ALF mode. Exist; otherwise, the parameter set corresponding to the currently determined ALF mode does not exist.

For example, if the currently determined ALF mode is the ALF mode 3 shown in Table 2 in the first embodiment, it will be verified whether the parameter set contains a variable adaptive_loop_filter_pattern_idx with a value of 3, and whether the parameter set contains Adaptive—loop—filter_period and adaptive—loop—whether the value of filter_period is the same as the preset ALF refresh period; If yes, the parameter set corresponding to the currently determined ALF mode exists. Otherwise, there is no parameter set corresponding to the currently determined ALF mode.

S508. The encoder generates a parameter set corresponding to the ALF mode, and allocates a parameter set index sequence number to the generated parameter set.

For example, if the currently determined ALF mode is the ALF mode 3 shown in Table 2 in the first embodiment, a parameter set that satisfies the following conditions is generated: The included variable adaptive_loop_filter_pattern_idx has a value of 3 and contains variables Adaptive—loop—filter_period and the variable adaptive_loop—filter_period have the same value as the preset ALF refresh period.

S509. The encoder adjusts the currently used parameter set to the parameter set corresponding to the currently determined ALF mode, and jumps to step S503.

This embodiment can be applied to applications where the computing resources of the transmitting end are sufficient and the sensitivity to the complexity is low, but the computing capacity of the receiving end is limited and the complexity control is high. For example, mobile handheld TV, video download and playback of mobile terminals, and the like.

Embodiment 4

This embodiment describes in detail the implementation process of selecting a filter in the encoding process.

Specifically, as shown in FIG. 6, determining a specific process of the filter used in the encoding process may include the following steps:

S601. After determining the used ALF mode, the encoder groups all candidate ALFs according to the complexity and denoising performance of each candidate filter to obtain one or more candidate filter banks.

Specifically, the grouping process of the encoder to the candidate filter may include: according to the complexity The candidate filters are sorted in order from low to high, and each candidate filter bank corresponding to different complexity segments is obtained, and then each candidate filter in the candidate filter group is sequentially performed according to the denoising performance from high to low. Sort

Or, the candidate filters are sequentially sorted according to the denoising ability from high to low, and each candidate filter group corresponding to different denoising capability segments is obtained, and then in the candidate filter group according to the complexity from low to high. Each candidate filter is sorted.

The filter complexity includes: computational complexity (such as processing the number of basic data operations required for addition, multiplication, and shifting per pixel), storage complexity (such as processing the required storage space per pixel), and the like. Numerical calculation related factors.

The complexity and denoising performance of the filter can be evaluated based on the type of filter itself. For example, in general, high-order filters have better denoising ability than low-order filters; same-order filters, filters with high coefficient precision and high precision are better than those with low coefficient accuracy and precision. . High-order filters are more complex than low-order filters; for the same-order filters, filters with high coefficient accuracy and high precision are more complex than filters with low coefficient accuracy and precision.

The complexity and denoising performance of the filter can also be evaluated based on existing statistical data or performance statistics accumulated during the encoding process. For example, statistical analysis can be performed on the denoising ability or complexity of each candidate filter according to empirical data obtained from a large number of offline statistics; according to the reduction of image distortion before and after filtering processing in the encoding process and the number of filter coefficient encoding bit overhead, the filter The complexity or denoising performance is evaluated. The specific implementation process is a common technical tool in the field, and will not be described here.

S602. The encoder calculates an image importance parameter of the current coding unit.

Here, the coding unit may be an image or a slice of an image.

The image importance parameter is used to indicate the effect of each coding unit on the overall coding efficiency of the video sequence. For example, the importance of intra-coded pictures is usually higher than the inter-predictive coded picture; low time The level of bidirectional predictive coded pictures is of higher importance than the same type of pictures of high temporal levels.

The image importance parameter may be determined according to factors such as an image coding type, a parameter on the motion compensation prediction loop accuracy, an image activity degree, and the like. The image importance parameter may be a value directly calculated according to an algorithm, for example, an image activity parameter calculated using an average absolute error; or may be a score value obtained by evaluating an image according to a common rule, such as intra prediction coding. The image has the highest importance coefficient, and the inter-frame predictive coded image is of the second most important importance; for the same type of image, the image importance can be further distinguished according to the video content of the image, for example, for an image that is also intra-predictive coded, The image at the scene switch has an image importance that is higher than the image importance of the image at the non-scene switch.

The use of existing methods capable of reflecting the effect of the image on the overall coding efficiency can be used to quantify the importance of the image in this embodiment.

