WO2022247735A1 - 数据处理的方法和系统 - Google Patents
数据处理的方法和系统 Download PDFInfo
- Publication number
- WO2022247735A1 WO2022247735A1 PCT/CN2022/094004 CN2022094004W WO2022247735A1 WO 2022247735 A1 WO2022247735 A1 WO 2022247735A1 CN 2022094004 W CN2022094004 W CN 2022094004W WO 2022247735 A1 WO2022247735 A1 WO 2022247735A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- boundary
- frame
- data
- amplitude
- frequency
- Prior art date
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 73
- 238000013144 data compression Methods 0.000 claims abstract description 148
- 230000006837 decompression Effects 0.000 claims abstract description 107
- 238000000034 method Methods 0.000 claims abstract description 85
- 238000007906 compression Methods 0.000 claims description 72
- 230000006835 compression Effects 0.000 claims description 68
- 238000012545 processing Methods 0.000 claims description 60
- 238000003860 storage Methods 0.000 claims description 52
- 230000008569 process Effects 0.000 claims description 38
- 230000000694 effects Effects 0.000 claims description 28
- 230000002829 reductive effect Effects 0.000 claims description 24
- 230000002238 attenuated effect Effects 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 19
- 230000000717 retained effect Effects 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 238000003825 pressing Methods 0.000 claims 2
- 230000006870 function Effects 0.000 description 234
- 230000005540 biological transmission Effects 0.000 description 37
- 238000001228 spectrum Methods 0.000 description 25
- 238000004891 communication Methods 0.000 description 14
- 238000012546 transfer Methods 0.000 description 13
- 230000008859 change Effects 0.000 description 10
- 230000007704 transition Effects 0.000 description 10
- 230000001965 increasing effect Effects 0.000 description 8
- 230000015572 biosynthetic process Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 238000003786 synthesis reaction Methods 0.000 description 7
- 230000003287 optical effect Effects 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000013500 data storage Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000010355 oscillation Effects 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 238000004590 computer program Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005562 fading Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000004020 conductor Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000036961 partial effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/117—Filters, e.g. for pre-processing or post-processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/172—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a picture, frame or field
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/119—Adaptive subdivision aspects, e.g. subdivision of a picture into rectangular or non-rectangular coding blocks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
- H04N19/147—Data rate or code amount at the encoder output according to rate distortion criteria
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/18—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/80—Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation
Definitions
- This specification relates to the field of data processing, and in particular to a data processing method and system.
- video and image compression techniques have been used more and more in the past few decades. Videos often contain a huge amount of information. From traditional radio, film and television to a large number of monitoring and Internet applications, compressed images, videos and images are taking up more and more network and storage resources. This makes it take up a lot of network resources if the original data of a piece of video is transmitted from one terminal to another terminal through the network. This makes it difficult to achieve smooth transmission of pictures in the case of some real-time video transmissions. Therefore, before video data is transmitted, compression processing must first be performed at the data compression device to facilitate transmission. After the compressed video is transmitted to the data decompression device via the transmission medium, the data decompression device then decompresses the video to at least partially restore the video image.
- the main video compression standards in the prior art are the H.264 and H.265 standards.
- the encoder Before transmission, usually according to the H.264 and H.265 standards, the encoder is used to compress the video as a whole, and after transmission, the video is decompressed through the decoder according to the H.264 and H.265 standards.
- the above-mentioned overall video compression processing method is still unsatisfactory in terms of the balance between the amount of computation and the clarity of the decompressed video. This is because, when the H.264 and H.265 standards process the original video, various complex algorithms are used to generate the predicted frame of the original frame, and then the residual between the original frame and the predicted frame is recorded .
- a common method is to reduce the high-frequency information in the original frame image by filtering the original frame. It can be seen from the Fourier transform that the frequency information of the boundary part of the object in the picture is often relatively rich, and the high-frequency components of the boundary part are usually larger than the high-frequency components of other gentle regions. Therefore, although the frame image with reduced high-frequency information becomes blurred visually (that is, the definition of the image is reduced), it can make the residual error between the predicted frame and the filtered original frame smaller. In this way, the calculation amount required for video encoding and the encoded data stream are greatly reduced.
- the technique of frame prediction is very complex and consumes a lot of computing resources.
- the video codec system every time the coding efficiency is increased by 30% to 40% on average, the amount of calculation will increase by about 10 times.
- the clarity of the transmitted data is reduced after decompression, and there are often various noises, such as block effects or ringing effects.
- the block effect refers to that in image processing, a block-based Fourier transform causes a discontinuity phenomenon to appear at an image boundary.
- the ringing effect refers to that in image processing, when performing spectrum adjustment processing on an image, if the selected spectrum adjustment function has a faster change in value (that is, there is a region where the derivative changes sharply), it will make the output
- the image produces grayscale oscillations where the grayscale changes sharply, just like the air vibration produced by a clock being struck.
- the noise mostly appears at the image boundary. If an output image has strong noise, it cannot meet people's increasing requirements for data clarity. Therefore, how to further improve the compression efficiency of data, improve the clarity of decompressed data, and eliminate noise at the same time has always been the goal pursued in the field of data compression and decompression technology.
- the data processing method and system can divide the initial frame in the initial video data into multiple units, and obtain the amplitude of the intermediate frequency to high frequency region of each unit, and use different boundary adjustment coefficients Adjusting the amplitude of the intermediate frequency to the high frequency region in each unit, so as to reduce the amplitude of the initial frame in the intermediate frequency to the high frequency region.
- the amplitude of the mid-frequency to high-frequency region in the current unit is large, it means that the current unit contains a strong boundary, then use the boundary adjustment coefficient less than 1 and greater than 0 to adjust the amplitude of the mid-frequency to high-frequency region of the current unit , to reduce the amplitude of the mid-frequency to high-frequency region of the current unit, thereby reducing the signal strength in the mid-frequency to high-frequency region of the current unit, thereby reducing the amount of data information, and improving the efficiency of data compression when making predictions and calculating residuals efficiency.
- the amplitude of the mid-frequency to high-frequency region in the current unit is small, it means that the current unit contains a weak boundary, then use a boundary adjustment coefficient greater than 1 to adjust the amplitude of the mid-frequency to high-frequency region of the current unit to enhance The amplitude of the mid-frequency to high-frequency region of the current unit to avoid the loss of weak boundaries in the current unit during data compression (prediction and residual) to avoid loss of details.
- the data processing method and system can enhance the data information volume of weak boundaries while improving the efficiency of data compression, so as to avoid loss of details during the data compression process, that is, reduce data distortion while improving data compression efficiency.
- the method and system can use the unit during data compression as the data decompression unit, and use the boundary compensation coefficient corresponding to the boundary adjustment coefficient to adjust the amplitude of the mid-frequency to high-frequency region for each unit Boundary compensation to compensate for the reduced amplitude of the mid-to-high frequency region during data compression resulting in decompressed frames.
- the boundary compensation corresponds to the boundary adjustment, and there is a corresponding relationship between the boundary compensation coefficient and the boundary adjustment coefficient.
- the boundary compensation can restore the compressed data after the boundary adjustment to a resolution of the original frame that is even higher than that of the original frame. That is to say, without significantly increasing the calculation amount of codec, the decoding end can at least restore the data in the important frequency of the decompressed data to the definition of the original frame, and even obtain the definition beyond the original frame.
- the boundary adjustment coefficients of the initial frame in the boundary adjustment process are all greater than 0, the information in the compressed frame is not missing, so the boundary adjustment can be designed according to the relationship between the boundary adjustment coefficient and the boundary compensation coefficient and their respective characteristics Coefficients and boundary compensation coefficients to restore the information in the compressed frame.
- the method and system can significantly improve data compression efficiency, improve data transmission efficiency, reduce data loss, avoid detail loss, eliminate noise, and improve the clarity of decompressed data.
- this specification provides a data processing method, including: selecting an initial frame in the initial data, the initial frame including initial data with a preset number of bytes; and performing data compression on the initial frame , to obtain a compressed frame, wherein the data compression includes performing boundary adjustment on the compressed frame, and the compressed frame includes the initial frame and any frame before the initial frame becomes the compressed frame during the data compression process A data state, wherein the boundary adjustment includes adjusting the amplitude of each unit in the plurality of units in the compressed frame using its corresponding boundary adjustment coefficient in the mid-frequency to high-frequency region, so as to reduce the The amplitude of the compression frame in the middle frequency to the high frequency region, the boundary adjustment coefficient is greater than 0.
- the adjusting the boundary of the compression frame includes: dividing the compression frame into the plurality of units based on a preset unit size; and using the corresponding unit for each unit The above-mentioned boundary adjustment coefficient is used to adjust its amplitude in the mid-frequency to high-frequency region.
- the adjusting the amplitude of each unit in the mid-frequency to high-frequency region by using the boundary adjustment coefficient corresponding thereto includes: for each unit: from a preset encoding Select a function in the function group as the encoding function, adjust it through the encoding function, and obtain the first unit, so that the components of the low frequency region in the frequency domain are retained and the components of the intermediate frequency to the high frequency region are attenuated; Calculate the difference with the first unit to obtain its corresponding first boundary, the first boundary includes its components in the intermediate frequency to the high frequency region; and use the corresponding boundary adjustment coefficient to the adjusting the magnitude of the first boundary to obtain its corresponding coding boundary; and superimposing the first unit on the coding boundary.
- the adjusting the magnitude of the first boundary by using the boundary adjustment coefficient corresponding thereto includes: determining that the boundary value of the first boundary is smaller than a preset first threshold, by being greater than 1 The boundary adjustment coefficient of 1 increases the amplitude of the first boundary; or it is determined that the boundary value of the first boundary is greater than a preset second threshold, and the amplitude of the first boundary is reduced by the boundary adjustment coefficient less than 1 value.
- enhancing the magnitude of the first boundary by using the boundary adjustment coefficient greater than 1 includes: selecting a coefficient from a preset first boundary adjustment coefficient group as the boundary adjustment coefficient, Enhancing the amplitude of the first boundary, the coefficients in the first boundary adjustment coefficient group are all greater than 1; the reducing the amplitude of the first boundary by the boundary adjustment coefficient less than 1 includes: A coefficient is selected from the set second boundary adjustment coefficient group as the boundary adjustment coefficient to reduce the amplitude of the first boundary, and the coefficients in the second boundary adjustment coefficient group are all less than 1.
- the adjusting the magnitude of the first boundary by using the corresponding boundary adjustment coefficient includes: taking the weighted value of the distortion rate and the code rate as an optimization target, and based on an optimization algorithm, obtaining the The boundary adjustment coefficient is used to adjust the amplitude of the first boundary.
- the performing data compression on the initial frame includes at least one of the following methods: performing the boundary adjustment on the initial frame first, and then predicting the initial frame after the boundary adjustment and calculating the residual; first predicting the initial frame to obtain a predicted frame, then performing the boundary adjustment and calculating the residual on the initial frame and the predicted frame; and first predicting and calculating the residual on the initial frame difference, and then perform the boundary adjustment on the residual.
- the compressed frame further includes the encoding function and the boundary adjustment coefficient corresponding to each unit in the plurality of units.
- this specification also provides a data processing system, including: at least one storage medium and at least one processor, the at least one storage medium stores at least one instruction set for data processing; the at least one processing The processor is connected in communication with the at least one storage medium, wherein, when the system is running, the at least one processor reads the at least one instruction set, and executes the first step of the present specification according to the instructions of the at least one instruction set.
- a data processing system including: at least one storage medium and at least one processor, the at least one storage medium stores at least one instruction set for data processing; the at least one processing The processor is connected in communication with the at least one storage medium, wherein, when the system is running, the at least one processor reads the at least one instruction set, and executes the first step of the present specification according to the instructions of the at least one instruction set.
- this specification also provides a data processing method, including: obtaining compressed data, the compressed data includes a compressed frame obtained by performing data compression on an initial frame, and the data compression includes boundary adjustment; Perform data decompression on the frame to obtain a decompressed frame, the data decompression includes performing boundary compensation on the decompressed frame, the decompressed frame includes the compressed frame and the compressed frame becomes the decompressed frame before the data decompression process Any data state of , wherein there is a preset association between the boundary compensation and the boundary adjustment.
- the boundary adjustment includes adjusting the amplitude of each unit in the multiple units of the compression frame using its corresponding boundary adjustment coefficient in the mid-frequency to high-frequency region, so as to reduce the The amplitude of the compression frame from the intermediate frequency to the high frequency region, the boundary adjustment coefficient is greater than 0, the compression frame includes the initial frame and the initial frame becomes the compression during the data compression process
- the boundary compensation includes using the boundary corresponding to the boundary adjustment coefficient based on the association relationship for each of the multiple units in the unframed The compensation coefficient compensates its amplitude in the mid-to-high frequency region.
- the adjusting the boundary of the compression frame includes: dividing the compression frame into the plurality of units based on a preset unit size; and using the corresponding unit for each unit
- the boundary adjustment coefficient adjusts its amplitude in the middle frequency to high frequency region, including for each unit: selecting a function from the preset encoding function group as the encoding function, and adjusting it through the encoding function , to obtain the first unit, so that the components in the low-frequency region in the frequency domain are retained and the components in the mid-frequency to high-frequency region are attenuated; and the difference between it and the first unit is obtained to obtain its corresponding first boundary, the The first boundary includes its components in the intermediate frequency to the high frequency region; and the amplitude of the first boundary is adjusted using the boundary adjustment coefficient corresponding thereto to obtain its corresponding coding boundary; and The first unit is superimposed on the encoding boundary.
- the performing boundary compensation on the de-frame includes: dividing the de-frame into the plurality of units based on the preset unit size; The boundary compensation coefficient corresponding to the boundary adjustment coefficient compensates its amplitude in the middle frequency to high frequency region.
