Video decoding method and apparatus, electronic device, and storage medium

Abstract

The application discloses a video decoding method and device, electronic equipment and a storage medium. The method comprises the following steps: obtaining a cumulative probability table and a count value of a video coding format from cumulative probability distribution information stored in a memory and storing the cumulative probability table and the count value into a random access memory, wherein the cumulative probability distribution information comprises a storage mode and the cumulative probability table and the count value arranged together according to the storage mode; decoding the coding format of a to-be-decoded region of the video according to the cumulative probability table in the random access memory, and adding one to the count value corresponding to the cumulative probability table; if a next syntax is different from a currently decoded syntax during the decoding process, writing back the cumulative probability table of the currently decoded syntax to the random access memory; and when the decoding of the coding format of the to-be-decoded region of the video is completed, writing the cumulative probability table written back to the random access memory into the memory, and clearing the count value corresponding to all the cumulative probability tables to update the cumulative probability distribution information.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of video decoding, in particular to a video decoding method and device, an electronic device and a storage medium. BACKGROUND

[0002] Context-based adaptive binary arithmetic coding (cabac) of video coding format AV1 needs to use a cumulative probability table cdf (15 bits), but the cdf needs to be stored in a random access memory in advance before decoding, and since the number of elements of different syntaxes is different, it brings great inconvenience to the storage of the cdf.

[0003] If each group of cdf of each syntax is stored in a unit, then the resource consumption is large.

[0004] If all the syntaxes are seamlessly spliced and stored, then it will be troublesome to calculate the storage address, a multiplier needs to be used to calculate the address, and most of the time, the address will cross the unit.

[0005] In addition, the count value (6 bits) matched with the cdf is independently stored, and an additional set of read-write control is also needed. SUMMARY

[0006] The technical problem to be solved by the present application is to provide a video decoding method and device, an electronic device and a storage medium, which do not need to additionally set a read-write control unit for the cumulative probability table, thereby saving the resources of the random access memory.

[0007] In one aspect of the present application, a video decoding method is provided. The method comprises the steps of: obtaining a cumulative probability table and a count value of a video coding format from cumulative probability distribution information stored in a memory to store in a random access memory, the cumulative probability distribution information comprising a storage mode and the cumulative probability table and the count value arranged together according to the storage mode; decoding a video coding format of a to-be-decoded region of the video according to the cumulative probability table in the random access memory, and adding one to the count value corresponding to the cumulative probability table; if the next syntax is different from the currently decoded syntax during the decoding process, writing the cumulative probability table of the currently decoded syntax back to the random access memory; and when the decoding of the video coding format of the to-be-decoded region of the video ends, writing all the cumulative probability tables written back to the random access memory into the memory, and clearing the count values corresponding to all the cumulative probability tables to update the cumulative probability distribution information.

[0008] In another aspect of the present application, a video decoding device is provided. The device comprises a memory configured to store a video; and one or more processors configured to perform the above-mentioned video decoding method.

[0009] In yet another aspect of the present application, an electronic device is provided. The electronic device includes a memory configured to store a processor-executable program; and a processor configured to execute the program such that the electronic device performs the above-mentioned video decoding method.

[0010] In still another aspect of the present application, a computer readable medium is provided. The medium has stored thereon a computer program which is executed by a processor to implement the above-mentioned video decoding method.

[0011] According to the present application, the cumulative probability table and its count value of the video coding format are obtained from the cumulative probability distribution information stored in the memory and stored in the random access memory. Since the cumulative probability distribution information includes the storage mode and the cumulative probability table and the count value arranged together according to the storage mode, the cumulative probability table and its count value can be stored together, the read-write control unit for the count value is reduced, the random access memory used is combined into one, the resource of the random access memory is saved, and the data position can be quickly located according to the corresponding storage. The cumulative probability table in the random access memory is used to decode the coding format of the to-be-decoded region of the video, and the count value corresponding to the cumulative probability table is incremented. In the decoding process, if the next syntax is different from the current decoded syntax, the cumulative probability table of the current decoded syntax is written back to the random access memory. In this way, the cumulative probability table used by each syntax can be stored in the random access memory. When the decoding is completed, all the used cumulative probability tables are written back to the memory, and the count value is cleared, so as to be directly used in the subsequent region decoding. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 Flow chart of the video decoding method according to the embodiment of the present application;