S603. The encoder selects a filter used for encoding according to the calculated image importance parameter and the obtained one or more candidate filter groups for the current coding unit.

Specifically, the encoder first determines a candidate filter bank of the current coding unit according to the calculated image importance parameter, and then selects a filter to be used for the current coding unit from the determined candidate filter group.

The process of selecting, by the encoder, the filter used for encoding the current coding unit according to the calculated image importance parameter and the obtained one or more candidate filter groups may include:

First, selecting a candidate filter bank from each candidate filter bank corresponding to different complexity segments according to available computing resources and storage resources, and then, according to the image importance parameter, from the selected candidate filter bank. Select a filter that matches the noise cancellation capability and use it as the filter for the current coding unit.

Or first selecting a candidate filter bank from each candidate filter bank corresponding to different complexity segments according to available computing resources and storage resources, and then, according to the image importance a parameter, selecting, from the selected candidate filter bank, a plurality of filters whose denoising ability is close to the required denoising capability, and then using the existing slice-based loop filter selection method based on the rate distortion criterion The filter used by the current coding unit is selected from the candidate filters.

Or first selecting a candidate filter bank from each candidate filter bank corresponding to different denoising capability segments according to the image importance parameter, and then, from the candidate filter according to available computing resources and storage resources. A filter with a matching complexity is selected in the group as the filter used by the current coding unit.

Or first selecting a candidate filter bank from each candidate filter bank corresponding to different denoising capability segments according to the image importance parameter, and then, from the candidate filter according to available computing resources and storage resources. Selecting a plurality of filters whose complexity is close to the required complexity requirement is selected as a candidate, and then using the existing slice-based loop filter selection method based on rate distortion criterion to select the current coding unit from the candidate filters. Filter.

Specifically, for an image with high image importance, a filter with relatively high denoising performance is selected, and for an image with low image importance, a filter with relatively low denoising performance can be selected. When there are relatively many available computing resources and storage resources, you can select a filter with higher complexity. When the computing resources and storage resources are relatively small, you can select a filter with lower complexity.

In particular, other complexity, denoising performance, image importance parameters, and performance metrics similar to those described in the present embodiment are used to group candidate filters and use low coding layer optimization tools (eg, rate-based distortion). The sliced layer loop filter selection method of the criterion) selecting the filter used from the candidate filter banks can be attributed to the present embodiment.

This embodiment can be applied to the case where both the transmitting end and the receiving end have sufficient computing resources and low sensitivity to complexity, but have certain requirements for real-time performance. For example, digital TV, interactive web TV

(IPTV), etc.

Although the ALF is taken as an example for description in this embodiment, the method for selecting a filter in this embodiment can also be used in other loop filtering applications, such as SAO. Embodiment 5

In this embodiment, a plurality of ALF modes and encoder complexity allocation methods are used in combination to encode to meet the tradeoff between the available source coding rate, the codec's computational complexity, and the coding efficiency. In this embodiment, a detailed description of the process of encoding the video image multiple times is provided.

Specifically, as shown in FIG. 7, the process of encoding the video image twice in this embodiment may include the following steps:

5701, identical to S501;

5702. The encoder performs the first encoding on the input video sequence according to the parameter set. Specifically, the specific implementation process of the first encoding of the video sequence by the encoder is as follows: Embodiment 2, Embodiment 3, Embodiment 4 As shown, it is not mentioned here.

Here, the first encoding process may be a complete encoding process, or may be a partial encoding process. For example, the first encoding process is usually a partial encoding process that does not include an output stream step.

5703, the encoder obtains information such as a source characteristic parameter and an encoder control auxiliary parameter in the first encoding process, and according to the obtained source characteristic parameter and the encoder control auxiliary parameter in the first encoding process, Determining the ALF mode used by each image or each video segment at the time of the second encoding;

Here, the encoder may further determine an encoding module parameter in the second encoding process according to the obtained source characteristic parameter, the encoder control auxiliary parameter, and the like in the first encoding process.

The encoding module parameters may include: a length of a group of pictures (GOP, Group Of Pictures) and an encoding type of each image in the GOP (such as intra prediction, inter-frame unidirectional prediction, inter-frame bidirectional prediction, etc.), Filter parameters further determined using the coding unit of ALF, as well as other encoder control parameters, such as quantization parameters and the like. The image coding type is related to the first loop filtering mode and the second loop filtering mode in the first embodiment.