- the using the boundary compensation coefficient corresponding to the boundary adjustment coefficient for each unit to compensate its amplitude in the mid-frequency to high-frequency region includes: : Determining the decoding function, adjusting it through the decoding function to obtain the second unit, so that the components in the low-frequency region in the frequency domain are retained and the components in the mid-frequency to high-frequency region are attenuated; and the second unit Calculate the difference of the unit, obtain its corresponding second boundary, the second boundary includes its components in the middle frequency to the high frequency region; and use the boundary compensation coefficient corresponding to the boundary adjustment coefficient to adjust the Compensating the magnitude of the second boundary to obtain its corresponding decoding boundary; and superimposing the current unit on the decoding boundary.
- the determining the decoding function includes: selecting a function from a preset decoding function group as the decoding function.
- the using the boundary compensation coefficient corresponding to the boundary adjustment coefficient to compensate the magnitude of the second boundary includes: selecting one from a set of preset boundary compensation coefficients as the The boundary compensation coefficient is used to compensate the amplitude of the second boundary.
- the compressed frame includes: the encoding function and the boundary adjustment coefficient corresponding to each unit in the plurality of units in the compressed frame.
- the determining the decoding function includes: selecting a function corresponding to the encoding function from a preset decoding function group as the decoding function.
- the compensating the magnitude of the second boundary by using the boundary compensation coefficient corresponding to the boundary adjustment coefficient includes: based on the relationship between the boundary adjustment coefficient and the boundary compensation coefficient An association relationship, determining the boundary compensation coefficient, and compensating the amplitude of the second boundary.
- the data decompression of the compressed frame includes at least one of the following methods: decoding the compressed frame first, and then performing the boundary compensation; performing the decompression on the compressed frame performing the boundary compensation during decoding; and performing the boundary compensation on the compressed frame before performing the decoding.
- the association relationship includes: the boundary compensation makes the amplitude of the decompressed frame at any frequency in the low frequency to intermediate frequency region not less than 85% of the original frame.
- the association relationship further includes: the boundary compensation makes the amplitude of the decompressed frame increase steadily in the intermediate frequency region relative to the initial frame.
- the association relationship further includes: the boundary compensation makes the amplitude of the decompressed frame decrease smoothly in the high-frequency region relative to the initial frame.
- this specification also provides a data processing system, including at least one storage medium and at least one processor, the at least one storage medium stores at least one instruction set for data processing; and the at least one processing
- the processor is connected in communication with the at least one storage medium, wherein when the system is running, the at least one processor reads the at least one instruction set, and executes the third aspect of the present specification according to the instructions of the at least one instruction set The method of data processing described.
- FIG. 1 shows a schematic diagram of a data processing system provided according to an embodiment of this specification
- Fig. 2 shows a schematic diagram of a data compression device for data processing provided according to an embodiment of this specification
- FIG. 3A shows a flow chart of data compression and data decompression provided according to an embodiment of this specification
- Fig. 3B shows a flow chart of data compression and data decompression provided according to an embodiment of this specification
- Fig. 3C shows a flow chart of data compression and data decompression provided according to an embodiment of this specification
- FIG. 4A shows a flow chart of a data processing method for compressing data provided according to an embodiment of the specification
- Fig. 4B shows a flow chart of boundary adjustment provided according to an embodiment of this specification
- Fig. 5 shows a structural block diagram of boundary adjustment provided according to an embodiment of the present specification
- FIG. 6 shows a graph of an encoding function provided according to an embodiment of the specification
- Fig. 7A shows a flowchart of a data processing method for decompressing a compressed frame according to an embodiment of the specification
- Fig. 7B shows a flow chart of boundary compensation provided according to an embodiment of this specification
- FIG. 8 shows a structural flowchart of a boundary compensation provided according to an embodiment of this specification
- FIG. 9A shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification
- FIG. 9B shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 9C shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 9D shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 10A shows a graph of an overall adjustment function H 0 (f), a boundary adjustment function H 1 (f) and a decoding function H 2 (f) in a normal mode according to an embodiment of the present specification.
- Fig. 10B shows a graph of an overall adjustment function H 0 (f), a boundary adjustment function H 1 (f) and a decoding function H 2 (f) of an enhancement mode provided according to an embodiment of the present specification.
- this specification provides a data processing system 100 (hereinafter referred to as the system 100).
- this specification describes a data processing method P200 for compressing data
- this specification describes a data processing method P300 for decompressing compressed frames.
- FIG. 1 shows a schematic diagram of a data processing system 100 .
- the system 100 may include a data compression device 200 , a data decompression device 300 and a transmission medium 120 .
- the data compression device 200 may receive an initial frame in the initial data to be compressed, and use the data processing method P200 proposed in this specification to compress the initial data to generate a compressed frame.
- the data compression device 200 may store data or instructions for performing the data processing method P200 described in this specification, and execute the data and/or instructions.
- the data decompression device 300 can receive the compressed frame, and use the data processing method P300 proposed in this specification to decompress the compressed frame to obtain the decompressed frame.
- the data decompression device 300 may store data or instructions for executing the data processing method P300 described in this specification, and execute the data and/or instructions.
- Data compression apparatus 200 and data decompression apparatus 300 may comprise a wide range of devices.
- data compression device 200 and data decompression device 300 may include desktop computers, mobile computing devices, notebook (e.g., laptop) computers, tablet computers, set-top boxes, handheld devices such as smart phones, televisions, cameras, display devices, digital media Players, video game consoles, in-vehicle computers, or the like.
- Transmission media 120 may facilitate the transfer of information and/or data.
- the transmission medium 120 may be any data carrier capable of transmitting compressed frames from the data compression device 200 to the data decompression device 300 .
- the transmission medium 120 may be a storage medium (eg, an optical disk), a wired or wireless communication medium.
- the communication medium may be a network.
- transmission medium 120 may be any type or combination of wired or wireless networks.
- the transmission medium 120 may include a cable network, a wired network, an optical fiber network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), Wide Area Network (WAN), Public Switched Telephone Network (PSTN), Bluetooth network, ZigBee network, Near Field Communication (NFC) network or similar network.
- LAN local area network
- WAN wide area network
- WLAN wireless local area network
- MAN metropolitan area network
- WAN Wide Area Network
- PSTN Public Switched Telephone Network
- Bluetooth network ZigBee network
- NFC Near Field Communication
- One or more components in data decompression device 300 and data compression device 200 may be connected to transmission medium 120 to transmit data and/or information.
- Transmission medium 120 may include routers, switches, base stations, or other devices that facilitate communication from data compression device 200 to data decompression device 300 .
- the transmission medium 120 may be a storage medium, such as mass storage, removable storage, volatile read-write storage, read-only memory (ROM), or the like, or any combination thereof.
- exemplary mass storage might include non-transitory storage media such as magnetic disks, optical disks, solid-state drives, and the like.
- Removable storage may include flash drives, floppy disks, compact discs, memory cards, zip disks, tapes, and more.
- Typical volatile read-write memory might include random access memory (RAM).
- RAM may include dynamic RAM (DRAM), dual date rate synchronous dynamic RAM (DDR SDRAM), static RAM (SRAM), thyristor RAM (T-RAM), and zero-capacitance RAM (Z-RAM), among others.
- DRAM dynamic RAM
- DDR SDRAM dual date rate synchronous dynamic RAM
- SRAM static RAM
- T-RAM thyristor RAM
- Z-RAM zero-capacitance RAM
- transmission medium 120 may be a cloud platform.
- the cloud platform may include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, inter-cloud cloud, etc., or a form similar to the above-mentioned forms, or any combination of the above-mentioned forms.
- the data compression device 200 receives the initial data, and executes the instructions of the data processing method P200 described in this specification, performs data compression on the initial data, and generates a compressed frame; the compressed frame is transmitted to the data through the transmission medium 120
- FIG. 2 shows a schematic diagram of a data compression device 200 for data processing.
- the data compression device 200 can execute the data processing method P200 described in this specification.
- the data processing method P200 is introduced in other parts of this specification.
- the data compression device 200 includes at least one storage medium 230 and at least one compression end processor 220 .
- the data compression device 200 may also include a communication port 250 and an internal communication bus 210 .
- the data compression device 200 may further include an I/O component 260 .
- Internal communication bus 210 may connect various system components, including storage media 230 and compression end processor 220 .
- I/O component 260 supports input/output between data compression device 200 and other components.
- the storage medium 230 may include a data storage device.
- the data storage device may be a non-transitory storage medium or a temporary storage medium.
- the data storage device may include one or more of a magnetic disk 232 , a read-only storage medium (ROM) 234 or a random-access storage medium (RAM) 236 .
- the storage medium 230 also includes at least one instruction set stored in the data storage device.
- the instructions are computer program codes, and the computer program codes may include programs, routines, objects, components, data structures, procedures, modules, etc. that execute the data processing methods provided in this specification.
- the communication port 250 is used for data communication between the data compression device 200 and the outside world.
- the data compression device 200 can be connected to the transmission medium 120 through the communication port 250 .
- At least one compression end processor 220 is communicatively connected with at least one storage medium 230 through the internal communication bus 210 . At least one compression end processor 220 is configured to execute the above at least one instruction set. When the system 100 is running, at least one compression processor 220 reads the at least one instruction set, and executes the data processing method P200 according to instructions of the at least one instruction set. The compression end processor 220 may execute all the steps included in the data processing method P200. The compression end processor 220 may be in the form of one or more processors.
- the compression end processor 220 may include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), Application Specific Integrated Circuit (ASIC), Application Specific Instruction Set Processor (ASIP), Central Processing Unit (CPU), Graphics Processing Unit (GPU), Physical Processing Unit (PPU), Microcontroller Unit, Digital Signal Processor (DSP), Field Programmable Gate Array (FPGA), Advanced RISC Machine (ARM), Programmable Logic Device (PLD), any circuit or processor capable of performing one or more functions, etc., or any combination thereof .
- RISC reduced instruction set computer
- ASIC Application Specific Integrated Circuit
- ASIP Application Specific Instruction Set Processor
- CPU Central Processing Unit
- GPU Graphics Processing Unit
- PPU Physical Processing Unit
- Microcontroller Unit Microcontroller Unit
- DSP Digital Signal Processor
- FPGA Field Programmable Gate Array
- ARM Programmable Logic Device
- PLD Programmable Logic Device
- the data compression device 200 in this specification may also include multiple processors, therefore, the operations and/or method steps disclosed in this specification may be executed by one processor as described in this specification, or by multiple processors. joint execution of the processors.
- the compression end processor 220 of the data compression device 200 executes step A and step B in this specification, it should be understood that step A and step B may also be jointly or separately executed by two different compression end processors 220 (for example , the first processor performs step A, and the second processor performs step B, or the first and second processors jointly perform steps A and B).
- the data decompression device 300 may execute the data processing method P300 described in this specification.
- the data processing method P300 is introduced in other parts of this specification.
- the data processing methods P200, P300 and the system 100 can be used for data compression and decompression, so as to improve the transmission efficiency of the data and save resources and space.
- the data may be non-real-time data or real-time data.
- the data may be non-real-time video data, audio data or image data, and so on.
- the data may also be real-time map data, real-time sensor data, real-time video monitoring data, network monitoring data, meteorological data, aerospace data, and so on.
- the data may be map data received from a base station while the self-driving car is driving.
- This manual does not limit the specific category of the data.
- the data processing methods P200, P300 described in this manual and the methods and steps taken by the system 100 when processing different types of data are consistent. For the convenience of presentation, this manual will take the processing of video data as an example. describe.
- a frame is a processing unit that makes up a data sequence.
- One or more initial frames may be included in the initial data.
- Each initial frame includes initial data of a preset number of bytes.
- the initial data may be original video data, and the initial frame may be a frame image in the original video data.
- H.264 and H.265 standards are usually used to encode initial video data, so as to achieve the purpose of compressing the video data.
- the H.264 and H.265 standards mainly adopt predictive coding when encoding video data, that is, predict the initial data in the video data to obtain the predicted value, and then compare the predicted value with the initial value of the initial data.
- the residual value is obtained by subtracting, so as to compress the video data.
- recovering and decompressing that is, decoding
- the initial frame can be recovered by adding the residual value and the predicted value.
- the data processing methods P200, P300 and system 100 provided in this manual can combine boundary adjustment and encoding when performing data compression, so as to reduce the amount of data during encoding, improve the compression efficiency of video data, and improve the transmission efficiency of video;
- boundary compensation and decoding can be combined to decompress the compressed data that has undergone boundary adjustment and encoding, so that the decompressed data can be restored to the original data.
- the data processing method P200 may perform data compression on the video data.
- the data processing method P200 may use a combination of coding (ie, prediction and residual calculation) and boundary adjustment to perform data compression on the initial frame to obtain a compressed frame.
- the data processing method P200 may perform the boundary adjustment and the encoding on the compressed frame.
- the compressed frame includes the initial frame and any data state before the initial frame becomes the compressed frame during the data compression process.
- the boundary adjustment refers to adjusting the amplitude of the spectrogram of the data to be processed.
- the boundary adjustment can adjust the amplitude of the selected area of the data to be processed in its frequency domain, such as the amplitude of the intermediate frequency area, the amplitude of the high frequency area, and the amplitude of the low frequency to intermediate frequency area. , and for example, the amplitude of the mid-frequency to high-frequency region, and so on.
- the boundary adjustment can be implemented by a boundary adjustment coefficient greater than 0 and less than 1 to attenuate the amplitude of a selected region in the frequency domain, thereby reducing the amount of data information in the data to be processed.
- the data processing method P300 may perform data decompression on the compressed frames subjected to the data compression by the data processing method P200 to obtain decompressed frames, so as to restore the video data.
- the data processing method P300 may adopt a method combining decoding (that is, recovering the compressed frame according to the residual value and the predicted value) and boundary compensation to perform data decompression on the compressed frame, so as to restore the data in the compressed frame.
- the data processing method P300 may perform the boundary compensation and the decoding on the deframed frame.
- the decompressed frame may include the compressed frame and any data state before the compressed frame becomes the decompressed frame during the data decompression process.
- the boundary compensation can make the data after the boundary adjustment be completely restored or approximately restored to the state before the boundary adjustment without considering other calculation errors.
- the data processing method P200, P300, and system 100 can significantly improve the compression efficiency of video data, reduce data loss during video data compression, improve video transmission efficiency, restoration rate and clarity of decompressed video, and reduce video after decompression. noise.