[0013] Figure 2 Module chart of the video decoding method according to the embodiment of the present application;

[0014] Figure 3 Storage schematic diagram of the cumulative probability table and its count value in the cumulative probability distribution information according to the embodiment of the present application;

[0015] Figure 4 Storage schematic diagram of the cumulative probability table and its count value in the cumulative probability distribution information according to the embodiment of the present application;

[0016] Figure 5 Schematic diagram of the video decoding device according to the embodiment of the present application. DETAILED DESCRIPTION

[0017] The technical content, the achieved purposes and effects of the present application will be described in detail below in combination with the embodiments and the accompanying drawings.

[0018] In the prior art, the cumulative probability table and the count value associated therewith are inconvenient to store due to the different number of elements of each syntax.

[0019] To solve at least the above technical problems, the present disclosure provides a video decoding method. According to the present disclosure, the cumulative probability table and the count value of the video coding format are stored together according to the corresponding storage mode, and the used cumulative probability table and the count value thereof are stored in the random access memory during the decoding process. After decoding the region to be decoded each time, the used cumulative probability table is written back to the memory. In this way, the embodiments according to the present disclosure can reduce the additional read-write control unit of the count value, and facilitate the direct use of the decoded used cumulative probability table during subsequent region decoding, thereby improving the decoding efficiency.

[0020] Hereinafter, the technical solutions according to the present disclosure will be described with reference to specific embodiments and in conjunction with the accompanying drawings.

[0021] Figure 1 is a flowchart showing a video decoding method 100 according to an embodiment of the present disclosure. Referring to Figure 1 , the method 100 includes the following steps 102 to 108.

[0022] At step 102, the cumulative probability table and the count value of the video coding format are obtained from the cumulative probability distribution information stored in the memory to be stored in the random access memory, and the cumulative probability distribution information includes the storage mode and the cumulative probability table and the count value arranged together according to the storage mode.

[0023] In some embodiments, the storage mode is determined according to the highest four bits in each unit of the cumulative probability distribution information. The storage mode includes the number of groups of the cumulative probability table and the count value that can be stored, and the maximum number of elements stored in each group of the cumulative probability table. The number of groups of the cumulative probability table and the count value are stored together. According to the storage mode of the region to be decoded, the unit corresponding to the storage mode of the region to be decoded is selected, and the cumulative probability table and the count value of the region to be decoded are stored.

[0024] In this way, the storage of the cumulative probability table and the count value is performed according to the storage mode, and through reasonable data storage, the space utilization rate is high, and the memory resources are saved.

[0025] In some embodiments, the position of the cumulative probability table and the count value is determined according to the highest four bits in each unit. In this way, the positioning efficiency of the data is improved, and compared with the existing seamless splicing storage method, the embodiments do not need to use a multiplier to calculate the storage address.

[0026] In some embodiments, the cumulative probability table and its corresponding count value are combined as a first data combination, which is continuously stored in one unit; or all cumulative probability tables and all count values that need to be stored in one unit are stored separately. In this way, the space utilization of storage is improved, and the search is facilitated.

[0027] In some embodiments, the number of cumulative probability tables and the number of elements of a cumulative probability table of the current syntax of the region to be decoded are obtained, and the number of elements is reduced by one and combined with the number of cumulative probability tables to match the storage mode in the random access memory, and the cumulative probability table and its count value are stored. In this way, since the last element in the cumulative probability table is the cumulative total probability of all elements, this value is the same, so this data does not need to be stored, thereby reducing the amount of data stored, and multiple syntaxes can use the same storage mode as long as the number of elements is the same, avoiding the case that each set of cumulative probability tables is stored in one unit.

[0028] In step 104, the encoding format of the region to be decoded of the video is decoded according to the cumulative probability table in the random access memory, and the count value corresponding to the cumulative probability table is incremented by one.

[0029] In step 106, if the next syntax is different from the current decoded syntax during the decoding process, the cumulative probability table of the current decoded syntax is written back to the random access memory.