The source characteristic parameter may include: image activity degree, scene switching image position, The degree of motion between images in each scene segment.

The encoder control assistance parameter may include: applicability of each coding prediction, block mode to the motion region of each image or video segment, different images or different video segments; encoder for each image or video segment Applicability, complexity, and coding efficiency of the content area, the degree of motion between images, or between video segments; each ALF mode for each image or each video segment content region, each prediction encoding type of image or each video segment Applicability, complexity, coding efficiency, etc.

The encoder control auxiliary parameter is mainly used to describe the adaptability of different coding tools (such as prediction, block mode, ALF mode, etc.) to video segments, images, and regions of different texture features in different characteristics of the input video sequence. It is thought that the second encoding process determines the encoding tools and parameters used for different texture feature regions in different video segments, images, and images, such as selecting a uni-directional inter prediction mode for the still scene portion and selecting a frame for the motion severe scene. Intra prediction mode; When using the third loop filtering mode, set a smaller refresh period value for the video segment with a fast scene change, and set a larger refresh period value for the video segment with slow scene change.

The encoder may divide the input video sequence into a plurality of video segments according to factors such as scene change, motion severity, and the like according to information such as a source characteristic parameter and an encoder control auxiliary parameter obtained in the first encoding process. Alternatively, the input video sequence can be divided into multiple images. The ALF mode of each image or each video segment is determined by the method of determining the ALF mode in the second embodiment or the third embodiment.

S704: The encoder re-assigns the parameter set index number to each parameter set corresponding to the ALF mode determined by S703.

Specifically, the encoder sorts the ALF modes according to the number of times of use of each ALF mode determined in the second encoding process after the first encoding process, according to the number of times, and according to the sorting result. And re-assigning the parameter set index number to the parameter set corresponding to each ALF mode. Specifically, the parameter set index number of the parameter set corresponding to the previous ALF mode is smaller, so that the number of bits of the parameter set index number can be saved, and the subsequent parameters are reduced. The time and resources required for the index index number to be written into the slice header information are conducive to saving resources.

S705. The encoder selects each parameter set after the parameter set index sequence number according to the S704, determines an ALF mode used by each image or each video segment when the second encoding is determined by S703, and selects in the first encoding process. The filter completes the second encoding of the input video sequence and outputs the code stream obtained by the second encoding.

Specifically, the encoder determines a parameter set used when encoding each image or each video segment according to the ALF mode used by each image or each video segment at the second encoding determined by S703, and each image or each video region is used. The parameter set used by the segment is written into the code stream, and the parameter set index number of the parameter set is written into the slice header information of each image or each video segment, and the filter selected in the first encoding process The relevant parameters are written into the slice header information of each image or each video segment.

Here, the encoder may encode the image or the slice of the image as a coding unit. The specific implementation process is similar to the coding process in the foregoing Embodiment 2 and/or Embodiment 3, and details are not described herein again.

This embodiment can be applied to applications in which both the transmitting end and the receiving end have sufficient computing resources and low sensitivity to complexity. For example, digital TV and IPTV have been used for on-demand playback, movie-on-demand, and high-definition digital home theater. This embodiment can also be applied to applications where the computing resources of the transmitting end are sufficient and the sensitivity to the complexity is low, but the computing power of the receiving end is limited and the complexity control is high. For example, video downloading and playback of a mobile terminal, movie on demand, live TV on-demand playback, and the like.

It should be noted that the ALF mode mentioned in this embodiment is the same as the ALF mode in the first embodiment. The code stream obtained by the encoder in this embodiment can be decoded by the decoding method provided in the first embodiment.

Embodiment 6

The embodiment provides an electronic device, and the electronic device may include a decoder provided by the present invention, and the decoder may decode the loop filter identification information by using a specific implementation process of the first embodiment. The electronic device may further include an encoder provided by the present invention, and the encoding is implemented by the methods provided in Embodiment 2 to Embodiment 5.

Specifically, the electronic device of the embodiment may be a related stream generating device and a receiving playing device in a video communication application, for example, a mobile phone, a computer, a server, a set top box, a portable mobile terminal, a digital television, a digital video camera, or the like.

It should be noted that the sending end is configured to send video information, and the receiving end is configured to receive video information. The above embodiments may be used arbitrarily in combination or separately.

The above is only the preferred embodiment of the present invention and is not intended to limit the scope of the present invention.

Claims

Claim

A loop filter coding method, the method comprising:

Setting a loop filtering mode for indicating a manner in which loop filtering is used in the encoding process; generating a parameter set according to the loop filtering mode;

Encoding is performed based on the generated parameter set.