- the specific process of the boundary adjustment and the boundary compensation will be described in detail later.
- the order of the boundary adjustment and the encoding can be interchanged, or can be performed in a crossed manner.
- the boundary adjustment may be before or after the encoding.
- the order of the boundary compensation and the decoding can be interchanged, or can be performed in a crossed manner.
- the sequence of the boundary compensation and the decoding in the data decompression is the same as the sequence of the boundary adjustment and the encoding in the data compression
- the order of should be corresponding, that is, the boundary compensation and the decoding can operate symmetrically and inversely to the boundary adjustment and the encoding.
- the compressed frame should first perform the decoding and then perform the boundary compensation when data is decompressed.
- the data in the initial frame before data compression processing P 0
- the data in the decompressed frame decompressed by the data decompression device 300 P 4 .
- the data compression device 200 when the data compression device 200 performs data compression on the initial frame, it may first perform the boundary adjustment on the initial frame, and then perform the encoding; The encoding is performed, and then the boundary adjustment is performed.
- 3A to 3C show some data compression and data decompression flow charts provided according to the embodiments of this specification.
- Fig. 3A shows a flow chart of data compression and data decompression provided according to the embodiment of this specification.
- the data compression device 200 performs data compression on the initial data by: the data compression device 200 first performs the boundary adjustment on the initial frame P0 , and then performs the encoding, that is, after the boundary adjustment
- the initial frame is predicted and the residual is obtained to obtain the predicted data PI and the residual data R, and the predicted data PI and the residual data R are input into the code stream generation module for synthesis to obtain the compressed frame.
- Said compressed frame comprises said predicted data PI and said residual data R.
- the under-pressure frame may be the initial frame P 0 .
- the boundary adjustment may divide the compression frame (initial frame P 0 ) into multiple units, and perform the boundary adjustment using a corresponding boundary adjustment coefficient for each unit.
- the data compression device 200 may also input the coding function and boundary adjustment coefficient corresponding to each unit in the boundary adjustment into the code stream generation module for synthesis. That is, the compressed frame may further include the encoding function and the boundary adjustment coefficient corresponding to each unit in the boundary adjustment.
- the data of the coding function and the boundary adjustment coefficient corresponding to each unit in the compressed frame as coded data RAMI (Regional AmplitudeModulationInformation).
- coded data RAMI Registered AmplitudeModulationInformation
- the data compression method shown in FIG. 3A can improve coding efficiency, further reduce the amount of data in the compressed frame, improve the compression ratio, and at the same time reduce data loss and avoid loss of details.
- the data decompression performed by the data decompression device 300 on the compressed frame may be: the data decompression device 300 first performs the decoding on the compressed frame, and then performs the boundary compensation. Specifically, the data decompression device 300 may first perform the decoding on the compressed frame, that is, analyze the compressed frame based on the code stream analysis module, and generate the predicted data PI, the residual data R, and the encoded data RAMI ; Then perform prediction according to the prediction data PI to obtain a prediction frame, and superpose it with the residual data R to obtain a decoded frame. For the convenience of description, we define the data in the decoded frame as P 2 .
- the data decompression device 300 uses the decoded data corresponding to the encoded data RAMI to perform the boundary compensation on the decoded frame P2 based on the encoded data RAMI, to obtain the decompressed frame P4 for output.
- the deframed frame may be the decoded frame P2 .
- the boundary compensation may divide the decoded frame (decoded frame P 2 ) into multiple units, and perform the boundary compensation for each unit using the boundary compensation coefficient corresponding to the boundary adjustment coefficient .
- the decoded data may include a decoding function and a boundary compensation coefficient corresponding to each unit.
- the decoding function corresponds to the encoding function
- the boundary compensation coefficient corresponds to the boundary adjustment coefficient.
- the data decompression device 300 may determine the decoding function and the boundary compensation coefficient corresponding to each unit based on the encoded data RAMI. Details about the decoding function and the boundary compensation coefficient corresponding to the boundary compensation will be described in detail later.
- the transfer function between the decompressed frame P 4 and the initial data P 0 as the overall spectrum adjustment function H 0 (f).
- the manner shown in FIG. 3A can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding loss of details.
- the data compression performed by the data compression device 200 on the initial data may also be: incorporating the boundary adjustment into the encoding process.
- the boundary adjustment can be done at any stage in the encoding process.
- the boundary compensation may also be performed at a corresponding stage of the decoding process.
- Fig. 3B shows a flow chart of data compression and data decompression provided according to the embodiment of this specification.
- the data compression device 200 may perform data compression on the initial data as follows: the data compression device 200 first predicts the initial frame P 0 to obtain the predicted frame and the predicted data PI, and then compresses the initial frame P 0 Performing the boundary adjustment and calculating the residual with the predicted frame to obtain the residual data R; inputting the predicted data PI, the residual data R, and the coded data RAMI into a code stream generation module for synthesis, Get the compressed frame.
- the under pressure frame may be the predicted frame and the initial frame P 0 .
- the specific operation of the data compression shown in FIG. 3B is the same as that shown in FIG. 3A , but the operation sequence is different. The content about the boundary adjustment will be introduced in detail in the following description.
- performing data decompression on the compressed frame by the data decompression device 300 may be: performing the boundary compensation during the decoding process on the compressed frame by the data decompression device 300 .
- the data decompression device 300 may first analyze the compressed frame based on the code stream analysis module, generate the predicted data PI, the residual data R, and the coded data RAMI; perform prediction according to the predicted data PI Obtaining a predicted frame; based on the coded data RAMI, using the corresponding decoded data to perform the boundary compensation on the predicted frame; superimposing the boundary-compensated predicted frame with the residual data R, and performing The superimposed data is subjected to the boundary compensation to obtain the decompressed frame P 4 .
- the decomposed frame may be the predicted frame and superimposed data of the predicted frame and the residual data R.
- the data in the superimposed frame is P 3 .
- the specific process of performing the boundary correction on the boundary in the superimposed frame P3 will be described in detail later.
- the manner shown in FIG. 3B can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding loss of details.
- Fig. 3C shows a flow chart of data compression and data decompression provided according to the embodiment of this specification.
- the data compression device 200 may perform data compression on the initial data as follows: the data compression device 200 first predicts and calculates the residual of the initial frame P 0 to obtain the predicted data PI and the residual R 1 , and then The residual R 1 is subjected to the boundary adjustment to obtain the residual data R; the residual data R after the boundary adjustment, the prediction data PI, and the coded data RAMI are input into a code stream generation module to synthesize, The compressed frame is generated.
- the compressed frame may be the residual R 1 .
- the specific operation of the data compression method shown in FIG. 3C is the same as that shown in FIG. 3A , but the operation sequence is different. The content about the boundary adjustment will be introduced in detail in the following description.
- the data decompression performed by the data decompression device 300 on the compressed frame may be: the data decompression device 300 analyzes the compressed frame based on the code stream analysis module, and generates the predicted data PI and the residual data. R; then perform prediction according to the prediction data PI to obtain a prediction frame; based on the encoded data RAMI, use the corresponding decoded data to perform the boundary compensation on the residual data R to obtain a residual R 1 ; and The difference R 1 is superimposed on the predicted frame to obtain the decompressed frame P 4 .
- the re-deframing may be the residual data R.
- the manner shown in FIG. 3C can reduce the amount of data in the compressed frame, thereby improving the compression ratio and encoding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding loss of details.
- FIG. 4A shows a flowchart of a data processing method P200 for compressing data.
- the data compression device 200 can execute the data processing method P200.
- the storage medium in the data compression device 200 may store at least one set of instruction sets.
- the instruction set is configured to instruct the compression processor 220 in the data compression device 200 to complete the data processing method P200.
- the compression processor 220 can read the instruction set and execute the data processing method P200.
- the method P200 may include:
- S220 Select an initial frame P 0 in the initial data.
- a frame is a processing unit that makes up a data sequence. During data processing, calculations are often performed in units of frames.
- the initial data may include one or more initial frames.
- the initial frame P 0 includes initial data of a preset number of bytes.
- video data is used as an example for description in this specification, therefore, the initial data may be initial video data, and the initial frame P 0 may be a frame image in the initial video data.
- the data compression device 200 may select a part of frame images from the initial data as the initial frame P 0 , or select all frame images in the initial data as the initial frame P 0 .
- the data compression device 200 may select the initial frame P 0 according to the initial data application scenario.
- a partial frame image can be selected as the initial frame P 0 .
- Many frames of the monitoring image in a secluded place are the same, and the data compression device 200 may select a part of the frame images as the initial frame P 0 for compression and transmission.
- the data compression device 200 may select all frame images as the initial frame P 0 for compression and transmission.
- Said data compression may comprise said boundary adjustment and said encoding of compressed frames.
- the performing the boundary adjustment on the in-compression frame may be inputting the in-compression frame into a boundary regulator to perform boundary adjustment.
- the compressed frame may include the initial frame P 0 and any data state before the initial frame P 0 becomes the compressed frame during the data compression process.
- the compressed frame includes the initial frame P 0 and any data state of the initial frame P 0 in the process of boundary adjustment and encoding, for example, an initial frame, a predicted frame, and a residual frame ,and many more.
- the boundary adjustment refers to adjusting the amplitude of the spectrogram in the compressed frame.
- the boundary adjustment can adjust the amplitude of the selected area of the compressed frame in its frequency domain, such as the amplitude of the intermediate frequency area, the amplitude of the high frequency area, and the amplitude of the low frequency to intermediate frequency area. , and for example, the amplitude of the mid-frequency to high-frequency region, and so on.
- the boundary adjustment can be realized by a boundary adjustment coefficient to adjust the amplitude of a selected region in the frequency domain.
- the boundary adjustment may use a boundary adjustment coefficient greater than 0 and less than 1 to attenuate the amplitude of a selected region in the frequency domain, thereby reducing the amount of data information in the compression frame.
- receivers have different degrees of sensitivity to frequency, so the data compression operation can select different regions in the frequency domain for amplitude attenuation according to different forms of data.
- the mid-frequency to high-frequency components in the frequency spectrum of a frame of data are mainly concentrated in the area where the data changes drastically in this frame of data, that is, the boundary data of the data.
- the medium-frequency to high-frequency data is mainly concentrated on the boundary of the object in the image, that is, the boundary data of this frame of image. Since the edge part of the object in the picture is rich in mid-frequency and high-frequency information, the mid-frequency and high-frequency areas will carry more data. Therefore, reducing the amplitude of the mid-frequency to high-frequency region will visually blur the boundary data of the compressed frame, and also greatly reduce the amount of information in the image. It should be noted that reducing the amplitude of the low-frequency region will also reduce the amount of information in the image.
- the boundary adjustment may be to adjust the amplitude of the intermediate frequency to high frequency region of the compressed frame, such as attenuating the amplitude of the intermediate frequency to high frequency region to reduce the intermediate frequency to the amount of data information in the high-frequency region.
- the amplitude of the intermediate frequency to high frequency region in the intermediate state frame processed by the boundary adjustment is attenuated, and the amount of data information is also reduced. Therefore, the intermediate state frame processed by the boundary adjustment will have a higher compression ratio in encoding.
- the data processing method P200 can compress the initial frame P0 by using a method combining boundary adjustment and encoding, so as to adjust the amplitude of the intermediate frequency to high frequency region, so as to reduce the amount of data information, Further improve the compression ratio of video data and improve the efficiency of video transmission.
- the order of the boundary adjustment and the encoding can be interchanged, or can be performed in a crossed manner.
- Step S240 may include at least one of the data compression methods shown in FIG. 3A , FIG. 3B and FIG. 3C . For the convenience of presentation, this specification will take the manner shown in FIG. 3A as an example to describe step S240 in detail.
- the data compression device 200 first performs the boundary adjustment on the initial frame P 0 to attenuate the amplitude of the initial frame P 0 in the mid-frequency to high-frequency region, so that the boundary of the initial frame P 0
- the information is fuzzy, and the encoding adjustment frame P 1 is obtained to reduce the amount of information in the initial frame P 0 , thereby reducing the space resources occupied by the compression of the initial frame P 0 ; and calculate the residual), predict the encoding adjustment frame P1 to obtain the prediction frame of the encoding adjustment frame P1 and the prediction data PI; then combine the prediction frame of the encoding adjustment frame P1 with the encoding
- the adjustment frame P1 is subtracted to obtain the residual data R; the residual data R, the predicted data PI and the coded data RAMI are input into the code stream generation module for synthesis to obtain the compressed frame.
- the data processing method P200 can improve the encoding efficiency of the encoding adjustment frame P1 , further reduce the amount of data in the compressed frame, improve the encoding efficiency, and increase the compression ratio. Since the object of the boundary adjustment is the initial frame P 0 , the frame under compression is the initial frame P 0 .
- performing the data compression on the compressed frame may include executing by at least one compression end processor 220 of the data compression device 200:
- step S242 Perform the boundary adjustment on the under-compression frame (initial frame P 0 ) to obtain the coding adjustment frame P 1 .
- Fig. 4B shows a flow chart of boundary adjustment provided according to an embodiment of this specification;
- Fig. 5 shows a structural block diagram of a boundary adjustment provided according to an embodiment of this specification.
- step S242 may include being executed by at least one compression end processor 220 of the data compression device 200:
- the data processing unit processed may be a frame of data or a part of a frame of data.
- the area may be a frame or a field of images, or a part of a frame/field of images.
- an image is further divided into a slice, a tile, a coding unit (CU), a macroblock, a block, or a subblock.
- a region is usually an NxN square, or an MxN rectangle. Regions include but are not limited to the above names. For convenience of description, we define each region as a unit.
- the boundary adjustment can take the above-mentioned units as adjustment objects, and perform the boundary adjustment on each unit.
- the size of the unit can be arbitrarily selected according to needs, that is, the values of M and N can be any integers, such as 4, 8, 16, 32, 16, 128, or 256, or even smaller, such as 2 .
- the cell may contain only one pixel.