[0030] In step 108, when the decoding of the encoding format of the region to be decoded of the video is completed, all cumulative probability tables written back to the random access memory are written into the memory, and the count values corresponding to all the cumulative probability tables are cleared to zero to update the cumulative probability distribution information.

[0031] In some embodiments, the updated cumulative probability table is obtained from the memory, and the next region to be decoded of the video is decoded according to the updated cumulative probability table, until all regions to be decoded in a frame of the video are decoded; the cumulative probability table is updated according to the decoding of a selected region to be decoded in all regions to be decoded in a frame of the video, and the decoding of the next frame is performed.

[0032] In this way, when each frame is updated to the memory, the corresponding count value can be cleared to zero according to the storage mode, so that it can be directly used in the subsequent frame.

[0033] In the following, the application scenarios of the video decoding method and device according to the embodiments of the present application will be described by examples.

[0034] Figure 2 Fig. 200 is a module diagram showing the video decoding method according to the embodiments of the present application.

[0035] First, the information of the cumulative probability table is loaded from the double rate memory (DDR) and stored in the random access memory.

[0036] Second, each frame of the video is divided into multiple regions, and the corresponding cumulative probability table is read from the random access memory for the context-based adaptive binary arithmetic coding (AV1_CABAC_DEC) of the video coding format AV1 during the decoding process.

[0037] If the next syntax is different from the current syntax, the cumulative probability table of the current syntax is written back to the random access memory.

[0038] After the decoding of the current region is completed, the cumulative probability table of the current region needs to be updated, and all the cumulative probability tables in the random access memory are written back to the DDR memory, and the corresponding count values of the cumulative probability table are cleared at the same time.

[0039] The region to be decoded is a rectangular region, i.e., a frame of image is divided into multiple rectangular regions, each region is independently encoded and decoded. For the application of the cumulative probability table, the rectangular regions are independent of each other, and the initial value of the cumulative probability table used by all the rectangular regions of a frame is the same, while the internal rectangular region is updated according to the decoding condition of each rectangular region.

[0040] The cumulative probability table of which rectangular region needs to be stored for the next frame is specified in the code stream in advance, and when the corresponding rectangular region is decoded, the corresponding cumulative probability table is written to the DDR memory after decoding.

[0041] Since the decoding count of the syntax needs to be restarted at the beginning of each rectangular region, the count value corresponding to the cumulative probability table can be cleared at the beginning of the decoding of the rectangular region, or the count value can be cleared after the decoding is completed. The selection of clearing the count value corresponding to the cumulative probability table at the time of writing back can be determined according to the overall hardware design.

[0042] Figure 3 FIG. 1 is a schematic diagram showing the storage of the cumulative probability table and its count value in the cumulative probability distribution information according to an embodiment of the present application.

[0043] cdf is the cumulative distribution function, which is the cumulative probability table, the integral of the probability density function, which can completely describe the probability distribution of a random variable. Count is used to calculate the number of times a variable or symbol is decoded, and the decoding times are at most 32 times. After a variable or symbol is decoded, if its corresponding count is less than 32, the count will be increased by 1, and the new cdf value is obtained by updating the formula. The function of count is to update cdf, and a variable needs a count to count.

[0044] This embodiment is to arrange the cdf and count of AV1 together, each element of cdf is 15 bits, the cdf with the most elements has a group of 15, corresponding to a count (6 bits), plus the highest N bits, which is used to represent the position of count. Here, N can be divided into multiple grades, for example, when cdf_reg

[255] is 1, the unit has 5 counts, and each syntax has at most 3 elements.

[0045] When cdf_reg

[255] is 0, the value of cdf_reg[254:252] needs to be judged, and currently there are five kinds in total, and the remaining ones are for future expansion.

[0046] Referring to Figure 3 , Figure 3 “b” in “bit”, cdf_reg[255:0] corresponds to the content of each unit, the highest 4 bits cdf_reg[255:252] are storage mode cdf_count_mode, which is used to mark the position of count, Figure 3 The slash part in the above formula is the reserve space, that is, it is not used temporarily.

[0047] When cdf_count_mode==0, it means that the unit has only one count, each syntax has at most 15 elements, and the position of count is cdf_reg[15*15+5:15*15].