The loop filter encoding method according to claim 1, wherein the method is: determining two or more loop filtering modes that can be used in an encoding process; filtering the loops a mode, respectively generating one or more parameter sets including loop filter identification information of the loop filtering mode;

3. The loop filter encoding method according to claim 2, wherein the two or more loop filtering modes comprise any two or more of the following loop filtering modes: Fragmentation uses a zero-loop filtering mode of loop filtering;

Indicates that when the image sequence number POC of the slice is an integer multiple of the loop filter refresh period, the third loop filter mode that uses loop filtering in the slice is allowed;

The loop filtering and encoding method according to claim 1, wherein the method is: determining a loop filtering mode used by a current encoding process;

The loop filter encoding method according to claim 4, wherein the loop filter The wave mode is any of the following loop filtering modes:

Representing a third loop filtering mode that allows loop filtering in the slice when the POC value of the slice is an integer multiple of the loop filter refresh period;

The loop filter encoding method according to claim 1, wherein the determining a loop filtering mode for indicating a manner in which the loop filtering is used in the encoding process is:

According to one or more of the following factors: specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, source coding rate, and receiving end processing capability, determined to indicate loop filtering in The loop filtering mode used in the encoding process.

The loop filter coding method according to any one of claims 1 to 6, wherein after the generating the parameter set including the loop filter identification information of the loop filter mode, the method further includes : Assign the parameter set index number used to identify the parameter set to the generated parameter set.

The loop filter encoding method according to claim 7, wherein the encoding according to the generated parameter set comprises: writing information of the parameter set into a code stream, in a slice layer encoding process The parameter set index number is written into the slice header information in the current coding unit, and the filter used for encoding is determined, and the parameters of the filter are written to the slice header information.

The loop filter coding method according to claim 8, wherein the determining the filter used by the coding comprises:

All candidate filters are divided according to the complexity and denoising performance of each candidate filter. Grouping, obtaining one or more candidate filter banks;

Calculating an image importance parameter of the current coding unit;

The loop filter coding method according to claim 1, 2, 3, 4, 5, 6, 8, or 9, wherein the method further comprises: verifying a loop determined by the current coding unit when performing coding The path filtering mode is the same as the loop filtering mode determined when encoding by the previous coding unit; if not, the currently used parameter set is adjusted; if yes, the currently used parameter is not adjusted.

The loop filter encoding method according to claim 10, wherein the adjusting the currently used parameter set comprises: verifying whether a parameter set corresponding to the loop filtering mode determined by the current coding unit is encoded, if Yes, directly adjusting the currently used parameter set to a parameter set corresponding to the loop filtering mode determined when the current coding unit performs coding; if not, generating a parameter set corresponding to the loop filtering mode determined when the current coding unit performs coding, and Adjusting the currently used parameter set to correspond to the loop filtering mode determined by the current coding unit for encoding

The loop filter encoding method according to claim 10, wherein the loop filtering mode determined when the current encoding unit performs encoding is verified, and the loop filtering mode determined when encoding by the previous encoding unit Before the method is the same, the method further includes: dynamically monitoring a change of available resources in the current encoding process, obtaining available resource information in the current encoding process; determining, according to the available resource information, loop filtering when encoding the current coding unit mode.

The method of claim 1, wherein the method further comprises:

According to the obtained source characteristic parameters in the last encoding process, and encoder control assistance a parameter, determining a loop filtering mode used by each image or each video segment in the current encoding process; according to the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process, respectively Each image or each video segment is encoded.

14. The loop filter encoding method according to claim 13, wherein:

The method further includes: reallocating a parameter set index number for a parameter set corresponding to a loop filtering mode used by each image or each video segment in the current encoding process;

The loop filter encoding method according to claim 14, wherein the parameter set index number is reassigned to a parameter set corresponding to a loop filtering mode used by each image or each video segment in the current encoding process, including :

And sorting the loop filtering modes according to how many times the loop filtering mode of each image or each video segment is used in the current encoding process, according to the number of uses;

According to the sorting order of the loop filtering modes, the parameter sets corresponding to the loop filtering modes are sequentially assigned index numbers of d to large parameter sets.