- the under-compression frame (initial frame P 0 ) can be divided into multiple units. For the convenience of description, we define the data of the unit in the i-th row and j-th column in the initial frame P0 as
- the boundary adjustment may be to use the corresponding boundary adjustment coefficient for each unit in the multiple units of the compressed frame (initial frame P 0 ) to adjust its amplitude in the mid-frequency to high-frequency region The value is adjusted to reduce the overall amplitude of the under-compressed frame (initial frame P 0 ) in the mid-frequency to high-frequency region.
- the boundary adjustment coefficient will be introduced in detail in the following description.
- step S242-4 may include the current unit implement:
- S242-42 Select a function from the preset coding function group as the coding function Through the encoding function for the current unit Adjust to get the first unit In the frequency domain, the components in the low frequency region are preserved and the components in the mid to high frequency region are attenuated.
- encoding function In the frequency domain there can be a low-pass filter such that the current unit in the initial frame P 0 The magnitude of is smoothly reduced in the frequency domain, so that the current unit in the initial frame P 0 In the frequency domain, the components of the low-frequency region are preserved and the components of the mid-frequency to high-frequency region are attenuated, thereby obtaining the current unit Corresponding first unit In order to save the amount of calculation required in the implementation process and avoid the occurrence of ringing effects, the encoding function should make the current unit Amplitudes in the frequency domain transition smoothly.
- encoding function It may be a low-pass filter with smooth transition in any form, which is not limited in this specification.
- the encoding function is a smooth transition curve, avoiding sharp changes in the amplitude adjustment gain in the curve.
- the ringing effect refers to that in image processing, when performing spectrum adjustment processing on an image, if the selected coding function With faster changes, the image will "ring".
- the so-called “ringing” refers to the vibration generated at the sharp change of the gray level of the output image, just like the air vibration generated after the clock is struck. The ringing effect mostly appears at the image boundary.
- the adjustment can be performed in the time domain using an encoded convolution kernel on the current unit Do convolution.
- the encoding function for the current unit adjustments can be expressed as in the current unit Multiply the transfer function in the frequency domain (ie encoding function) or perform corresponding convolution calculations in the time domain. If the current unit For digitized data, the convolution operation can be selected with the encoding function The corresponding coded convolution kernel performs the convolution operation.
- this specification will take convolution in the time domain as an example to describe the for the current unit Make adjustments. But those skilled in the art should understand that by multiplying the coding function in the frequency domain The method is also the scope to be protected in this specification.
- the encoding function group may be stored in the storage medium of the data compression device 200 .
- the set of encoding functions may include at least one different encoding function.
- Each encoding function corresponds to an encoding convolution kernel. That is to say, the storage medium of the data compression device 200 may include at least one coded convolution kernel.
- data compression device 200 for the current unit When performing convolution, one of the encoding function groups can be arbitrarily selected as the current unit Corresponding encoding function And use its corresponding convolution kernel as the coded convolution kernel, for the current unit Do convolution.
- a convolution kernel or a combination of convolution kernels with a higher order is required in the implementation process. This means an unnecessary increase in computation.
- high-order convolution kernels are more likely to cause strong color oscillations in the output image where the grayscale or color changes sharply, which is called the ringing effect.
- the ringing effect mostly appears at the image boundary.
- the encoding function can be analyzed in the frequency domain
- the amplitude of the mid-low frequency region is adjusted, so that the change of the amplitude adjustment gain in the mid-low frequency region is smooth and continuous.
- the ratio of the absolute value of the sum of negative coefficients to the sum of non-negative coefficients in the corresponding coded convolution kernel should be less than 0.1.
- the convolution kernel coefficients in the coded convolution kernel may all be non-negative numbers. Taking video data as an example, when there are many negative coefficients in the encoding convolution kernel, the pixel values at the image boundary are very different, and a large pixel value multiplied by a negative coefficient will make the final convolution The result is smaller, reflected in the image as darker pixels.
- the coefficients of the coded convolution kernel can be all non-negative numbers, or the absolute value of the sum of the negative coefficients in the coded convolution kernel and the value of the non-negative coefficient The ratio of the sum should be less than 0.1, that is, a small number of negative coefficients with small absolute values are allowed to appear in the encoding convolution kernel.
- Fig. 6 shows a kind of encoding function provided according to the embodiment of this description
- the horizontal axis is the normalized frequency f
- the vertical axis is the encoding function
- the amplitude adjustment gain H 1 .
- the normalized frequency f of the horizontal axis can be divided into a low frequency region, a middle and low frequency region, a middle frequency region, a middle and high frequency region and a high frequency region. Different types of data may have different definitions for low frequency, middle frequency and high frequency regions.
- the maximum value of the normalized frequency on the horizontal axis is 0.5.
- the high-frequency region may include frequencies between (d, 0.5] in the normalized frequency domain.
- d is the frequency lower limit of the high-frequency region.
- d may be in the normalized frequency domain 0.35, 0.36, 0.37, 0.38, 0.39, 0.4, 0.41, 0.42, 0.43, 0.44, and any one frequency in 0.45.
- the high frequency may include normalized frequency domain (0.33, 0.5].
- the high frequency may include 0.33, 0.34, 0.35, 0.36, 0.37, 0.38, 0.39, 0.4, 0.41, 0.42, 0.43, 0.44, 0.45, The interval between any two frequencies in 0.46, 0.47, 0.48, 0.49, 0.5, wherein 0.5 is the maximum frequency of the normalization.
- the intermediate frequency region can include the frequency between (b, c], where b is The lower frequency limit of the intermediate frequency region, c is the upper frequency limit of the intermediate frequency region.Such as, the lower frequency limit b of the intermediate frequency region can be 0.15,0.16,0.17,0.18,0.19, Any frequency among 0.2, 0.21, 0.22, 0.23, 0.24, 0.25, 0.26, 0.27 and 0.28; the frequency upper limit c of the intermediate frequency region can be 0.35, 0.34, 0.33, 0.32 in the normalized frequency domain and any frequency in 0.31.
- the low frequency region can include the frequency between [0, a] in the normalized frequency domain. Wherein a is the frequency upper limit of the low frequency region.
- the frequency upper limit a of the low frequency region It can be any one of 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.10, 0.12, 0.13, 0.14 and 0.15 in the normalized frequency domain.
- the frequency region between the two is called the middle and low frequency region.
- the frequency region between the two is called the middle and high frequency region .
- the stop band interval in can be any interval between the frequency 0.15 and 0.50.
- the stop band interval in can be any two intervals specified in 0.15, 0.17, 0.19, 0.21, 0.23, 0.25, 0.27, 0.29, 0.31, 0.33, 0.35, 0.37, 0.39, 0.41, 0.43, 0.45 and 0.50.
- the passband interval in can be any interval between frequency 0 and 0.35.
- the passband interval in can be any two of 0, 0.02, 0.04, 0.06, 0.08, 0.10, 0.12, 0.14, 0.15, 0.17, 0.19, 0.21, 0.23, 0.25, 0.27, 0.29, 0.21, 0.23 and 0.35. within the specified range.
- Fig. 6 is only illustrated by taking video data as an example, and those skilled in the art should understand that the encoding function The curves are not limited to the form shown in Figure 6, all capable of making the current unit The magnitude of is smoothly reduced in the frequency domain, so that the current unit in the initial frame P 0
- m ⁇ 1 Represents a linear combination of n functions
- k m represents the weight corresponding to the mth function
- q ⁇ 1 Represents the product combination of n functions
- k q represents the weight corresponding to the qth function
- Table 1 shows a parameter table of a coded convolution kernel provided according to an embodiment of this specification.
- Table 1 exemplarily lists the parameters of an encoding convolution kernel, where each row in Table 1 represents an encoding convolution kernel.
- the gray value of the pixels in the middle is within 0-255, therefore, in this embodiment, the convolution result needs to be divided by 16.
- the encoding convolution kernel is based on the encoding function obtained by Fourier transform.
- Table 1 is just an exemplary illustration, and those skilled in the art should know that the encoding convolution kernel is not limited to the parameters shown in Table 1, all of which can make the current unit
- the magnitude of is smoothly reduced in the frequency domain, so that the current unit in the initial frame P 0 In the frequency domain, the components of the low-frequency region are preserved and the components of the mid-frequency to high-frequency regions are attenuated. All belong to the scope of protection of this specification.
- the data compression device 200 may perform steps S242-42 for each unit. per unit in the coded function After adjustment, a corresponding first unit is obtained Encoding functions corresponding to all units Combined to form the encoding function H 1 (f). Encoding function corresponding to each unit
- the encoding function H 1 (f) can be generated by combining according to the positions of the units, and the encoding function H 1 (f) can be regarded as a matrix. It should be noted that when performing steps S242-42 on different units, the same encoding function (ie encoding convolution kernel) may be selected, or different encoding functions (ie encoding convolution kernel) may be selected. i.e. different units Corresponding encoding function Can be the same or different.
- All units correspond to the first unit Combined to form the first frame P 1b .
- the first unit corresponding to each unit The first frame P 1b can be generated by combining according to the positions of each unit, and the first frame P 1b is a blurred image.
- the relationship between P 0 and P 0b can be expressed as the following formula:
- step S242-4 may include the current unit implement:
- the mid-frequency to high-frequency components in the data spectrum of each frame are mainly concentrated in the area where the data changes sharply in this frame of data, that is, the boundary data of the data.
- the intermediate-frequency to high-frequency data is mainly concentrated on the boundary of the object in the image, that is, the boundary data of this frame of image.
- the encoding function make the current unit The amplitude in the frequency domain is smoothly reduced to attenuate components in the mid to high frequency region. Therefore, the first unit It can be understood as removing the current unit The data of the boundary information in .
- the first boundary include current unit Components in the middle frequency to the high frequency region.
- the first boundary include current unit Boundary information in . per unit
- a corresponding first boundary is obtained First boundary for all cells Combined to form the first boundary frame E 1b .
- the first boundary corresponding to each cell The first boundary frame E 1b may be generated by combining the units according to their positions.
- the first boundary frame E 1b can be expressed as the following formula:
- step S242-4 may include the current unit implement:
- the boundary adjustment coefficient corresponding to the boundary adjustment is defined as g 1 .
- the corresponding boundary adjustment coefficient is defined as For a frame of image, there may be a strong boundary, a weak boundary, or both a strong boundary and a weak boundary.
- the strong boundary may be a boundary in which pixel values between adjacent pixels differ greatly.
- the weak boundary may be a boundary with a small difference in pixel values between adjacent pixels.
- the initial frame P 0 is divided into multiple units through step S242-2, thereby dividing the initial frame P 0 into multiple small areas. The smaller the size of the cell, the more singular the boundaries contained in the cell. When the cells are small enough, each cell may include only strong boundaries or only weak boundaries. Therefore, by dividing the initial frame P0 into multiple units in step S242-2, the data compression device 200 can individually adjust the boundary information of each unit.
- the boundary adjustment can be performed on those weak boundaries with small differences between adjacent pixels at mid-frequency to high-frequency
- the amplitude of the area is enhanced, thereby increasing the difference between the pixel values of the adjacent pixels in the weak boundary, avoiding the loss of the weak boundary during the encoding process, and avoiding the loss of details, so that after encoding and decoding, the details can still be reserve.
- the boundary adjustment can attenuate the amplitude of the strong boundary in the mid-frequency to high-frequency region with a large difference between adjacent pixels, so as to reduce the pixel value between adjacent pixels in the strong boundary The difference, thereby reducing the data information contained in the strong boundary and improving the compression ratio.
- the difference between adjacent pixels in the strong boundary is large enough, even after the boundary adjustment is performed to attenuate the amplitude, the retained boundary is still large enough, and will not disappear after encoding and decoding processing.
- the boundary adjustment coefficient when using the boundary adjustment coefficient to the first boundary
- the information contained in the initial frame P 0 should be preserved as much as possible without loss, so that the information can be better recovered during decompression. Therefore, the boundary adjustment coefficient Should be greater than 0.
- Adjustment coefficient at the border Processed Encoding Boundary The amplitude in the mid-to-high frequency region is also greater than zero without any data loss. Therefore, all data can be recovered when the compressed data is decompressed.
- the boundary adjustment coefficient If there is a zero point, the data in the mid-frequency to high-frequency region in the unit corresponding to the zero point may be lost, and the decoding end will not be able to recover the lost data during decompression, so the original data cannot be recovered.
- step S242-46 the data compression device 200 can Corresponding to the first boundary
- the size of the boundary value determines the current cell Corresponding boundary adjustment coefficient different units May have different boundary adjustment factors
- the boundary value may be the first boundary The value corresponding to each pixel in .
- Step S242-46 may include:
- S242-462 Determine the current unit corresponding to the first boundary The boundary value is less than the preset first threshold value, and the boundary adjustment coefficient is greater than 1 enhance the first boundary the magnitude of .
- S242-464 Determine the current unit corresponding to the first boundary The boundary value is greater than the preset second threshold, and the boundary adjustment coefficient is less than 1 Lower the first boundary the magnitude of .
- steps S242-462 if the current unit Corresponding to the first boundary
- the boundary value in is smaller than the first threshold, which means that the current unit contains weak boundaries, which contain many small details.
- use a boundary adjustment factor greater than 1 for the current unit Corresponding to the first boundary Adjust the amplitude of the mid- to high-frequency region to enhance the current unit Corresponding to the first boundary
- the amplitude of the mid-frequency to high-frequency region maintain the current unit SNR to avoid current cell
- the weak boundary in is lost in the process of data compression (prediction and residual), avoiding the loss of details and ensuring the coding effect.
- the data processing method P200 can improve the efficiency of data compression and at the same time enhance the amount of data information of weak boundaries, so as to avoid loss of details during the process of data compression, that is, reduce data distortion while improving the efficiency of data compression.
- the first threshold may be 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, and so on. In some embodiments, the first threshold may be smaller, such as 1, 2, 3, 4, and so on. In some embodiments, the first threshold may be any number between 5 and 30.