[0048] When cdf_count_mode==1, it means that the unit has two counts, each syntax has at most 8 elements, and the positions of count are respectively:

[0049] cdf_reg[15*16+6+5:15*16+6];

[0050] cdf_reg[15*8+5:15*8].

[0051] When cdf_count_mode == 2, it means that there are 4 counts in this unit, and each syntax has 4 or 3 elements, and the positions of the counts are as follows:

[0052] cdf_reg[15*15+6*3+5:15*15+6*3];

[0053] cdf_reg[15*12+6*2+5:15*12+6*2];

[0054] cdf_reg[15*8+6*1+5:15*8+6*1];

[0055] cdf_reg[15*4+5:15*4].

[0056] When cdf_count_mode == 3, it means that there are 12 counts in this unit, and each syntax has at most 1 element, and the positions of the counts are as follows:

[0057] cdf_reg[15*12+6*11+5:15*12+6*11];

[0058] cdf_reg[15*11+6*10+5:15*11+6*10];

[0059] cdf_reg[15*10+6*9+5:15*10+6*9];

[0060] cdf_reg[15*9+6*8+5:15*9+6*8];

[0061] cdf_reg[15*8+6*7+5:15*8+6*7];

[0062] cdf_reg[15*7+6*6+5:15*7+6*6];

[0063] cdf_reg[15*6+6*5+5:15*6+6*5];

[0064] cdf_reg[15*5+6*4+5:15*5+6*4];

[0065] cdf_reg[15*4+6*3+5:15*4+6*3];

[0066] cdf_reg[15*3+6*2+5:15*3+6*2];

[0067] cdf_reg[15*2+6*1+5:15*2+6*1];

[0068] cdf_reg[15*1+5:15*1].

[0069] If a syntax contains multiple probability tables, it will be stored as multiple syntaxes.

[0070] For example, Default_Partition_W8_Cdf has 4 groups, and each group has 4 elements. Since the last element is fixed at 32768, only 3 elements need to be stored, and the count stored in each group does not need to be stored. Therefore, Default_Partition_W8_Cdf can be stored using the storage method of cdf_count_mode == 2.

[0071] Several CDFs and counts are arranged together, and the storage location of the counts is different because the values ​​of most AV1 syntax variables are different. For example, the range of Default_Segment_Id_Cdf is 0 to 7, indicating that the decoded value of the segment_id syntax is in the range of 0 to 7. Every time segment_id is decoded, count is incremented by 1 until the count of segment_id equals 32, at which point count remains at 32. Assuming the currently decoded segment_id is 4, the CDF values ​​of elements 4 to 7 will be incremented by a certain probability value, while the CDF values ​​of elements 0 to 3 will be decremented by the corresponding probability value.

[0072] Figure 4 This is a schematic diagram illustrating the storage of the cumulative probability table and its count values ​​in the cumulative probability distribution information of this invention.

[0073] In some embodiments, refer to Figure 4 , Figure 4 The "b" in the code stands for "bit," and the syntax is stored compactly, as is the corresponding count. The number of elements stored here actually represents the range of the syntax. For example, if the syntax range of segment_id is 0 to 7, it will be stored with cdf_count_mode == 1.

[0074] It can be seen that as long as the number of elements of multiple syntaxes is the same, the same cdf_count_mode can be used. For example, the cdf of mv_bit has 10 probability tables, each of which has 2 elements, indicating that the value of mv_bit has two, 0 and 1, the first column is the probability of 0 appearing, the second column is the total probability of 0 and 1 appearing, which is 32768, and the value of the second column is fixed, so it does not need to be stored, and the third column is the initial value of count. For the cdf of mv_bit, 10 groups of cdf, a total of 10 elements, are stored in the storage mode corresponding to cdf_count_mode == 3, and the maximum number of storage modes is 12, so the lower 10 positions are stored, and the upper 2 positions are idle and not stored.

[0075] For example, the cdf of cfl_sign has only one probability table, which has 8 elements, indicating that the value of cfl_sign has 8, 0 to 7, the first column is the probability of 0 appearing, the second column is the total probability of 0 and 1 appearing, and so on, and the value of the eighth column is fixed, so it does not need to be stored, and the ninth column is the initial value of count. For the cdf of cfl_sign, 7 elements are stored in the storage mode corresponding to cfl_count_mode == 1, and another cdf with no more than 8 elements is found to be stored.