An encoder, the encoder comprising: a determining module, a generating module, and an encoding module; wherein

17. The encoder of claim 16 wherein:

The determining module is further configured to determine that the two or more types are capable of being used in the encoding process Loop filtering mode;

18. The encoder of claim 16 wherein:

The determining module is further configured to determine a loop filtering mode used by the current encoding process; the generating module is further configured to generate a parameter set that includes loop filtering identifier information of the loop filtering mode determined by the determining module;

The encoder according to claim 16, 17 or 18, wherein the determining module is further configured to determine, according to one or more of the following factors, that the loop filtering is used in the encoding process. The loop filtering mode used in the mode: specific application requirements, available computing resources and storage resources of the encoding process, video source characteristics, source coding rate, and receiving end processing capability.

20. The encoder of claim 16 wherein:

The generating module is further configured to allocate, to the generated parameter set, a parameter set index sequence number for identifying the parameter set;

The encoding module is further configured to write the information of the parameter set into the code stream, and write the parameter set index number into the slice layer header information in the current coding unit in the slice layer coding process, and determine the code. A filter is used to write parameters of the filter to the slice header information.

The encoder according to claim 16, wherein the encoding module is further configured to determine a filter used for encoding by using the following method:

All candidate filters are grouped according to the complexity and denoising performance of each candidate filter to obtain one or more candidate filter banks; Calculating an image importance parameter of the current coding unit;

The encoder according to claim 16, wherein the encoder further includes a verification module and an adjustment module, where

An adjustment module for adjusting a currently used parameter set;

23. The encoder of claim 22, wherein

The adjusting module is further configured to verify whether a parameter set corresponding to the loop filtering mode determined by the current coding unit is present, and if yes, directly adjusting the currently used parameter set to a loop determined by the current coding unit for encoding a parameter set corresponding to the filtering mode; if not, generating, by the generating module, a parameter set corresponding to the loop filtering mode determined by the current coding unit for encoding, and adjusting the currently used parameter set to be determined by the current coding unit for encoding The parameter set corresponding to the loop filtering mode.

24. An encoder according to claim 16, 17, 18, 20, 21, 22, or 23, characterized in that

The determining module is further configured to determine, according to the obtained source characteristic parameter in the last encoding process, and the encoder control auxiliary parameter, a loop filtering mode used by each image or each video segment in the current encoding process;

The encoding module is further configured to: respectively, according to the parameter set corresponding to the loop filtering mode used by each image or each video segment in the current encoding process determined by the determining module, respectively, for each image or each video segment Encode.

25. A loop filter decoding method, the method comprising: Decoding the loop filtering identification information of the loop filtering mode in the parameter set related code stream; decoding the loop filtering identification information of the loop filtering mode in the slice layer;

The loop filtering mode is used to indicate the manner in which loop filtering is used in the encoding process.

The loop filter decoding method according to claim 25, wherein the decoding of the loop filter identification information of the loop filter mode in the parameter set related code stream comprises:

Reading the bits in the parameter set related code stream, obtaining values of the syntax units for identifying the loop filter identification information, and assigning the values of the syntax units to the parameter set data structure Corresponding variable.

The loop filter decoding method according to claim 25, wherein the decoding of the loop filter identification information of the loop filtering mode in the slice layer comprises:

Obtaining values of the syntax units for identifying the loop filter identification information in the parameter set, and passing the values of the syntax unit to corresponding variables of the slice layer data structure.

The loop filter decoding method according to any one of claims 25 to 27, wherein the method further comprises: correlating a loop filter in the slice layer according to the loop filter identification information obtained by decoding The parameters are parsed.

A decoder, the decoder comprising: a first decoding unit and a second decoding unit, wherein

The decoder according to claim 29, wherein the first decoding unit is further configured to read bits in a parameter set related code stream, to obtain each of the loop filter identification information used to identify the loop filter identification information. a value of a syntax unit, and assigning values of the syntax units to the parameter set data Corresponding variables in the structure.

The decoder according to claim 29, wherein the second decoding unit is further configured to obtain values of the syntax units used to identify the loop filter identification information in the parameter set, and Passing the value of the syntax unit to the corresponding variable of the slice layer data structure.

The decoder according to any one of claims 29 to 31, wherein the second decoding unit is further configured to filter the loop in the slice layer according to the decoded loop filter identification information. The relevant parameters of the device are parsed.

An electronic device, comprising: an encoder and/or a decoder, wherein

The encoder includes a determining module, a generating module, and an encoding module, the determining module is configured to set a loop filtering mode for indicating a manner in which the loop filtering is used in the encoding process; and the generating module is configured to use the loop set according to the determining module a filtering mode, generating a parameter set; the encoding module is configured to perform encoding according to the parameter set generated by the generating module;