- steps S242-464 if the current unit Corresponding to the first boundary
- the boundary value in is greater than the second threshold, representing the current unit contains a strong boundary, use a boundary adjustment coefficient less than 1 and greater than 0 for the current unit Corresponding to the first boundary
- the amplitude of the mid- to high-frequency region is adjusted to reduce the current unit Corresponding to the first boundary amplitude in the mid- to high-frequency region, reducing the current unit Corresponding to the first boundary
- the signal strength in the mid-frequency to high-frequency region is reduced, thereby reducing the amount of data information, and improving the efficiency of data compression when making predictions and calculating residuals.
- the second threshold may be greater than the first threshold, or equal to the first threshold.
- the boundary value in is at the boundary value (including the first threshold and the second threshold) between the first threshold and the second threshold, and no boundary adjustment can be made.
- the boundary adjustment coefficient Can be 1.
- the first boundary adjustment coefficient group and the second boundary adjustment coefficient group may be pre-stored in the storage medium of the data compression device 200 .
- the first set of boundary adjustment coefficients includes at least one coefficient. And the coefficients in the first boundary adjustment coefficient group are all greater than 1.
- the coefficients in the first boundary adjustment coefficient group may be any number greater than 1.
- the data compression device 200 executes steps S242-462, it may select a coefficient from the preset first boundary adjustment coefficient group as the boundary adjustment coefficient enhance the first boundary the magnitude of .
- the data compression device 200 selects the boundary adjustment coefficient from the preset first boundary adjustment coefficient group , according to the first boundary Choose the size of the boundary value in .
- first border The larger the boundary value of , the corresponding boundary adjustment coefficient smaller; first boundary The smaller the boundary value of , the corresponding boundary adjustment coefficient bigger.
- the second set of boundary adjustment coefficients includes at least one coefficient. And the coefficients in the second boundary adjustment coefficient group are all coefficients greater than 0 and less than 1.
- the coefficients in the second boundary adjustment coefficient group may be any number greater than 0 and less than 1.
- the data compression device 200 executes steps S242-464, it may select a coefficient from the preset second boundary adjustment coefficient group as the boundary adjustment coefficient reduce the magnitude of the first bound
- the data compression device 200 selects the boundary adjustment coefficient from the preset second boundary adjustment coefficient group , according to the first boundary Choose the size of the boundary value in .
- first border The larger the boundary value of , the corresponding boundary adjustment coefficient smaller; first boundary The smaller the boundary value of , the corresponding boundary adjustment coefficient bigger.
- the data compression device 200 can also determine the current unit through an optimization algorithm Corresponding boundary adjustment coefficient Specifically, the data compression device 200 can establish an optimization equation. For example, the data compression device 200 can use the current unit The weighted value of the corresponding distortion rate and code rate is an optimization target. the current unit The minimum weighted value of the corresponding distortion rate and code rate is the optimization goal. Based on the optimization algorithm, the boundary adjustment coefficient perform iterative calculations to determine the boundary adjustment coefficient Adjust the coefficients with the bounds to the first boundary The amplitude is adjusted.
- the data compression device 200 may perform steps S242-46 for each unit. per unit Corresponding to the first boundary Adjustment coefficient at the border After adjustment, a corresponding coding boundary is obtained Boundary adjustment coefficients for all elements Combined to form the boundary adjustment coefficient g 1 . Boundary adjustment coefficients corresponding to each unit The boundary adjustment coefficient g 1 can be generated by combining according to the position of each unit. The boundary adjustment coefficient g 1 can be regarded as a matrix. It should be noted that different units Corresponding boundary adjustment coefficient Can be the same or different.
- Coding boundaries corresponding to all units Combined to form the coded boundary frame E 1m Coding boundaries corresponding to each unit
- the encoding boundary frame E 1m can be generated by combining the units according to their positions.
- the relationship between E 1m and E 1b can be expressed as the following formula:
- step S242-4 may include the current unit implement:
- H E (f) g 1 (1-H 1 (f)) + H 1 (f) formula (6)
- the initial frame P0 is divided into multiple units by step S242-2, thereby dividing the initial frame P0 into multiple small areas, so that different boundary adjustment coefficients can be selected according to the characteristics of each unit Therefore, it is more flexible to adjust the signal strength of the mid-frequency to high-frequency region components, and achieve a better balance between the coding effect and the size of the code stream.
- step S240 may also include:
- S244 Perform the encoding (prediction and residual calculation) on the encoding adjustment frame P1 to obtain the prediction data PI and the residual data R.
- step S240 may also include:
- S248 The encoding function of each unit of the initial frame P 0 in the boundary adjustment process and boundary adjustment coefficient input to the code stream generation module for synthesis to obtain the compressed frame.
- the encoding function H 1 (f) and the boundary adjustment coefficient g 1 corresponding to the initial frame P 0 are input into the code stream generation module for synthesis to obtain the compressed frame.
- the compressed frame not only includes the prediction data PI and the residual data R, but also includes the encoding function corresponding to each unit in the plurality of units and the boundary adjustment coefficient
- the encoding function corresponding to each unit and the boundary adjustment coefficient That is, the aforementioned coded data RAMI.
- the boundary adjustment coefficient g1 may be the boundary adjustment coefficient
- the difference between, that is, with one of the multiple units as the basic data, the boundary adjustment coefficient corresponding to the other units is the difference between it and the base data.
- the boundary adjustment coefficient g 1 can also be the boundary adjustment coefficient adjustment coefficient with a boundary near it poor. Wherein, both m and n are integers and are not zero at the same time. In this way, the amount of data in the boundary adjustment coefficient g 1 can be further reduced, and the compression ratio can be further improved.
- the coding adjustment frame P 1 is obtained.
- the encoding adjusts the mid-to-high amplitude attenuation of strong boundaries in frame P1 .
- the data compression device 200 makes the coding adjustment frame P 1 as a whole after performing the boundary adjustment on the initial frame P 0
- the amount of data information is reduced.
- the data compression device 200 performs encoding and code stream generation calculation on the encoding adjustment frame P1 , which can improve the encoding efficiency of the initial frame P0 , thereby increasing the compression ratio of the initial frame and improving the transmission efficiency of the initial data.
- the encoding adjusts the amplitude enhancement from the middle frequency to the high frequency of the weak boundary in the frame P 1 , so as to avoid the loss of details.
- the data processing method P200 can perform the boundary adjustment on the initial frame at the same time, while improving the compression ratio of the initial frame, improving the encoding efficiency and the transmission efficiency of the initial data, it can also reduce the Data loss, avoid loss of details.
- FIG. 7A shows a flowchart of a data processing method P300 for decompressing compressed frames.
- the data decompression device 300 can execute the data processing method P300.
- the storage medium in the data decompression device 300 may store at least one set of instructions.
- the instruction set is configured to instruct the decompression processor in the data decompression device 300 to complete the data processing method P300.
- the processor at the decompression end can read the instruction set and execute the data processing method P300.
- the method P300 may include:
- the compressed data includes the compressed frame.
- the compressed data may include the compressed frame obtained by performing data compression on the initial frame in the initial data by the data processing method P200.
- the compressed frame comprises compressed prediction data PI and residual data R.
- the compressed frame further includes an encoding function corresponding to each unit in the plurality of units in the compressed frame and boundary adjustment coefficient That is, the coding function H 1 (f) and the boundary adjustment coefficient g 1 , that is, the aforementioned coded data RAMI.
- the compressed frame also includes a coding function H 1 (f) and a boundary adjustment coefficient g 1 . As shown in FIG.
- step S320 may include: inputting the compressed frame into the code stream analysis module for analysis and calculation, obtaining the prediction data PI and the residual data R, and the encoding function H 1 (f) and Boundary adjustment factor g 1 .
- a frame is a commonly used processing unit that composes a data sequence. During data processing, calculations are often performed in units of frames.
- the initial data may be compressed in units of frames.
- the data decompression device 300 decompresses the compressed frames
- data decompression may also be performed in units of frames.
- S340 Perform data decompression on the compressed frame to obtain a decompressed frame.
- the data decompression refers to performing decompression calculation on the compressed frame to obtain a decompressed frame, restoring or substantially restoring the decompressed frame to the original data, or making the decompressed frame clearer than the initial data.
- the threshold can be any value between 80%-90%.
- the threshold can be in the closed interval defined by any two values among 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, and 90%. any value of .
- the data decompression should make the amplitude of the decompressed frame at any frequency from low frequency to intermediate frequency not less than 85% ⁇ 3% of the initial frame.
- the data decompression includes performing boundary compensation on the decompressed frame, and decoding the data after the boundary compensation, so as to obtain the required decompressed frame.
- the frame of data being decompressed during decompression includes the compressed frame and any data state before the compressed frame becomes the decompressed frame during the decompression process.
- the data processing method P200 uses a combination of boundary adjustment and encoding to compress the initial frame, so as to further increase the compression ratio of video data and improve the efficiency of video transmission.
- the data processing method P300 can decompress the compressed frame by decoding (that is, recovering the compressed frame according to the residual data R and the predicted data PI) and boundary compensation to obtain the required decompressed frame , to recover the data in the compressed frame.
- the frame being decompressed may include the compressed frame and any data state of the compressed frame during decoding according to the prediction data PI and residual data R.
- the decompressed frame may be the compressed frame, a decoded frame obtained through decoding, or a predicted frame obtained through prediction, and so on.
- the data compression may be to adjust the amplitude of the mid-frequency to high-frequency region of the compressed frame through the boundary adjustment, for example, attenuating the amplitude of the mid-frequency to high-frequency region to reduce the mid-frequency to high-frequency range.
- the amount of data information in the high-frequency region thereby reducing the amount of data information in the compression frame.
- video data because the edge part of the object in the image is rich in mid-frequency and high-frequency information, and the mid-frequency and high-frequency regions will carry more data, so reducing the amplitude of the mid-frequency to high-frequency region will visually make the
- the data at the boundary of the compression frame is blurred, and the amount of information in the image will also be greatly reduced.
- the data decompression may be to compensate the amplitude of the mid-frequency to high-frequency region of the compressed frame through boundary compensation, for example, to enhance the magnitude of the mid-frequency to high-frequency region to restore it to the original frame , or enhanced relative to the state in the initial frame.
- the boundary compensation applied when performing the data decompression on the compressed frame refers to inputting the decompressed frame into a boundary compensator to perform boundary compensation.
- the boundary compensation may correspond to the boundary adjustment, that is to say, there should be a preset relationship between the boundary compensation and the boundary adjustment.
- boundary compensation There should be a preset relationship between the boundary compensation and the boundary adjustment, which may be that there is a preset relationship between the decoded data (decoding function H 2 (f) and boundary compensation coefficient g 2 ) in the boundary compensation relation.
- decoded data decoding function H 2 (f) and boundary compensation coefficient g 2
- encoded data RAMI encoded data RAMI
- decoding function H 1 (f) and boundary adjustment coefficient g 1 decoding data
- decoding function H 2 (f) and boundary The correlation between the compensation coefficients g 2 ) will be introduced in detail in the following description.
- step S340 may include:
- said de-frame may be said decoded frame P2 .
- the compressed frame may be obtained by encoding the encoding adjustment frame P1 by the data compression device 200 .
- the data decompression device 300 may decode the compressed frame to obtain the decoded frame P 2 . That is, a predicted frame is obtained by predicting according to the predicted data PI, and superimposed on the residual data R to obtain decoded data P 2 , which is the data P 2 of the decoded frame.
- the relationship can be expressed as the following formula:
- S344 Perform the boundary compensation on the decompressed frame (decoded frame P 2 ), to obtain the decompressed frame P 4 .
- FIG. 7B shows a flow chart of boundary compensation provided according to an embodiment of this specification
- FIG. 8 shows a flow chart of a structure of boundary compensation provided according to an embodiment of this specification.
- step S344 may include being executed by at least one decompression end processor of the data decompression device 300:
- S344-2 Divide the decoding frame (decoded frame P 2 ) into the plurality of units based on the preset unit size.
- the division method of the decompression frame (decoded frame P 2 ) can be consistent with the division method of the compression frame (initial frame P 0 ) in step S242-2, and each unit corresponds to each other, which will not be repeated here repeat.
- S344-4 Use the boundary compensation coefficient corresponding to the boundary adjustment coefficient for each unit to compensate its amplitude in the middle frequency to high frequency region.
- the data processing method P300 can use the unit during data compression as the data decompression unit, and use the boundary compensation coefficient corresponding to the boundary adjustment coefficient to perform boundary on the amplitude of the mid-frequency to high-frequency region for each unit. Compensation to compensate for the reduced amplitude of the mid-to-high frequency region during data compression resulting in decompressed frames.
- the boundary compensation corresponds to the boundary adjustment, and there is a corresponding relationship between the boundary compensation coefficient and the boundary adjustment coefficient.
- the boundary compensation can restore the compressed data after the boundary adjustment to a resolution of the original frame that is even higher than that of the original frame.
- the boundary compensation may be to use the boundary compensation corresponding to the boundary adjustment coefficient based on the association relationship for each of the multiple units in the deframe (decoded frame P 2 ). coefficients to compensate for its amplitude in the mid to high frequency region.
- the association relationship and the boundary compensation coefficient will be described in detail in the following description.
- step S344-4 may include the current unit implement:
- S344-42 Determine the decoding function Through the decoding function for the current unit Adjustment is made to obtain the second unit In the frequency domain, the components in the low frequency region are preserved and the components in the mid to high frequency region are attenuated.
- decoding function In the frequency domain there can be a low-pass filter such that the current unit in the decoded frame P2 The magnitude of is smoothly reduced in the frequency domain, so that the current unit in the decoded frame P2 In the frequency domain, the components of the low-frequency region are preserved and the components of the mid-frequency to high-frequency region are attenuated, thereby obtaining the current unit corresponding to the second unit In order to save the amount of calculation required in the implementation process and avoid the occurrence of ringing effects, the decoding function should make the current unit Amplitudes in the frequency domain transition smoothly.
- decoding function It may be a low-pass filter with smooth transition in any form, which is not limited in this specification.
- the decoding function is a smooth transition curve, avoiding sharp changes in the amplitude adjustment gain in the curve.
- the ringing effect refers to that in image processing, when performing spectrum adjustment processing on an image, if the selected decoding function With faster changes, the image will "ring".