[0076] According to another aspect of the present application, Figure 5 is a schematic diagram showing a video decoding apparatus 500 according to an embodiment of the present application. Referring to Figure 5 The video decoding apparatus 500 includes a memory 502 and a processor 504.

[0077] The memory 502 is configured to store videos, etc.

[0078] The processor 504 is configured to obtain the cumulative probability table and the count value of the video coding format from the cumulative probability distribution information stored in the memory, and store them in the random access memory. The cumulative probability distribution information includes a storage mode and the cumulative probability table and the count value arranged together according to the storage mode. The processor 504 is configured to decode the coding format of the to-be-decoded region of the video according to the cumulative probability table in the random access memory, and add one to the count value corresponding to the cumulative probability table. The processor 504 is configured to write back the cumulative probability table of the currently decoded syntax to the random access memory if the next syntax is different from the currently decoded syntax during the decoding process. In addition, the processor 504 is configured to write all the cumulative probability tables written back to the random access memory into the memory and clear the count values corresponding to all the cumulative probability tables when the decoding of the coding format of the to-be-decoded region of the video is completed, so as to update the cumulative probability distribution information.

[0079] In some embodiments, the processor 504 is configured to perform the steps of the video decoding method as described above, which will not be repeated here.

[0080] In some embodiments, the video decoding apparatus 500 is a chip.

[0081] According to another aspect of the present application, a computer readable medium is provided. The computer readable medium stores a computer program which is executed by a processor to implement the video decoding method as described above.

[0082] According to another aspect of the present application, an electronic device is provided. The electronic device includes a memory and a processor. The memory stores a processor-executable program. The processor executes the program to cause the electronic device to implement the video decoding method as described above.

[0083] In summary, the video decoding method and device, electronic equipment and storage medium provided by the present application store the cumulative probability table and the count value of the video coding format from the memory into the random access memory according to the corresponding storage mode, can store the cumulative probability table and the count value together, reduce the read-write control unit for the count value, use the random access memory as one, save the resources of the random access memory, and can improve the fast positioning of the data position according to the corresponding storage. The storage method in the present application belongs to semi-tight ordering, inserts the count value information into the corresponding cumulative probability table, improves the utilization rate of the storage space, and the positioning efficiency of the cumulative probability table and the count value is high, without the need of a multiplier to calculate the address, and the update of the cumulative probability table and the count value is synchronous, and the read-write control unit only needs a set. At the same time, the storage method of the cumulative probability table and the count value is 256 bits, which can correspond to the bus bit width of 128 bits.

[0084] The cumulative probability table in the random access memory is used to decode the coding format of the to-be-decoded region of the video, and the count value corresponding to the cumulative probability table is added by one. In the decoding process, if the next syntax is different from the current decoded syntax, the cumulative probability table of the current decoded syntax is written back to the random access memory. In this way, the cumulative probability table used by each syntax in the random access memory can be stored. When the decoding is completed, all the used cumulative probability tables are written back to the memory, and the count value is cleared, so as to be directly used in subsequent region and subsequent frame decoding.

[0085] The above is only an embodiment of the present application, and does not limit the patent range of the present application. Any equivalent transformation or direct or indirect application in the related technical field according to the content of the present application specification and drawings is also included in the patent protection range of the present application.

Claims

1. A method of video decoding, the method comprising: The method comprises the following steps: acquiring a cumulative probability table and a count value of a video coding format from cumulative probability distribution information stored in a memory to store in a random access memory, wherein the cumulative probability distribution information comprises a storage mode and the cumulative probability table and the count value arranged together according to the storage mode; decoding a coding format of a to-be-decoded region of a video according to the cumulative probability table in the random access memory, and adding one to the count value corresponding to the cumulative probability table; if a next syntax is different from a currently decoded syntax in the decoding process, writing back the cumulative probability table of the currently decoded syntax to the random access memory; when the decoding of the coding format of the to-be-decoded region of the video is completed, writing all the cumulative probability tables written back to the random access memory into the memory, and clearing the count values corresponding to all the cumulative probability tables to update the cumulative probability distribution information, wherein the step of acquiring the cumulative probability table and the count value of the video coding format from the cumulative probability distribution information stored in the memory to store in the random access memory comprises the following steps: determining a storage mode according to a preset mapping relationship between the value of the highest four bits in each unit of the cumulative probability distribution information and the storage mode, wherein the storage mode comprises the number of groups of the cumulative probability table and the count value that can be stored, and the maximum number of elements of each group of the cumulative probability table, and the groups of the cumulative probability table and the count value are stored in combination; selecting a unit corresponding to the storage mode of the to-be-decoded region according to the storage mode of the to-be-decoded region, and storing the cumulative probability table and the count value of the to-be-decoded region. The step of storing the groups of the cumulative probability table and the count value in combination comprises the following steps:

2. The video decoding method of claim 1, wherein, combining the cumulative probability table and the corresponding count value as a first data group, and storing the first data group continuously in one unit; or storing all the cumulative probability tables and all the count values that need to be stored in one unit separately. The step of determining the storage mode according to the highest four bits in each unit of the cumulative probability distribution information comprises the following steps:

3. The video decoding method of claim 1, wherein, judging whether the highest bit in each unit is 1; if yes, the storage mode of the unit is to store 5 groups of the cumulative probability table and the count value, and each group of the cumulative probability table stores a maximum of 3 elements; otherwise, determining the number of groups of the cumulative probability table and the count value that can be stored, and the maximum number of elements of each group of the cumulative probability table according to the value of the highest four bits in each unit. The step of determining the number of groups of the cumulative probability table and the count value that can be stored, and the maximum number of elements of each group of the cumulative probability table according to the value of the highest four bits in each unit comprises the following steps:

4. The video decoding method of claim 3, wherein, if the value of the highest four bits in each unit is 0, the storage mode is to store 1 group of the cumulative probability table and the count value, and each group of the cumulative probability table stores a maximum of 15 elements; if the value of the highest four bits in each unit is 1, the storage mode is to store 2 groups of the cumulative probability table and the count value, and each group of the cumulative probability table stores a maximum of 8 elements; ​ If the value of the highest four bits in each unit is 2, the storage mode is to store 4 sets of cumulative probability tables and their count values, each set of cumulative probability table stores at most 4 elements; If the value of the highest four bits in each unit is 3, the storage mode is to store 12 sets of cumulative probability tables and their count values, each set of cumulative probability table stores at most 1 element.

5. The video decoding method of claim 1, wherein, Determining the storage mode according to the highest four bits in each unit of the cumulative probability distribution information further comprises: Determining the position of the cumulative probability table and its count value according to the highest four bits in each unit.

6. The video decoding method of claim 1, wherein, According to the storage mode of the to-be-decoded region, selecting a unit corresponding to the storage mode of the to-be-decoded region, and storing the cumulative probability table and its count value of the to-be-decoded region comprises: Obtaining the number of cumulative probability tables and the number of elements of the cumulative probability table of the current syntax of the to-be-decoded region, matching the storage mode in the random access memory by combining the number of elements minus one with the number of cumulative probability tables, and storing the cumulative probability table and its count value.

7. The video decoding method of claim 1, wherein, After the count values corresponding to all the cumulative probability tables are cleared, further comprising: Obtaining the updated cumulative probability table from the memory, and decoding the next to-be-decoded region of the video according to the updated cumulative probability table until all the to-be-decoded regions in a frame of the video are decoded.

8. The video decoding method of claim 7, wherein, After all the to-be-decoded regions in a frame of the video are decoded, further comprising: Updating the cumulative probability table according to the decoding situation of a to-be-decoded region selected from all the to-be-decoded regions in a frame of the video, and decoding the next frame. 9.A video decoding apparatus, comprising: a memory configured to store a video; and one or more processors configured to perform the video decoding method according to any one of claims 1 to 8.

10. A computer readable medium having stored thereon a computer program, characterized in that, The computer program is executed by a processor to implement the video decoding method according to any one of claims 1 to 8. 11.An electronic device, comprising: a memory configured to store a processor-executable program; and a processor configured to execute the program so that the electronic device performs the video decoding method according to any one of claims 1 to 8.

Images