- the so-called “ringing” refers to the vibration generated at the sharp change of the gray level of the output image, just like the air vibration generated after the clock is struck. The ringing effect mostly appears at the image boundary.
- the data decompression described in this specification uses a smooth transition decoding function Adjust the decoded frame P2 , filter the components of the intermediate frequency to high frequency region in the decoded frame P2 , and then process the decoded frame P2 and the decoding function
- the boundary information can be obtained by calculating the difference of the adjusted data; the boundary information is adjusted using a boundary compensation coefficient to restore it to an initial state or to enhance it relative to the initial state.
- a decoding convolution kernel can be designed so that all coefficients are non-negative numbers, or the ratio of the absolute value of the sum of negative coefficients to the sum of non-negative coefficients is less than 0.1, and ringing can be avoided the emergence of the effect.
- the adjustment can also be expressed as using the decoding convolution kernel in the time domain to check the current unit Do convolution. Therefore, there should also be a corresponding relationship between the decoding convolution kernel used for the boundary compensation and the coding convolution kernel used for the boundary adjustment. That is to say, the encoding function and decode function Corresponding relationships should also exist. That is, by selecting and encoding functions The decoding function corresponding to the encoding convolution kernel And decoding convolution kernel, two ways can achieve the same effect. For the convenience of description, this specification will take convolution in the time domain as an example to describe the boundary compensation, but those skilled in the art should understand that by multiplying the decoding function in the frequency domain The way of performing frequency spectrum adjustment is also the scope to be protected in this specification.
- a convolution kernel or a combination of convolution kernels with a higher order is required in the implementation process. This means an unnecessary increase in computation.
- high-order convolution kernels are more likely to cause strong color oscillations in the output image where the grayscale or color changes sharply, which is called the ringing effect.
- the ringing effect mostly appears at the image boundary.
- the decoding function The current unit can be analyzed in the frequency domain
- the amplitude of the middle and low frequency region is adjusted, so that the change of the amplitude adjustment gain in the middle and low frequency region is smooth and continuous.
- the ratio of the absolute value of the sum of negative coefficients to the sum of non-negative coefficients in the decoding convolution kernel is smaller than a threshold.
- the threshold may be any one of 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.1, 0.2, 0.3, 0.4 or any value in the interval defined by any two numbers.
- the convolution kernel coefficients in the decoding convolution kernel may all be selected as non-negative numbers.
- decoding function In the middle, the amplitude adjustment gain for the mid-frequency to high-frequency region is equal to 0, and can fluctuate within a certain error range.
- the error range can be 0, ⁇ 1%, ⁇ 2%, ⁇ 3%, ⁇ 4%, ⁇ 5%, ⁇ 6%, ⁇ 7%, ⁇ 8%, ⁇ 9%, ⁇ 10%, ⁇ 11% %, ⁇ 12%, ⁇ 13%, ⁇ 14%, ⁇ 15%, ⁇ 16%, ⁇ 17%, ⁇ 18%, ⁇ 19%, ⁇ 20%, ⁇ 21%, ⁇ 22%, ⁇ 23%, ⁇ 24%, ⁇ 25%, ⁇ 26%, ⁇ 27%, ⁇ 28%, ⁇ 29%, ⁇ 30%, ⁇ 31%, ⁇ 32%, ⁇ 33%, ⁇ 34%, ⁇ 35% and other values within any two specified intervals.
- the pass decode function The DC part can be kept, that is, the amplitude adjustment gain of the part where the frequency is 0 is 1, so as to ensure that the basic information in the initial frame can be preserved. Therefore, by the decoding function for the current unit When adjusting, the amplitude adjustment gain of the low frequency region is smoothly transitioned from the amplitude adjustment gain of 1 at the position where the frequency is 0 to the amplitude adjustment gain of the intermediate frequency region close to 0.
- the decoding function group may be stored in the storage medium of the data decompression device 300 .
- the set of decoding functions may include at least one different decoding function.
- Each decoding function corresponds to a decoding convolution kernel. That is to say, the storage medium of the data decompression device 300 may include at least one decoding convolution kernel.
- the storage medium of the data decompression device 300 may also store a decoding function with encoding function Correspondence between.
- the compressed frame includes the encoded data RAMI, that is, the encoding function H 1 (f) and the boundary adjustment coefficient g 1 , that is, the encoding function corresponding to each unit and boundary adjustment coefficient Data decompression device 300 for the current unit
- the encoding function H 1 (f) and the decoding function with encoding function The corresponding relationship between the current unit is selected from the preset decoding function group encoding function
- the corresponding function as the current unit Corresponding decoding function
- the corresponding convolution kernel is used as the decoding convolution kernel, and the current unit Do convolution.
- the encoded data RAMI that is, the encoding function H 1 (f) and the boundary adjustment coefficient g 1 are not included in the compressed frame.
- the data decompression device 300 performs a When performing convolution, a function can be arbitrarily selected from the preset decoding function group as the decoding function And use its corresponding convolution kernel as the decoding convolution kernel to check the current unit Do convolution.
- the data decompression device 300 can also rely on empirical values to select a decoding function from the decoding function group For example, the decoding function is selected by machine learning
- the decoding function can be encoded with the function same.
- the curve shown in Figure 6 can also be a decoding function Those skilled in the art should understand all that can make the current unit The magnitude of is smoothly reduced in the frequency domain, so that the current unit in the decoded frame P2 A decoding function in which components in the low-frequency region are preserved and components in the mid- to high-frequency region are attenuated in the frequency domain and the decoding function linear combination of or the decoding function product combination Or the combination of linear combination and product combination all belong to the protection scope of this specification.
- m ⁇ 1 Represents a linear combination of n functions
- k m represents the weight corresponding to the mth function
- q ⁇ 1 Represents the product combination of n functions
- k q represents the weight corresponding to the qth function
- the data decompression device 300 may perform steps S344-42 for each unit. per unit After decoding function After adjustment, a corresponding second unit is obtained Decoding functions corresponding to all units Combined to form the decoding function H 2 (f).
- the decoding function corresponding to each unit The decoding function H 2 (f) can be generated by combining the units according to their positions.
- the decoding function H 2 (f) can be regarded as a matrix.
- All units correspond to the second unit Combined to form the second frame P 2b .
- the second unit corresponding to each unit The second frame P 2b may be generated by combining the units according to their positions.
- the second frame P 2b is a blurred image.
- the relationship between P 2 and P 2b can be expressed as the following formula:
- step S344-4 may also include the current unit implement:
- the mid-frequency to high-frequency components in the data spectrum of each frame are mainly concentrated in the area where the data changes sharply in this frame of data, that is, the boundary data of the data.
- the intermediate-frequency to high-frequency data is mainly concentrated on the boundary of the object in the image, that is, the boundary data of this frame of image.
- the decoding function make the current unit The amplitude in the frequency domain is smoothly reduced to attenuate components in the mid to high frequency region. Therefore, the second unit It can be understood as removing the current unit The data of the boundary information in .
- the second boundary include current unit Components in the middle frequency to the high frequency region.
- the second boundary include current unit Boundary information in .
- a corresponding second boundary is obtained Second boundary for all cells Combined to form the second boundary frame E 2b .
- the second boundary corresponding to each cell The second boundary frame E 2b may be generated by combining the units according to their positions.
- the second boundary frame E 2b can be expressed as the following formula:
- step S344-4 may also include the current unit implement:
- the boundary compensation coefficient corresponding to the boundary compensation is defined as g 2 .
- the boundary adjustment can attenuate the amplitude of the strong boundary in the frequency domain from the middle frequency to the high frequency region of the strong boundary in the compressed frame, so as to blur the boundary data in the compressed frame, thereby Reduce the amount of data generated by encoding.
- the boundary compensation can restore or even enhance the data after the boundary adjustment. That is to say, the boundary compensation can completely restore the amplitude of the mid-frequency to high-frequency region in the de-framing or basically restore to the state before fading, or even enhance the state before fading.
- boundary adjustment coefficient and Boundary Compensation Coefficient there is a preset association relationship, that is, the boundary adjustment coefficient and Boundary Compensation Coefficient Corresponding.
- a set of boundary compensation coefficients may be stored in the storage medium of the data decompression device 300 .
- the set of boundary compensation coefficients may include at least one different coefficient.
- the storage medium of the data decompression device 300 may also store a boundary adjustment coefficient and Boundary Compensation Coefficient Correspondence between.
- the compressed frame includes the encoded data RAMI, that is, the encoding function H 1 (f) and the boundary adjustment coefficient g 1 , that is, the encoding function corresponding to each unit and boundary adjustment coefficient
- the data decompression device 300 executes steps S344-46, it can adjust the coefficient based on the boundary and Boundary Compensation Coefficient
- the association relationship of the current unit is selected from the preset boundary compensation coefficient group
- the boundary adjustment coefficient of Corresponding boundary compensation coefficient as the current unit Corresponding boundary compensation coefficient to the second boundary The amplitude is compensated.
- the encoded data RAMI that is, the encoding function H 1 (f) and the boundary adjustment coefficient g 1 are not included in the compressed frame.
- the data decompression device 300 executes steps S344-46, it can arbitrarily select one of the preset boundary compensation coefficient groups as the current unit Corresponding boundary compensation coefficient to the second boundary The amplitude is compensated.
- the data decompression device 300 can also rely on empirical values to select boundary compensation coefficients from the set of boundary compensation coefficients For example, the selection of boundary compensation coefficients through machine learning
- the decoding boundary corresponding to each unit can be generated by combining the units according to their positions.
- the relationship between E 2m and E 2b can be expressed as the following formula:
- step S344-4 may include the current unit implement:
- the decompressed frame P 4 can be generated by combining the units according to their positions.
- the relationship between P 4 and E 2m and P 2 can be expressed as the following formula:
- the design of the boundary adjustment only attenuates the amplitude of the intermediate-frequency to high-frequency region in the initial frame P 0 , so that the encoding adjustment
- the frequency information of the low frequencies in the initial frame P 0 is preserved in the frame P 1 .
- the decoded frame P2 is basically the same as the encoding adjustment frame P1 , therefore, the decoded frame P2 also retains low-frequency frequency information.
- the design of the boundary compensation only compensates the amplitude of the mid-frequency to high-frequency region in the decoded frame P2 .
- the frequency information of the low frequency in the original frame P 0 is preserved in the decompressed frame P 4 .
- the decompressed frame P 4 obtained by compensating the amplitude of the mid-frequency to high-frequency region in the decoded frame P 2 through the boundary compensation can be completely restored Or basically restore all the frequency information of the intermediate frequency in the initial frame P 0 . That is to say, the data decompression can restore or even enhance the compressed data at any frequency of the intermediate frequency.
- the amplitude of the decompressed frame P 4 at any frequency in the low frequency region should be approximately equal to the initial frame P 0
- the amplitude at any frequency in the intermediate frequency region should be approximately equal to or greater than the initial frame P 0 .
- Frame P 0 The approximately equal means that the magnitude of the decompressed frame P 4 is equal to the magnitude of the initial frame P 0 and fluctuates within a certain error range.
- the amplitude of the decompressed frame P4 returns to 85% or more of the original frame P0 at any frequency in the low-frequency to intermediate-frequency region, it is difficult for human eyes to perceive the amplitude of the original frame P0.
- the amplitude of the decompressed frame P4 at any frequency in the low frequency to intermediate frequency region should not be less than 85% of the initial frame P0 . That is, the error range should not make the amplitude of the decompressed frame P 4 lower than 85% of the original frame P 0 at any frequency in the low frequency to intermediate frequency region. Since the human eye is not sensitive to information in the high-frequency region, the information in the decompressed frame P 4 for the high-frequency region can be retained to adapt to high-quality scenes, and can also be attenuated to suppress unnecessary high-frequency noise.
- the amplitude of the decompressed frame P 4 at any high-frequency frequency should be approximately equal to the initial frame P 0 , or lower than the initial frame P 0 , or greater than the initial frame P 0 0 .
- the relationship between P 0 and P 4 can be expressed as the following formula:
- the formula here only lists the basic relationship formula between P 4 and P 0 , without writing the error into the formula. Those skilled in the art should understand that fluctuations within the error range make the amplitude of the decompressed frame P 4 slightly smaller than the initial frame P 0 in the low-frequency to intermediate-frequency region, which also falls within the protection scope of this specification.
- f 0 is the cut-off value of the frequency sensitive to the human eye.
- f 0 may be 0.33, or other values larger or smaller than 0.33.
- the value of f0 is different.
- the decompressed frame P 4 when H 0 (f) ⁇ 1 in the selected frequency domain interval, the decompressed frame P 4 can be in the selected frequency domain The data in the interval is restored to the initial frame P 0 ; when the selected frequency domain interval H 0 (f)>1, the data of the decompressed frame P 4 in the selected frequency domain interval can be Enhanced, that is, the amplitude of the decompressed frame P 4 in the selected area is higher than that of the original frame P 0 .
- the initial frame P 0 is a frame in the video
- H 0 (f)>1 as enhanced mode.
- FIG. 9A shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 9B shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 9C shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- FIG. 9D shows a graph of an overall adjustment function H 0 (f) provided according to an embodiment of the present specification.
- the horizontal axis is the normalized frequency f
- the vertical axis is the amplitude adjustment gain H 0 of the overall spectrum adjustment function H 0 (f).
- the curves in Figures 9A to 9D represent different overall spectral adjustment functions H 0 (f).
- the normalized frequency maximum value on the horizontal axis is 0.5.
- the normalized frequency f of the horizontal axis can be divided into a low frequency region, a middle and low frequency region, a middle frequency region, a middle and high frequency region and a high frequency region.
- the frequency between (0, a] belongs to low frequency; the frequency between (a, b] belongs to low frequency; the frequency between (b, c] belongs to intermediate frequency; the frequency between (c, d] belongs to medium and high frequency;
- the frequency between (d, 0.5] belongs to high frequency.
- the values of a, b, c, d, e are described with reference to FIG. 6, and will not be repeated here.
- the decompressed frame P4 should be kept in the low-frequency to medium-frequency region relative to the initial frame P0 . That is to say, the overall spectrum adjustment function H 0 (f) should make the amplitude of the decompressed frame P 4 in the low frequency to intermediate frequency region not less than 85% of the initial frame P 0 , and can even be greater than the Describe the initial frame P 0 . Since the human eye is not sensitive to information in the high-frequency region, the amplitude of the decompressed frame P4 in the high-frequency region can be selected according to different application scenarios.
- the decompressed frame P4 The amplitude of P 4 in the high frequency region may be smaller than the initial frame P 0 .
- the amplitude of the decompressed frame P 4 in the high-frequency region may be approximately equal to or greater than the initial frame P 0 .
- the amplitude adjustment gain H 0 of the overall adjustment function H 0 (f) at any frequency f in the low frequency to intermediate frequency region (including the low frequency and intermediate frequency regions) is greater than 1 or approximately equal to 1, so that after decompression
- the amplitude of the decompressed frame P4 is not less than 85% of the initial frame P0 , so that the clarity is restored or enhanced, and the visual observation effect is improved.
- Said approximately equal to 1 may fluctuate within a certain error range of being equal to 1 here.
- the error range can be 0, ⁇ 1%, ⁇ 2%, ⁇ 3%, ⁇ 4%, ⁇ 5%, ⁇ 6%, ⁇ 7%, ⁇ 8%, ⁇ 9%, ⁇ 10%, ⁇ 11% %, ⁇ 12%, ⁇ 13%, ⁇ 14%, ⁇ 15% and other numerical values within the range specified by any two.
- the amplitude adjustment gain of the overall adjustment function H 0 (f) in the high frequency region as the first amplitude adjustment gain
- the amplitude adjustment gain in the intermediate frequency region as the second amplitude adjustment gain.
- the amplitude adjustment gain in the low frequency region is defined as a third amplitude adjustment gain.
- the third amplitude adjustment gain value, the second amplitude adjustment gain value and the first amplitude adjustment gain value may fluctuate within the error range.
- the third amplitude adjustment gain value, the second amplitude adjustment gain value and the first amplitude adjustment gain value of the overall adjustment function H 0 (f) in the low-frequency to high-frequency region are all approximately equal to 1, so that The amplitude of the decompressed frame P4 in the low-frequency to high-frequency region is not less than 85% of the initial frame P0 , so that the data of the decompressed frame P4 in the low-frequency to high-frequency region can be restored smoothly or basically to the state of the initial frame P 0 .
- the third amplitude adjustment gain value and the second amplitude adjustment gain value of the overall adjustment function H 0 (f) in the low frequency to intermediate frequency region are approximately equal to 1, so that the decompressed frame P 4 is in the low frequency to intermediate frequency region.
- the data of the region can be restored smoothly or basically to the state of the initial frame P0 .
- the first amplitude adjustment gain value of the overall adjustment function H 0 (f) in the high-frequency region is less than 1, so that the amplitude of the decompressed frame P 4 in the high-frequency region decreases smoothly relative to the initial frame P 0 , so that Suppresses high frequency noise.
- the steady reduction of the amplitude may be that the amplitude is attenuated by the first amplitude adjustment gain value, or that the amplitude is attenuated within a certain error range around the first amplitude adjustment gain value.
- the first amplitude adjustment gain may be any value between 0 and 1.
- the first amplitude adjustment gain value can be 0, 0.04, 0.08, 0.12, 0.16, 0.20, 0.24, 0.28, 0.32, 0.36, 0.40, 0.44, 0.48, 0.52, 0.56, 0.60, 0.64, 0.68, 0.72 , 0.76, 0.80, 0.84, 0.88, 0.92, 0.96 and 1, etc. within the range specified by any two values. As shown in FIG.
- the first amplitude adjustment gain of the overall adjustment function H 0 (f) in the high-frequency region is about 0.6.
- Both the second and third amplitude adjustment gain values are around 1.
- the second and third amplitude adjustment gain values can fluctuate within a certain error range, for example, the second and third amplitude adjustment gain values can be in values such as 0.85, 0.90, 0.95, 1, 1.05, 1.10, and 1.15 within any two specified intervals.
- the third amplitude adjustment gain value of the overall adjustment function H 0 (f) in the low frequency region is approximately equal to 1, so that the data of the decompressed frame P 4 in the low frequency region can be restored smoothly or basically to the original Status of frame P 0 .
- the second amplitude adjustment gain value of the overall adjustment function H 0 (f) in the middle frequency region and the first amplitude adjustment gain value in the high frequency region are both greater than 1, so that the decompressed frame P 4 in the middle frequency to high frequency region
- the amplitude increases smoothly relative to the initial frame P 0 , resulting in enhanced data clarity in the mid-to-high frequency region.
- the steady increase of the amplitude may be that the amplitude is enhanced by the second amplitude adjustment gain value and the first amplitude adjustment gain value, or that the amplitude is increased by the second amplitude adjustment gain value and the first amplitude adjustment gain value. Enhancement is performed within a certain error range around the first amplitude adjustment gain value.
- the second amplitude adjustment gain value may be substantially the same as the first amplitude adjustment gain value, or the second amplitude adjustment gain value may be greater than the first amplitude adjustment gain value, or the The second amplitude adjustment gain value is smaller than the first amplitude adjustment gain value. In the curve shown in FIG. 9C , the second amplitude adjustment gain value is substantially the same as the first amplitude adjustment gain value.
- the second amplitude adjustment gain value and the first amplitude adjustment gain value may be any value greater than 1.
- the second amplitude adjustment gain value and the first amplitude adjustment gain value may be 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 2.1, 2.2, and 2.4 within the range specified by any two of the equal values.
- the second amplitude adjustment gain and the first amplitude adjustment gain of the overall adjustment function H 0 (f) in the mid-frequency to high-frequency region are about 1.2.
- the third amplitude adjustment gain value of the overall adjustment function H 0 (f) in the low frequency region is approximately equal to 1, so that the data of the decompressed frame P 4 in the low frequency region can be restored smoothly or basically to the original Status of frame P 0 .
- the second amplitude adjustment gain value of the overall adjustment function H 0 (f) in the intermediate frequency region is greater than 1, so that the amplitude of the decompressed frame P 4 in the intermediate frequency increases smoothly relative to the initial frame P 0 , so that the intermediate frequency region Enhanced data clarity.
- the first amplitude adjustment gain value of the overall adjustment function H 0 (f) in the high-frequency region is less than 1, so that the amplitude of the decompressed frame P 4 in the high-frequency region decreases smoothly relative to the initial frame P 0 , so that Reduce the amount of data in the insensitive high-frequency region to suppress high-frequency noise.
- the curve shown in Fig. 9D can enhance clarity while reducing the amount of data.
- the second amplitude adjustment gain value may be any value greater than 1.
- the first amplitude adjustment gain may be any value between 0 and 1. As shown in FIG. 9D , the second amplitude adjustment gain of the overall adjustment function H 0 (f) in the middle frequency region is about 1.2, and the first amplitude adjustment gain in the high frequency region is about 0.6.
- the overall spectrum adjustment function H 0 (f) can also adjust the amplitude of the high frequency region, so that the amplitude adjustment gain
- the changes in the mid-high frequency region are smooth and continuous.
- the overall spectrum adjustment function H 0 (f) can also adjust the amplitude of the mid-low frequency region, so that the amplitude adjustment gain is at The changes in the mid-low frequency region are continuous.
- the curve of the overall adjustment function H 0 (f) is a curve with a smooth transition.
- the curve of the overall adjustment function H0 (f) can be allowed to exist in a small range The fluctuation does not affect the effect of decompression.
- the parameters of the overall adjustment function H 0 (f) can be set according to the receiver's sensitivity to the data. Different forms of data have different receiver sensitivity to frequency.
- Fig. 10A shows an overall adjustment function H 0 (f), an encoding function H 1 (f), an encoding transfer function HE (f) and a decoding transfer function H D (f) in the normal mode provided by an embodiment of this specification. ) graph.
- Fig. 10B shows an overall adjustment function H 0 (f), an encoding function H 1 (f), an encoding transfer function HE (f) and a decoding transfer function HD ( f) Graph.
- the encoding convolution kernel and the decoding convolution kernel used in FIG. 10A and FIG. 10B are the same as those shown in Table 1.
- the boundary adjustment coefficient g 1 0.5
- the boundary compensation coefficient g 2 0.96.
- the boundary adjustment coefficient g 1 0.5
- the boundary compensation coefficient g 2 1.6.
- the horizontal axis is the normalized frequency f
- the vertical axis is the amplitude adjustment gain H.
- the overall spectrum adjustment function H 0 (f) ⁇ 1 in any frequency region
- the overall spectrum adjustment function H 0 (f) performs normal mode spectrum adjustment on the decompressed frame, that is, the overall spectrum adjustment function H 0 In (f)
- the information for all frequencies is completely preserved, and the data in the decompressed frame can basically be restored to the data in the initial frame.
- FIG. 10A the overall spectrum adjustment function H 0 (f) ⁇ 1 in any frequency region
- the overall spectrum adjustment function H 0 (f) performs normal mode spectrum adjustment on the decompressed frame, that is, the overall spectrum adjustment function H 0 In (f)
- the information for all frequencies is completely preserved, and the data in the decompressed frame can basically be restored to the data in the initial frame.
- the overall spectrum adjustment function H 0 (f) performs enhanced mode spectrum adjustment on the intermediate frequency to high frequency region of the decompressed frame, that is, the overall spectrum adjustment function H 0 (f) enhances the information in the intermediate frequency to high frequency region, so The data in the mid-frequency to high-frequency region in the decompressed frame is enhanced compared with the data in the mid-frequency to high-frequency region in the initial frame. It should be noted that the curves shown in Fig. 10A and Fig.
- H 0 (f), H 1 (f), HE (f) and HD (f) The curves are not limited to the form shown in Figure 10A and Figure 10B, all H 0 (f), H 1 (f), H E (f) and HD (f ) curves belong to the protection scope of this specification.
- the data compression device 200 executes the method P200 to divide the initial frame in the initial video data into multiple units, and Acquire the amplitude of the intermediate frequency to high frequency region of each unit, and use different boundary adjustment coefficients to adjust the amplitude of the intermediate frequency to high frequency region in each unit, so as to reduce the initial frame.
- the amplitude of the mid-frequency to high-frequency region in the current unit is large, it means that the current unit contains a strong boundary, then use the boundary adjustment coefficient less than 1 and greater than 0 to adjust the amplitude of the mid-frequency to high-frequency region of the current unit , to reduce the amplitude of the mid-frequency to high-frequency region of the current unit, thereby reducing the signal strength in the mid-frequency to high-frequency region of the current unit, thereby reducing the amount of data information, and improving the data when making predictions and calculating residuals Compression efficiency.
- the amplitude of the mid-frequency to high-frequency region in the current unit is small, it means that the current unit contains a weak boundary, then use a boundary adjustment coefficient greater than 1 to adjust the amplitude of the mid-frequency to high-frequency region of the current unit to enhance The amplitude of the mid-frequency to high-frequency region of the current unit to avoid the loss of weak boundaries in the current unit during data compression (prediction and residual) to avoid loss of details.
- the data processing method P200 and system 100 can enhance the data information volume of weak boundaries while improving the efficiency of data compression, so as to avoid loss of details during the data compression process, that is, reduce data distortion while improving data compression efficiency .
- the data processing system 100 when decompressing the compressed frame, executes the method P300 through the data decompression device 300, uses the unit during data compression as the data decompression unit, and uses the boundary adjustment coefficient for each unit
- the corresponding boundary compensation coefficient performs boundary compensation on the amplitude of the mid-frequency to high-frequency region, so as to compensate the amplitude of the mid-frequency to high-frequency region reduced during the data compression process, and obtain the decompressed frame.
- the boundary compensation corresponds to the boundary adjustment, and there is a corresponding relationship between the boundary compensation coefficient and the boundary adjustment coefficient.
- the boundary compensation can restore the compressed data after the boundary adjustment to a resolution of the original frame that is even higher than that of the original frame.
- the decoding end can at least restore the data in the important frequency of the decompressed data to the definition of the original frame, and even obtain the definition beyond the original frame. Since the boundary adjustment coefficients of the initial frame in the boundary adjustment process are all greater than 0, the information in the compressed frame is not missing, so the boundary adjustment can be designed according to the relationship between the boundary adjustment coefficient and the boundary compensation coefficient and their respective characteristics Coefficients and boundary compensation coefficients to restore the information in the compressed frame.
- the method and system can significantly improve data compression efficiency, improve data transmission efficiency, reduce data loss, avoid detail loss, eliminate noise, and improve the clarity of decompressed data.
- This specification also provides a non-transitory storage medium, which stores at least one set of executable instructions for data processing.
- the executable instructions When the executable instructions are executed by a processor, the executable instructions instruct the processor to implement The steps of the data processing method P200.
- various aspects of this specification can also be implemented in the form of a program product, which includes program codes.
- the program product runs on the data compression device 200, the program code is used to make the data compression device 200 execute the steps of data processing described in this specification.
- the program product for realizing the above method can adopt a portable compact disk read only memory (CD-ROM) and include program codes, and can run on the data compression device 200, such as a personal computer.
- CD-ROM portable compact disk read only memory
- the program product in this specification is not limited thereto.
- a readable storage medium may be any tangible medium containing or storing a program, and the program may be used or combined with an instruction execution system (such as the compression end processor 220) use.
- the program product may reside on any combination of one or more readable media.
- the readable medium may be a readable signal medium or a readable storage medium.
- the readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof.
- readable storage media include: electrical connections with one or more conductors, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
- the computer readable storage medium may include a data signal carrying readable program code in baseband or as part of a carrier wave traveling as a data signal. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.
- a readable storage medium may also be any readable medium other than a readable storage medium that can send, propagate or transport a program for use by or in conjunction with an instruction execution system, apparatus or device.
- the program code contained on the readable storage medium may be transmitted by any suitable medium, including but not limited to wireless, cable, optical cable, RF, etc., or any suitable combination of the above.
- Program code for performing the operations of this specification may be written in any combination of one or more programming languages, including object-oriented programming languages—such as Java, C++, etc., as well as conventional procedural programming languages. Programming language - such as "C" or a similar programming language.
- the program code may execute entirely on the data compression device 200, partly on the data compression device 200, as a stand-alone software package, partly on the data compression device 200 and partly on a remote computing device, or entirely on a remote computer. executed on a computing device.
- the remote computing device may be connected to data compression device 200 through transmission medium 120, or may be connected to an external computing device.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Description
Claims (24)
- 一种数据处理的方法,其特征在于,包括:选择初始数据中的初始帧,所述初始帧包括预设字节数的初始数据;以及对所述初始帧进行数据压缩,得到压缩帧,其中,所述数据压缩包括对在压帧进行边界调节,所述在压帧包括所述初始帧和所述初始帧在所述数据压缩过程中成为所述压缩帧之前的任一数据状态,其中,所述边界调节包括对所述在压帧的多个单元中的每个单元使用与其对应的边界调节系数对其在中频至高频区域的幅值进行调节,以降低所述在压帧在所述中频至所述高频区域的幅值,所述边界调节系数大于0。
- 如权利要求1所述的数据处理的方法,其特征在于,所述对在压帧进行边界调节,包括:基于预设的单元尺寸将所述在压帧划分为所述多个单元;以及对所述每个单元使用与其对应的所述边界调节系数对其在中频至高频区域的幅值进行调节。
- 如权利要求2所述的数据处理的方法,其特征在于,所述对所述每个单元使用与其对应的所述边界调节系数对其在中频至高频区域的幅值进行调节,包括对所述每个单元:从预设的编码函数组中选取一个函数作为编码函数,通过所述编码函数对其进行调节,得到第一单元,使其在频域内的低频区域的分量被保留而中频至高频区域的分量被衰减;对其和所述第一单元求差,获取其对应的第一边界,所述第一边界包括其在所述中频至所述高频区域的分量;以及使用与其对应的所述边界调节系数对所述第一边界的幅值进行调节,得到其对应的编码边界;以及将所述第一单元与所述编码边界进行叠加。
- 如权利要求3所述的数据处理的方法,其特征在于,所述使用与其对应的所述边界调节系数对所述第一边界的幅值进行调节,包括:确定所述第一边界的边界值小于预设第一阈值,通过大于1的所述边界调节系数增强所述第一边界的幅值;或者确定所述第一边界的边界值大于预设第二阈值,通过小于1的所述边界调节系数降低所述第一边界的幅值。
- 如权利要求4所述的数据处理的方法,其特征在于,所述通过大于1的所述边界调节系数增强所述第一边界的幅值,包括:从预设的第一边界调节系数组中选取一个系数作为所述边界调节系数,增强所述第一边界的幅值,所述第一边界调节系数组中的系数均大于1;所述通过小于1的所述边界调节系数降低所述第一边界的幅值,包括:从预设的第二边界调节系数组中选取一个系数作为所述边界调节系数,降低所述第一边界的幅值,所述第二边界调节系数组中的系数均小于1。
- 如权利要求3所述的数据处理的方法,其特征在于,所述使用与其对应的所述边界调节系数对所述第一边界的幅值进行调节,包括:以失真率和码率的加权值为优化目标,基于优化算法,获取所述边界调节系数,以所述边界调节系数对所述第一边界的幅值进行调节。
- 如权利要求3所述的数据处理的方法,其特征在于,所述对所述初 始帧进行数据压缩,包括以下方式中的至少一种:对所述初始帧先进行所述边界调节,再对所述边界调节后的初始帧进行预测和求残差;对所述初始帧先进行预测得到预测帧,再对所述初始帧和所述预测帧进行所述边界调节和求残差;以及对所述初始帧先进行预测和求残差,再对所述残差进行所述边界调节。
- 如权利要求7所述的数据处理的方法,其特征在于,所述压缩帧还包括所述多个单元中的每个单元对应的所述编码函数和所述边界调节系数。
- 一种数据处理的系统,其特征在于,包括:至少一个存储介质,存储有至少一个指令集,用于数据处理;以及至少一个处理器,同所述至少一个存储介质通讯连接,其中,当所述系统运行时,所述至少一个处理器读取所述至少一个指令集,并且根据所述至少一个指令集的指示执行权利要求1-8中任一项所述的数据处理的方法。
- 一种数据处理的方法,其特征在于,包括:获取压缩数据,所述压缩数据包括对初始帧进行数据压缩得到的压缩帧,所述数据压缩包括边界调节;以及对所述压缩帧进行数据解压,得到解压帧,所述数据解压包括对在解帧进行边界补偿,所述在解帧包括所述压缩帧和所述压缩帧在所述数据解压过程中成为所述解压帧前的任一数据状态,其中,所述边界补偿与所述边界调节存在预先设定的关联关系。
- 如权利要求10所述的数据处理的方法,其特征在于,所述边界调 节包括对在压帧的多个单元中的每个单元使用与其对应的边界调节系数对其在中频至高频区域的幅值进行调节,以降低所述在压帧在所述中频至所述高频区域的幅值,所述边界调节系数大于0,所述在压帧包括所述初始帧和所述初始帧在所述数据压缩过程中成为所述压缩帧之前的任一数据状态,所述边界补偿包括对所述在解帧的所述多个单元中的所述每个单元,基于所述关联关系,使用与所述边界调节系数相对应的边界补偿系数对其在中频至高频区域的幅值进行补偿。
- 如权利要求11所述的数据处理的方法,其特征在于,所述对在压帧进行边界调节,包括:基于预设的单元尺寸将所述在压帧划分为所述多个单元;以及对所述每个单元使用与其对应的所述边界调节系数对其在中频至高频区域的幅值进行调节,包括对所述每个单元:从预设的编码函数组中选取一个函数作为编码函数,通过所述编码函数对其进行调节,得到第一单元,使其在频域内的低频区域的分量被保留而中频至高频区域的分量被衰减;对其和所述第一单元求差,获取其对应的第一边界,所述第一边界包括其在所述中频至所述高频区域的分量;以及使用与其对应的所述边界调节系数对所述第一边界的幅值进行调节,得到其对应的编码边界;以及将所述第一单元与所述编码边界进行叠加。
- 如权利要求12所述的数据处理的方法,其特征在于,所述对在解帧进行边界补偿,包括:基于预设的所述单元尺寸将所述在解帧划分为所述多个单元;以及对所述每个单元使用与所述边界调节系数相对应的所述边界补偿系数 对其在中频至高频区域的幅值进行补偿。
- 如权利要求13所述的数据处理的方法,其特征在于,所述对所述每个单元使用与所述边界调节系数相对应的所述边界补偿系数对其在中频至高频区域的幅值进行补偿,包括对所述每个单元:确定解码函数,通过所述解码函数对其进行调节,得到第二单元,使其在频域内的低频区域的分量被保留而中频至高频区域的分量被衰减;对其和所述第二单元求差,获取其对应的第二边界,所述第二边界包括其在所述中频至所述高频区域的分量;以及使用与所述边界调节系数相对应的所述边界补偿系数对所述第二边界的幅值进行补偿,得到其对应的解码边界;以及将所述当前单元与所述解码边界进行叠加。
- 如权利要求14所述的数据处理的方法,其特征在于,所述确定解码函数,包括:从预设的解码函数组中选取一个函数作为所述解码函数。
- 如权利要求14所述的数据处理的方法,其特征在于,所述使用与所述边界调节系数相对应的所述边界补偿系数对所述第二边界的幅值进行补偿,包括:从预设的边界补偿系数组中选取一个作为所述边界补偿系数,对所述第二边界的幅值进行补偿。
- 如权利要求14所述的数据处理的方法,其特征在于,所述压缩帧包括:所述在压帧中的多个单元中的每个单元对应的所述编码函数和所述边 界调节系数。
- 如权利要求17所述的数据处理的方法,其特征在于,所述确定解码函数,包括:从预设的解码函数组中选取与所述编码函数相对应的函数作为所述解码函数。
- 如权利要求17所述的数据处理的方法,其特征在于,所述使用与所述边界调节系数相对应的所述边界补偿系数对所述第二边界的幅值进行补偿,包括:基于所述边界调节系数与所述边界补偿系数的关联关系,确定所述边界补偿系数,对所述第二边界的幅值进行补偿。
- 如权利要求10所述的数据处理的方法,其特征在于,所述对所述压缩帧进行数据解压,包括以下方式中的至少一种:对所述压缩帧先进行解码,再进行所述边界补偿;对所述压缩帧进行所述解码过程中进行所述边界补偿;以及对所述压缩帧先进行所述边界补偿,再进行所述解码。
- 如权利要求10所述的数据处理的方法,其特征在于,所述关联关系包括:所述边界补偿使所述解压帧在低频至中频区域的任意频率上的幅值不小于所述初始帧的85%。
- 如权利要求21所述的数据处理的方法,其特征在于,所述关联关系还包括:所述边界补偿使所述解压帧相对于所述初始帧在中频区域的幅值平稳地增加。
- 如权利要求21所述的数据处理的方法,其特征在于,所述关联关系还包括:所述边界补偿使所述解压帧相对于所述初始帧在高频区域的幅值平稳地降低。
- 一种数据处理的系统,其特征在于,包括:至少一个存储介质,存储有至少一个指令集,用于数据处理;以及至少一个处理器,同所述至少一个存储介质通讯连接,其中,当所述系统运行时,所述至少一个处理器读取所述至少一个指令集,并且根据所述至少一个指令集的指示执行如权利要求10-23中任一项所述的数据处理的方法。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2023568177A JP2024517851A (ja) | 2021-05-28 | 2022-05-20 | データ処理方法及びシステム |
EP22810462.6A EP4329304A1 (en) | 2021-05-28 | 2022-05-20 | Data processing method and system |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110594135.5A CN115412730A (zh) | 2021-05-28 | 2021-05-28 | 数据处理的方法和系统 |
CN202110594135.5 | 2021-05-28 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022247735A1 true WO2022247735A1 (zh) | 2022-12-01 |
Family
ID=84156379
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/094004 WO2022247735A1 (zh) | 2021-05-28 | 2022-05-20 | 数据处理的方法和系统 |
Country Status (4)
Country | Link |
---|---|
EP (1) | EP4329304A1 (zh) |
JP (1) | JP2024517851A (zh) |
CN (1) | CN115412730A (zh) |
WO (1) | WO2022247735A1 (zh) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102158701A (zh) * | 2011-04-19 | 2011-08-17 | 湖南大学 | 一种基于压缩感知理论的分类量化图像编码方法 |
CN106612439A (zh) * | 2016-02-04 | 2017-05-03 | 四川用联信息技术有限公司 | 一种自适应快速分形图像压缩方法 |
CN113518227A (zh) * | 2020-04-09 | 2021-10-19 | 于江鸿 | 数据处理的方法和系统 |
CN113766239A (zh) * | 2020-06-05 | 2021-12-07 | 于江鸿 | 数据处理的方法和系统 |
CN114079472A (zh) * | 2020-08-20 | 2022-02-22 | 于江鸿 | 数据处理的方法和系统 |
-
2021
- 2021-05-28 CN CN202110594135.5A patent/CN115412730A/zh active Pending
-
2022
- 2022-05-20 JP JP2023568177A patent/JP2024517851A/ja active Pending
- 2022-05-20 EP EP22810462.6A patent/EP4329304A1/en active Pending
- 2022-05-20 WO PCT/CN2022/094004 patent/WO2022247735A1/zh active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102158701A (zh) * | 2011-04-19 | 2011-08-17 | 湖南大学 | 一种基于压缩感知理论的分类量化图像编码方法 |
CN106612439A (zh) * | 2016-02-04 | 2017-05-03 | 四川用联信息技术有限公司 | 一种自适应快速分形图像压缩方法 |
CN113518227A (zh) * | 2020-04-09 | 2021-10-19 | 于江鸿 | 数据处理的方法和系统 |
CN113766239A (zh) * | 2020-06-05 | 2021-12-07 | 于江鸿 | 数据处理的方法和系统 |
CN114079472A (zh) * | 2020-08-20 | 2022-02-22 | 于江鸿 | 数据处理的方法和系统 |
Also Published As
Publication number | Publication date |
---|---|
JP2024517851A (ja) | 2024-04-23 |
CN115412730A (zh) | 2022-11-29 |
EP4329304A1 (en) | 2024-02-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20210067785A1 (en) | Video encoding rate control for intra and scene change frames using machine learning | |
WO2021204015A1 (zh) | 数据处理的方法和系统 | |
US10645383B2 (en) | Constrained directional enhancement filter selection for video coding | |
US10425641B2 (en) | Quantization offset and cost factor modification for video encoding | |
CN106664412B (zh) | 包括目标比特率的视频编码速率控制和质量控制 | |
US11418789B2 (en) | Content adaptive quantization for video coding | |
US20210144377A1 (en) | Method and system of video coding with content adaptive quantization | |
US11503306B2 (en) | Image and video data processing method and system | |
WO2022247735A1 (zh) | 数据处理的方法和系统 | |
US20210321119A1 (en) | Image and video data processing method and system | |
US20220078417A1 (en) | Image and video data processing method and system | |
WO2022037498A1 (zh) | 数据处理的方法和系统 | |
US8526506B1 (en) | System and method for transcoding with quality enhancement | |
CN114079472A (zh) | 数据处理的方法和系统 | |
US20220272325A1 (en) | Image and video data processing method and system | |
NL2029548A (en) | Determining adaptive quantization matrices using machine learning for video coding | |
CN113766239B (zh) | 数据处理的方法和系统 | |
JPH09130648A (ja) | 動画像信号処理装置 | |
EP3242476A1 (en) | Histogram segmentation based local adaptive filter for video encoding and decoding | |
CN113766239A (zh) | 数据处理的方法和系统 | |
CN114727110A (zh) | 数据处理的方法和系统 | |
US20210306640A1 (en) | Fine grain lookahead enhancement for video coding | |
JPH07274165A (ja) | 画像情報圧縮装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22810462 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2023568177 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022810462 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2022810462 Country of ref document: EP Effective date: 20231124 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |