Single model based variable bit rate learned image and video compression using adaptive quantization offset
By using an adaptive quantization offset method and a neural network model to obtain the quantization offset, the problem of insufficient compression efficiency and quality in existing video coding systems is solved, and more efficient video signal compression and decoding is achieved.
Patent Information
- Application Number
- CN202480066974.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-10-18
- Filing Date
- 2024-10-16
- Publication Date
- 2026-05-15
AI Technical Summary
Existing video coding systems struggle to effectively utilize quantization offset parameters for end-to-end learning during compression, resulting in insufficient compression efficiency and quality.
An adaptive quantization offset method is adopted, which obtains the quantization offset through a neural network model and calculates the distribution standard deviation of latent variables using a lookup table or neural network to obtain and reconstruct the quantization index, thereby improving the accuracy and efficiency of the quantization process.
It improves compression efficiency and quality in video encoding and decoding processes, enhances the flexibility and adaptability of quantization processes, and improves the storage and transmission performance of video signals.
Smart Images

Figure CN122055964A_ABST
Abstract
Description
Cross-reference to related applications
[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 544,698, filed October 18, 2023, the disclosure of which is incorporated herein by reference in its entirety. Background Technology
[0002] Video coding systems can be used to compress digital video signals, for example, to reduce the storage and / or transmission bandwidth required for such signals. Video coding systems can include, for example, block-based, wavelet-based, neural network (NN)-based, and / or object-based systems. Summary of the Invention
[0003] This document discloses systems, methods, and means for performing quantization processes within image or video compression using additional parameters (e.g., quantization offsets) to perform end-to-end learning. A video encoding and / or decoding device may include a processor. The device may be configured to acquire latent variables associated with a neural network (NN) model. The device may acquire a quantization offset associated with the latent variables. The device may reconstruct the quantized latent variables associated with the NN model based on the quantization offset.
[0004] The device may include one or more of the following features. The device may further reconstruct the quantized latent variables associated with the NN model based on a quantization step size. The device may obtain a quantization index based on the latent variable and the quantization step size. The device may further reconstruct the quantized latent variables associated with the NN model based on the quantization index. The quantization index may be an integer value. The NN may be trained without the quantization offset. The quantization offset associated with the latent variable may be obtained based on a lookup table. The quantization offset associated with the latent variable may be obtained using a second neural network. The quantization offset associated with the latent variable may be obtained based on the estimated standard deviation of the distribution associated with the latent variable.
[0005] A computer program product, stored on a non-transitory computer-readable medium, includes program code instructions that, when executed by a processor, implement the steps described herein. Video data, including information representing quantized latent variables, can be decoded / encoded according to the methods described herein.
[0006] This paper discloses systems, methods, and means for performing end-to-end learning within image or video compression using additional parameters (e.g., quantization offset).
[0007] A video decoding device can acquire quantized latent variables associated with a neural network (NN) model and can acquire quantization offsets associated with those latent variables. The video decoding device can reconstruct the latent variables associated with the NN model based on the quantization offsets.
[0008] The quantization offset associated with the latent variable can be obtained based on a lookup table and / or using a neural network. The quantization offset associated with the latent variable can also be calculated based on the estimated standard deviation of the latent variable's distribution.
[0009] A video coding device can acquire latent variables associated with a neural network (NN) model and can acquire quantization offsets associated with those latent variables. The video coding device can encode the latent variables associated with the NN model based on the quantization offsets.
[0010] The systems, methods, and means described herein may relate to a decoder. In some examples, the systems, methods, and means described herein may relate to an encoder. In some examples, the systems, methods, and means described herein may relate to a signal (e.g., a signal from an encoder and / or received by a decoder). A computer-readable medium may include instructions for causing one or more processors to perform the methods described herein. A computer program product may include instructions that, when executed by one or more processors, cause the one or more processors to perform the methods described herein. Attached Figure Description
[0011] Figure 1A This is a system diagram illustrating an exemplary communication system that can implement one or more of the disclosed embodiments.
[0012] Figure 1B The diagram can be found Figure 1A The illustrated system diagram shows an exemplary wireless transmit / receive unit (WTRU) used within a communication system.
[0013] Figure 1C The diagram can be found Figure 1A The diagram illustrates a system diagram of an exemplary radio access network (RAN) and an exemplary core network (CN) used within a communication system.
[0014] Figure 1D The diagram can be found Figure 1A The illustrated system diagram shows another exemplary RAN and another exemplary CN used within the communication system.
[0015] Figure 2 An exemplary video encoder is illustrated.
[0016] Figure 3 An exemplary video decoder is illustrated.
[0017] Figure 4 The illustration shows an example of a system that can achieve multiple aspects and examples.
[0018] Figure 5 An exemplary autoencoder-based compression architecture is illustrated.
[0019] Figure 6 The illustration shows an exemplary training of the total decomposition prior model.
[0020] Figure 7 An exemplary autoencoder employing a super-prior scheme is illustrated.
[0021] Figure 8 An exemplary comparison of the compression performance of several variable rate methods is illustrated.
[0022] Figure 9 The illustration shows an exemplary hyper-prior architecture autoencoder that utilizes quantization offset.
[0023] Figure 10 An exemplary comparison of the compression performance of several variable rate methods is illustrated. Detailed Implementation
[0024] A more detailed understanding can be obtained from the following description, which is given by way of example in conjunction with the accompanying drawings.
[0025] Figure 1A This is an illustration of an exemplary communication system 100 that can implement one or more of the disclosed embodiments. The communication system 100 can be a multiple access system that provides content such as voice, data, video, messaging services, and broadcasting to multiple wireless users. The communication system 100 enables multiple wireless users to access such content by sharing system resources, including wireless bandwidth. For example, the communication system 100 can employ one or more channel access methods, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal FDMA (OFDMA), Single Carrier FDMA (SC-FDMA), Zero-Tail Unique Word DFT-Spread Spectrum OFDM (ZT UW DTS-s OFDM), Unique Word OFDM (UW-OFDM), Resource Block Filtered OFDM, Filter Bank Multicarrier (FBMC), etc.
[0026] like Figure 1AAs shown, the communication system 100 may include wireless transmit / receive units (WTRUs) 102a, 102b, 102c, 102d, RAN 104 / 113, CN 106 / 115, public switched telephone network (PSTN) 108, Internet 110, and other networks 112. However, it should be understood that the disclosed embodiments contemplate any number of WTRUs, base stations, networks, and / or network elements. Each of the WTRUs 102a, 102b, 102c, 102d may be any type of device configured to operate and / or communicate in a wireless environment. For example, WTRUs 102a, 102b, 102c, and 102d (any of which may be referred to as a “station” and / or “STA”) may be configured to transmit and / or receive wireless signals and may include user equipment (UE), mobile stations, fixed or mobile user units, subscription-based units, pagers, cellular phones, personal digital assistants (PDAs), smartphones, laptops, netbooks, personal computers, wireless sensors, hotspots or Mi-Fi devices, Internet of Things (IoT) devices, watches or other wearable devices, head-mounted displays (HMDs), vehicles, drones, medical devices and applications (e.g., remote surgery), industrial devices and applications (e.g., robots and / or other wireless devices operating in the context of industrial and / or automated processing chains), consumer electronics devices, devices operating on commercial and / or industrial wireless networks, etc. Any of WTRUs 102a, 102b, 102c, and 102d may be interchangeably referred to as a UE.
[0027] The communication system 100 may also include base station 114a and / or base station 114b. Each of base stations 114a and 114b may be any type of device configured to wirelessly interface with at least one of WTRUs 102a, 102b, 102c, and 102d to facilitate access to one or more communication networks (such as CN 106 / 115, Internet 110, and / or other networks 112). For example, base stations 114a and 114b may be base transceiver stations (BTS), Node-B, eNode B, home Node B, home eNode B, gNB, NR Node B, site controller, access point (AP), wireless router, etc. Although base stations 114a and 114b are each depicted as a single element, it should be understood that base stations 114a and 114b may include any number of interconnected base stations and / or network elements.
[0028] Base station 114a may be part of RAN 104 / 113, which may also include other base stations and / or network elements (not shown), such as base station controllers (BSCs), radio network controllers (RNCs), relay nodes, etc. Base station 114a and / or base station 114b may be configured to transmit and / or receive radio signals on one or more carrier frequencies, which may be referred to as cells (not shown). These frequencies may be in licensed spectrum, unlicensed spectrum, or a combination of licensed and unlicensed spectrum. A cell may provide radio service coverage for a specific geographic area that may be relatively fixed or may vary over time. The cell may be further divided into cell sectors. For example, the cell associated with base station 114a may be divided into three sectors. Therefore, in one embodiment, base station 114a may include three transceivers, i.e., one transceiver for each sector of the cell. In embodiments, base station 114a may employ multiple-input multiple-output (MIMO) technology and may utilize multiple transceivers for each sector of the cell. For example, beamforming may be used to transmit and / or receive signals in a desired spatial direction.
[0029] Base stations 114a and 114b can communicate with one or more of WTRUs 102a, 102b, 102c, and 102d via air interface 116, which can be any suitable wireless communication link (e.g., radio frequency (RF), microwave, centimeter wave, micrometer wave, infrared (IR), ultraviolet (UV), visible light, etc.). Any suitable radio access technology (RAT) can be used to establish air interface 116.
[0030] More specifically, as described above, the communication system 100 can be a multiple access system and can employ one or more channel access schemes, such as CDMA, TDMA, FDMA, OFDMA, SC-FDMA, etc. For example, base stations 114a and WTRUs 102a, 102b, and 102c in RAN 104 / 113 can implement radio technologies such as Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access (UTRA), which can use Wideband CDMA (WCDMA) to establish air interfaces 115 / 116 / 117. WCDMA can include communication protocols such as High-Speed Packet Access (HSPA) and / or evolved HSPA (HSPA+). HSPA can include High-Speed Downlink (DL) Packet Access (HSDPA) and / or High-Speed UL Packet Access (HSUPA).
[0031] In the embodiment, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as evolved UMTS terrestrial radio access (E-UTRA), which can use Long Term Evolution (LTE) and / or LTE-Advanced (LTE-A) and / or LTE-Advanced Pro (LTE-A Pro) to establish air interface 116.
[0032] In the embodiment, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as NR radio access, which can use New Radio (NR) to establish air interface 116.
[0033] In the embodiments, base station 114a and WTRUs 102a, 102b, and 102c can implement multiple radio access technologies. For example, base station 114a and WTRUs 102a, 102b, and 102c can simultaneously implement LTE radio access and NR radio access, for example, using the dual connectivity (DC) principle. Therefore, the air interface utilized by WTRUs 102a, 102b, and 102c can be characterized by multiple types of radio access technologies and / or transmissions sent to / from multiple types of base stations (e.g., eNBs and gNBs).
[0034] In other embodiments, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as IEEE 802.11 (i.e., Wi-Fi), IEEE 802.16 (i.e., WiMAX), CDMA2000, CDMA2000 1X, CDMA2000 EV-DO, Provisional Standard 2000 (IS-2000), Provisional Standard 95 (IS-95), Provisional Standard 856 (IS-856), Global System for Mobile Communications (GSM), Enhanced Data Rate for GSM Evolution (EDGE), GSM EDGE (GERAN), etc.
[0035] Figure 1ABase station 114b can be, for example, a wireless router, a home Node B, a home eNode B, or an access point, and can utilize any suitable RAT to facilitate wireless connectivity in local areas such as business premises, homes, vehicles, campuses, industrial facilities, air corridors (e.g., for use by drones), roads, etc. In one embodiment, base station 114b and WTRUs 102c, 102d can implement radio technologies such as IEEE 802.11 to establish a wireless local area network (WLAN). In another embodiment, base station 114b and WTRUs 102c, 102d can implement radio technologies such as IEEE 802.15 to establish a wireless personal area network (WPAN). In yet another embodiment, base station 114b and WTRUs 102c, 102d can utilize cellular-based RATs (e.g., WCDMA, CDMA2000, GSM, LTE, LTE-A, LTE-A Pro, NR, etc.) to establish picocells or femtocells. Figure 1A As shown, base station 114b can have a direct connection to Internet 110. Therefore, base station 114b does not need to access Internet 110 via CN 106 / 115.
[0036] RAN 104 / 113 can communicate with CN 106 / 115, which can be any type of network configured to provide voice, data, application, and / or Voice over Internet Protocol (VoIP) services to one or more of WTRU 102a, 102b, 102c, and 102d. Data can have different Quality of Service (QoS) requirements, such as different throughput requirements, latency requirements, error tolerance requirements, reliability requirements, data throughput requirements, mobility requirements, etc. CN 106 / 115 can provide call control, billing services, location-based services, prepaid calling, internet connectivity, video distribution, etc., and / or perform high-level security functions, such as user authentication. Although in Figure 1A Although not shown, it should be understood that RAN 104 / 113 and / or CN106 / 115 can communicate directly or indirectly with other RANs that use the same RAT as or a different RAT than RAN 104 / 113. For example, in addition to being connected to RAN 104 / 113, which can utilize NR radio technology, CN 106 / 115 can also communicate with another RAN (not shown) that uses GSM, UMTS, CDMA 2000, WiMAX, E-UTRA, or WiFi radio technology.
[0037] CN 106 / 115 can also serve as a gateway for WTRU 102a, 102b, 102c, 102d to access PSTN 108, the Internet 110, and / or other networks 112. PSTN 108 may include a circuit-switched telephone network providing Common Old-Style Telephone Service (POTS). The Internet 110 may include a global system of interconnected computer networks and devices using common communication protocols such as Transmission Control Protocol (TCP), User Datagram Protocol (UDP), and / or Internet Protocol (IP) from the TCP / IP Internet Protocol suite. Network 112 may include wired and / or wireless communication networks owned and / or operated by other service providers. For example, network 112 may include another CN connected to one or more RANs, which may use the same RAT as RAN 104 / 113 or a different RAT.
[0038] Some or all of the WTRUs 102a, 102b, 102c, and 102d in communication system 100 may include multi-mode capability (e.g., WTRUs 102a, 102b, 102c, and 102d may include multiple transceivers for communicating with different wireless networks via different wireless links). For example, Figure 1A The WTRU 102c shown can be configured to communicate with base station 114a, which can employ cellular-based radio technology, and with base station 114b, which can employ IEEE 802 radio technology.
[0039] Figure 1B This is a system diagram illustrating an exemplary WTRU 102. (Example:) Figure 1B As shown, WTRU 102 may include a processor 118, a transceiver 120, a transmit / receive element 122, a speaker / microphone 124, a keyboard 126, a display / touchpad 128, non-removable memory 130, removable memory 132, a power supply 134, a Global Positioning System (GPS) chipset 136, and / or other peripheral devices 138, etc. It should be understood that WTRU 102 may include any sub-combination of the foregoing elements while remaining consistent with the embodiments.
[0040] Processor 118 may be a general-purpose processor, a special-purpose processor, a conventional processor, a digital signal processor (DSP), multiple microprocessors, one or more microprocessors associated with a DSP core, a controller, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) circuit, any other type of integrated circuit (IC), a state machine, etc. Processor 118 may perform signal encoding, data processing, power control, input / output processing, and / or any other functions that enable WTRU 102 to operate in a wireless environment. Processor 118 may be coupled to transceiver 120, which may be coupled to transmitting / receiving element 122. Although Figure 1B While the processor 118 and transceiver 120 are depicted as separate components, it should be understood that the processor 118 and transceiver 120 may be integrated together in an electronic package or chip.
[0041] Transmitting / receiving element 122 can be configured to transmit signals to or receive signals from a base station (e.g., base station 114a) via air interface 116. For example, in one embodiment, transmitting / receiving element 122 can be an antenna configured to transmit and / or receive RF signals. In another embodiment, transmitting / receiving element 122 can be a transmitter / detector configured to transmit and / or receive, for example, IR, UV, or visible light signals. In yet another embodiment, transmitting / receiving element 122 can be configured to transmit and / or receive both RF signals and optical signals. It should be understood that transmitting / receiving element 122 can be configured to transmit and / or receive any combination of wireless signals.
[0042] Although the transmitting / receiving element 122 is in Figure 1B While depicted as a single element, WTRU 102 may include any number of transmit / receive elements 122. More specifically, WTRU 102 may employ MIMO technology. Thus, in one embodiment, WTRU 102 may include two or more transmit / receive elements 122 (e.g., multiple antennas) for transmitting and receiving wireless signals via air interface 116.
[0043] Transceiver 120 can be configured to modulate signals transmitted by transmitting / receiving element 122 and demodulate signals received by transmitting / receiving element 122. As described above, WTRU 102 can have multimode capability. Therefore, transceiver 120 can include multiple transceivers to enable WTRU 102 to communicate via various RATs such as NR and IEEE 802.11.
[0044] The processor 118 of WTRU 102 can be coupled to a speaker / microphone 124, a keyboard 126, and / or a display / touchpad 128 (e.g., a liquid crystal display (LCD) unit or an organic light-emitting diode (OLED) display unit) and can receive user input data therefrom. The processor 118 can also output user data to the speaker / microphone 124, keyboard 126, and / or display / touchpad 128. Furthermore, the processor 118 can access information from any type of suitable memory (such as non-removable memory 130 and / or removable memory 132) and store data in that memory. Non-removable memory 130 may include random access memory (RAM), read-only memory (ROM), a hard disk, or any other type of memory storage device. Removable memory 132 may include a user identity module (SIM) card, memory stick, secure digital storage (SD) card, etc. In other embodiments, the processor 118 can access information from memory that is not physically located on WTRU 102 (such as on a server or home computer (not shown)) and store data in that memory.
[0045] The processor 118 can receive power from the power supply 134 and can be configured to distribute power to and / or control other components in the WTRU 102. The power supply 134 can be any suitable device for powering the WTRU 102. For example, the power supply 134 may include one or more dry cell batteries (e.g., nickel-cadmium (NiCd), nickel-zinc (NiZn), nickel metal hydride (NiMH), lithium-ion (Li-ion), etc.), solar cells, fuel cells, etc.
[0046] The processor 118 may also be coupled to a GPS chipset 136, which may be configured to provide location information (e.g., longitude and latitude) about the current location of the WTRU 102. In addition to or instead of the information from the GPS chipset 136, the WTRU 102 may receive location information from base stations (e.g., base stations 114a, 114b) via the air interface 116 and / or determine its location based on the timing of signals received from two or more nearby base stations. It should be understood that the WTRU 102 may acquire location information using any suitable location determination method while remaining consistent with the embodiments.
[0047] The processor 118 may be further coupled to other peripheral devices 138, which may include one or more software and / or hardware modules providing additional features, functions, and / or wired or wireless connectivity. For example, peripheral device 138 may include an accelerometer, electronic compass, satellite transceiver, digital camera (for photos and / or video), Universal Serial Bus (USB) port, vibration device, television transceiver, hands-free headset, Bluetooth® module, FM radio unit, digital music player, media player, video game console module, internet browser, virtual reality and / or augmented reality (VR / AR) device, activity tracker, etc. Peripheral device 138 may include one or more sensors, which may be one or more of the following: gyroscope, accelerometer, Hall effect sensor, magnetometer, orientation sensor, proximity sensor, temperature sensor, time sensor; geolocation sensor; altimeter, light sensor, touch sensor, magnetometer, barometer, gesture sensor, biometric sensor, and / or humidity sensor.
[0048] WTRU 102 may include a full-duplex radio for which the transmission and reception of some or all signals (e.g., associated with a specific subframe used for both UL (e.g., for transmission) and downlink (e.g., for reception)) may be concurrent and / or simultaneous. The full-duplex radio may include an interference management unit for reducing and / or substantially eliminating self-interference through signal processing via hardware (e.g., a choke) or via a processor (e.g., a separate processor (not shown) or via processor 118). In an embodiment, WTRU 102 may include a half-duplex radio for which the transmission and reception of some or all signals (e.g., associated with a specific subframe used for either UL (e.g., for transmission) or downlink (e.g., for reception)) may be concurrent and / or simultaneous.
[0049] Figure 1C This is a system diagram illustrating RAN 104 and CN 106 according to an embodiment. As described above, RAN 104 can employ E-UTRA radio technology to communicate with WTRUs 102a, 102b, and 102c via air interface 116. RAN 104 can also communicate with CN 106.
[0050] RAN 104 may include eNode-B 160a, 160b, 160c, but it should be understood that RAN 104 may include any number of eNode-Bs while remaining consistent with the embodiments. Each of eNode-B 160a, 160b, 160c may include one or more transceivers for communicating with WTRU 102a, 102b, 102c via air interface 116. In one embodiment, eNode-B 160a, 160b, 160c may implement MIMO technology. Therefore, for example, eNode-B 160a may use multiple antennas to transmit radio signals to and / or receive radio signals from WTRU 102a.
[0051] Each of the eNode-B 160a, 160b, and 160c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, and user scheduling in the UL and / or DL, etc. Figure 1C As shown, eNode-B 160a, 160b, and 160c can communicate with each other via the X2 interface.
[0052] Figure 1C The CN 106 shown may include a Mobility Management Entity (MME) 162, a Serving Gateway (SGW) 164, and a Packet Data Network (PDN) Gateway (or PGW) 166. While each of the foregoing elements is depicted as part of the CN 106, it should be understood that any of these elements may be owned and / or operated by an entity other than the CN operator.
[0053] The MME 162 can connect to each of the eNode-Bs 162a, 162b, and 162c in RAN 104 via the S1 interface and can act as a control node. For example, the MME 162 can be responsible for authenticating users of WTRUs 102a, 102b, and 102c, activating / deactivating bearers, selecting specific service gateways, etc., during the initial attachment of WTRUs 102a, 102b, and 102c. The MME 162 can provide control plane functions for handover between RAN 104 and other RANs (not shown) employing other radio technologies such as GSM and / or WCDMA.
[0054] The SGW 164 can connect to each of the eNode Bs 160a, 160b, and 160c in RAN 104 via the S1 interface. The SGW 164 typically routes and forwards user data packets to / from WTRUs 102a, 102b, and 102c. The SGW 164 can perform other functions, such as anchoring the user plane during inter-eNode B handover, triggering paging when DL data is available for WTRUs 102a, 102b, and 102c, and managing and storing the context of WTRUs 102a, 102b, and 102c.
[0055] SGW 164 can be connected to PGW 166, which can provide WTRU 102a, 102b, 102c with access to packet-switched networks (such as Internet 110) to facilitate communication between WTRU 102a, 102b, 102c and IP-enabled devices.
[0056] CN 106 can facilitate communication with other networks. For example, CN 106 can provide WTRUs 102a, 102b, and 102c with access to circuit-switched networks (such as PSTN 108) to facilitate communication between WTRUs 102a, 102b, and 102c and traditional landline communication equipment. For example, CN 106 may include or communicate with an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server) that acts as an interface between CN 106 and PSTN 108. Furthermore, CN 106 can provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers.
[0057] Despite WTRU in Figures 1A to 1D While described as a wireless terminal, it is conceivable that, in some representative embodiments, such a terminal may (e.g., temporarily or permanently) use a wired communication interface with a communication network.
[0058] In a representative embodiment, the other network 112 may be a WLAN.
[0059] A WLAN in Infrastructure Basic Services Set (BSS) mode may have an access point (AP) for the BSS and one or more stations (STAs) associated with the AP. The AP may have access or an interface for a distribution system (DS) carrying traffic flows to and from the BSS or another type of wired / wireless network. Traffic flows originating outside the BSS and destined for STAs can reach and be delivered to the STAs via the AP. Traffic flows originating from STAs and destined for destinations outside the BSS can be sent to the AP for delivery to the appropriate destination. Traffic flows between STAs within the BSS can be sent via the AP, for example, where a source STA can send a traffic flow to the AP and the AP can deliver the traffic flow to the destination STA. Traffic flows between STAs within the BSS can be considered and / or referred to as peer-to-peer traffic flows. These peer-to-peer traffic flows can be sent between the source STA and the destination STA (e.g., directly between them) using a direct link setup (DLS). In some representative embodiments, the DLS may use 802.11e DLS or 802.11z Tunneled DLS (TDLS). WLANs using the Standalone BSS (IBSS) mode can function without an access point (AP), and STAs within the IBSS or using the IBSS (e.g., all STAs) can communicate directly with each other. The IBSS communication mode may sometimes be referred to as a "self-organizing" communication mode in this document.
[0060] When operating in 802.11ac infrastructure mode or a similar mode, the AP can transmit beacons on a fixed channel, such as the primary channel. This primary channel can have a fixed width (e.g., a 20 MHz wide bandwidth) or a width dynamically set via signaling. The primary channel can be the operating channel of the BSS and can be used by the STA to establish a connection with the AP. In some representative embodiments, Carrier Sense Multiple Access with Collision Avoidance (CSMA / CA) can be implemented, for example, in an 802.11 system. With CSMA / CA, each STA (including the AP) can listen on the primary channel. If the primary channel is detected / discovered and / or determined to be busy by a particular STA, that STA can back off. A single STA (e.g., only one station) can transmit in a given BSS at any given time.
[0061] High-throughput (HT) STAs can communicate using a 40 MHz wide channel, for example, by combining a primary 20 MHz channel with adjacent or non-adjacent 20 MHz channels to form a 40 MHz wide channel.
[0062] Very High Throughput (VHT) STAs can support 20 MHz, 40 MHz, 80 MHz, and / or 160 MHz wide channels. 40 MHz and / or 80 MHz channels can be formed by combining consecutive 20 MHz channels. A 160 MHz channel can be formed by combining eight consecutive 20 MHz channels, or by combining two non-consecutive 80 MHz channels; this can be referred to as an 80+80 configuration. For the 80+80 configuration, after channel coding, data passes through a segment resolver, which divides the data into two streams. Inverse Fast Fourier Transform (IFFT) processing and time-domain processing can be performed separately on each stream. The streams can be mapped onto the two 80 MHz channels, and the data can be transmitted by the transmitting STA. At the receiver of the receiving STA, the above operations for the 80+80 configuration can be reversed, and the combined data can be sent to the Media Access Control (MAC).
[0063] 802.11af and 802.11ah support sub-1 GHz operating modes. The channel operating bandwidth and carrier are reduced in 802.11af and 802.11ah compared to those used in 802.11n and 802.11ac. 802.11af supports 5 MHz, 10 MHz, and 20 MHz bandwidths in the TV white space (TVWS) spectrum, while 802.11ah supports 1 MHz, 2 MHz, 4 MHz, 8 MHz, and 16 MHz bandwidths using non-TVWS spectrum. According to representative embodiments, 802.11ah can support instrument-type control / machine-type communications, such as MTC devices in macro coverage areas. MTC devices may have certain capabilities, such as limited capabilities, including support for (e.g., only) certain and / or limited bandwidths. MTC devices may include batteries with a battery life exceeding a threshold (e.g., to maintain a very long battery life).
[0064] WLAN systems that can support multiple channels and channel bandwidths (such as 802.11n, 802.11ac, 802.11af, and 802.11ah) include a channel that can be designated as the primary channel. This primary channel can have a bandwidth equal to the maximum common operating bandwidth supported by all STAs in the BSS. The bandwidth of the primary channel can be set and / or limited by the STAs supporting the minimum bandwidth operating mode (from all STAs operating in the BSS). In the example of 802.11ah, for STAs supporting (e.g., only supporting) the 1MHz mode (e.g., MTC type devices), the primary channel can still be 1 MHz wide, even if the AP and other STAs in the BSS support 2 MHz, 4 MHz, 8 MHz, 16 MHz, and / or other channel bandwidth operating modes. Carrier Sense and / or Network Allocation Vector (NAV) settings can depend on the state of the primary channel. If the primary channel is busy, for example, due to STAs supporting only the 1 MHz operating mode transmitting to the AP, the entire available band can be considered busy even if most of the band remains idle and potentially available.
[0065] In the United States, the available frequency band for 802.11ah is 902 MHz to 928 MHz. In South Korea, the available frequency band is 917.5 MHz to 923.5 MHz. In Japan, the available frequency band is 916.5 MHz to 927.5 MHz. Depending on the country code, the total bandwidth available for 802.11ah ranges from 6 MHz to 26 MHz.
[0066] Figure 1D This is a system diagram illustrating RAN 113 and CN 115 according to an embodiment. As described above, RAN 113 can employ NR radio technology to communicate with WTRUs 102a, 102b, and 102c via air interface 116. RAN 113 can also communicate with CN 115.
[0067] RAN 113 may include gNBs 180a, 180b, and 180c, but it should be understood that RAN 113 may include any number of gNBs while remaining consistent with the embodiments. Each of gNBs 180a, 180b, and 180c may include one or more transceivers for communicating with WTRUs 102a, 102b, and 102c via air interface 116. In one embodiment, gNBs 180a, 180b, and 180c may implement MIMO technology. For example, gNBs 180a and 180b may utilize beamforming to transmit signals to and / or receive signals from gNBs 180a, 180b, and 180c. Therefore, for example, gNB 180a may use multiple antennas to transmit radio signals to and / or receive radio signals from WTRU 102a. In an embodiment, gNBs 180a, 180b, and 180c may implement carrier aggregation technology. For example, gNB 180a can transmit multiple component carriers (not shown) to WTRU 102a. A subset of these component carriers may be on unlicensed spectrum, while the remaining component carriers may be on licensed spectrum. In embodiments, gNBs 180a, 180b, and 180c may implement Coordinated Multipoint (CoMP) technology. For example, WTRU 102a may receive coordinated transmissions from gNBs 180a and 180b (and / or gNB 180c).
[0068] WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using transmissions associated with scalable parameter sets. For example, OFDM symbol spacing and / or OFDM subcarrier spacing can vary for different transmissions, different cells, and / or different portions of the radio transmission spectrum. WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using subframes or transmission time intervals (TTIs) with multiple or scalable lengths (e.g., containing different numbers of OFDM symbols and / or varying absolute time lengths).
[0069] gNBs 180a, 180b, and 180c can be configured to communicate with WTRUs 102a, 102b, and 102c in standalone and / or non-standalone configurations. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c without simultaneously accessing other RANs (e.g., eNode-Bs 160a, 160b, and 160c). In standalone configuration, WTRUs 102a, 102b, and 102c can use one or more of gNBs 180a, 180b, and 180c as mobility anchors. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using signals in unlicensed frequency bands. In a non-standalone configuration, WTRUs 102a, 102b, and 102c can communicate / connect with gNBs 180a, 180b, and 180c, and also with another RAN (such as eNode-Bs 160a, 160b, and 160c). For example, WTRUs 102a, 102b, and 102c can implement DC principles to communicate substantially simultaneously with one or more gNBs 180a, 180b, and 180c and one or more eNode-Bs 160a, 160b, and 160c. In a non-standalone configuration, eNode-Bs 160a, 160b, and 160c can act as mobility anchors for WTRUs 102a, 102b, and 102c, and gNBs 180a, 180b, and 180c can provide additional coverage and / or throughput for serving WTRUs 102a, 102b, and 102c.
[0070] Each of gNBs 180a, 180b, and 180c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, user scheduling in UL and / or DL, network slicing support, dual connectivity, interoperability between NR and E-UTRA, routing of user plane data to User Plane Functions (UPF) 184a and 184b, routing of control plane information to Access and Mobility Management Functions (AMF) 182a and 182b, etc. Figure 1D As shown, gNB 180a, 180b, and 180c can communicate with each other via the Xn interface.
[0071] Figure 1DThe CN 115 shown may include at least one AMF 182a, 182b, at least one UPF 184a, 184b, at least one Session Management Function (SMF) 183a, 183b, and possibly a Data Network (DN) 185a, 185b. While each of the foregoing elements is depicted as part of the CN 115, it should be understood that any of these elements may be owned and / or operated by an entity other than the CN operator.
[0072] AMF 182a and 182b can connect to one or more of the gNBs 180a, 180b, and 180c in RAN 113 via the N2 interface and can act as control nodes. For example, AMF 182a and 182b can be responsible for authenticating users of WTRU 102a, 102b, and 102c, supporting network slicing (e.g., handling different PDU sessions with different requirements), selecting specific SMFs 183a and 183b, managing registration areas, terminating NAS signaling, mobility management, etc. AMF 182a and 182b can use network slicing to customize CN support for WTRU 102a, 102b, and 102c based on the type of service utilized by WTRU 102a, 102b, and 102c. For example, different network slices can be established for different use cases, such as services relying on Ultra Reliable Low Latency (URLLC) access, services relying on Enhanced Massive Mobile Broadband (eMBB) access, services for Machine Type Communication (MTC) access, etc. AMF 162 can provide control plane functions for switching between RAN 113 and other RANs (not shown) that employ other radio technologies such as LTE, LTE-A, LTE-APro and / or non-3GPP access technologies such as WiFi.
[0073] SMFs 183a and 183b can connect to AMFs 182a and 182b in CN 115 via the N11 interface. SMFs 183a and 183b can also connect to UPFs 184a and 184b in CN 115 via the N4 interface. SMFs 183a and 183b can select and control UPFs 184a and 184b, and configure the routing of service flows through UPFs 184a and 184b. SMFs 183a and 183b can perform other functions, such as managing and allocating UE IP addresses, managing PDU sessions, controlling policy enforcement and QoS, and providing downlink data notifications. PDU session types can be IP-based, non-IP-based, Ethernet-based, etc.
[0074] UPF 184a and 184b can connect to one or more of gNB 180a, 180b, and 180c in RAN 113 via the N3 interface. This N3 interface can provide WTRU 102a, 102b, and 102c with access to packet-switched networks (such as Internet 110) to facilitate communication between WTRU 102a, 102b, and 102c and IP-enabled devices. UPF 184 and 184b can perform other functions such as routing and forwarding packets, enforcing user plane policies, supporting multihomed PDU sessions, handling user plane QoS, buffering downlink packets, and providing mobility anchoring.
[0075] CN 115 can facilitate communication with other networks. For example, CN 115 may include or be able to communicate with an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server) that acts as an interface between CN 115 and PSTN 108. Furthermore, CN 115 can provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers. In one embodiment, WTRUs 102a, 102b, and 102c can be connected to the local data network (DN) 185a and 185b via the N3 interface to UPFs 184a and 184b and the N6 interface between UPFs 184a and 184b and DNs 185a and 185b.
[0076] Given Figures 1A to 1D as well as Figures 1A to 1D The functions described herein with respect to one or more of the following can be performed by one or more emulation devices (not shown): WTRU 102a to 102d, base stations 114a and 114b, eNode-B 160a to 160c, MME 162, SGW 164, PGW 166, gNB 180a to 180c, AMF 182a and 182b, UPF 184a and 184b, SMF 183a and 183b, DN 185a and 185b, and / or any other devices described herein. The emulation device can be one or more devices configured to emulate one or more of the functions described herein. For example, the emulation device can be used to test other devices and / or simulate network and / or WTRU functions.
[0077] The simulation device can be designed to perform one or more tests on other devices in a laboratory environment and / or a carrier network environment. For example, the one or more simulation devices can perform one or more of these functions while being fully or partially implemented and / or deployed as part of a wired and / or wireless communication network to test other devices within that communication network. The one or more simulation devices can also perform one or more of these functions while being temporarily implemented / deployed as part of a wired and / or wireless communication network. The simulation device can be directly coupled to another device for testing purposes and / or can use over-the-air wireless communication to perform the tests.
[0078] The one or more simulation devices can perform one or more (including all) of the functions without being implemented / deployed as part of a wired and / or wireless communication network. For example, the simulation device can be utilized in a test scenario in a test laboratory and / or in a non-deployed (e.g., test) wired and / or wireless communication network to perform testing on one or more components. The one or more simulation devices can be test equipment. The simulation device can transmit and / or receive data using direct RF coupling and / or wireless communication via RF circuitry (e.g., which may include one or more antennas).
[0079] This application describes various aspects, including tools, features, examples, models, schemes, etc. Many of these aspects are described in detail, and are generally described in a manner that may sound restrictive, at least to illustrate individual characteristics. However, this is for the sake of clarity and does not limit the scope of this application or those aspects. In fact, all the different aspects can be combined and interchanged to provide other aspects. Furthermore, these aspects can be combined and interchanged with aspects described in prior applications.
[0080] The aspects described and envisioned in this application can be implemented in many different forms. Figures 5 to 10 Some examples can be provided, but other examples are also welcome. Figures 5 to 10 The discussion does not limit the breadth of implementations. At least one of these aspects relates generally to video encoding and decoding, and at least one other aspect relates generally to transmitting a generated or encoded bitstream. These and other aspects can be implemented as methods, apparatus, computer-readable storage media storing instructions for encoding or decoding video data according to any of the methods, and / or computer-readable storage media storing bitstreams generated according to any of the methods.
[0081] In this application, the terms “reconstruction” and “decoding” are used interchangeably, the terms “pixel” and “sample” are used interchangeably, and the terms “image”, “picture” and “frame” are used interchangeably.
[0082] This document describes various methods, and each method includes one or more steps or actions for implementing the method. Unless the correct operation of the method requires a specific order of steps or actions, the order and / or use of specific steps and / or actions can be modified or combined. Furthermore, terms such as "first," "second," etc., can be used in various examples to modify elements, components, steps, operations, etc., such as, for example, "first decoding" and "second decoding." Unless specifically required, the use of such terms does not imply an ordering of the modified operations. Therefore, in this example, the first decoding need not be performed before the second decoding, and can occur, for example, before, during, or in a time period overlapping with the second decoding.
[0083] The various methods and other aspects described in this application can be used to modify, for example... Figure 2 and Figure 3 The illustrated video encoder 200 and decoder 300 modules are, for example, decoding modules. Furthermore, the subject matter disclosed herein can be applied to, for example, any type, format, or version of video encoding and decoding, whether described in standards or recommendations, whether pre-existing or future-developed, and any extensions to such standards and recommendations. Unless otherwise indicated or technically excluded, the aspects described in this application may be used individually or in combination.
[0084] Various numerical values are described in this application, such as exemplary performance of compression methods. These and other specific values are for illustrative purposes only, and the aspects described are not limited to these specific values.
[0085] Figure 2 This is an illustration of an exemplary video encoder. Variations of the exemplary encoder 200 are contemplated, but for clarity, encoder 200 is described below without describing all anticipated variations.
[0086] Before encoding, the video sequence may undergo pre-coding processing (201), such as applying color transformations to the input color images (e.g., a conversion from RGB 4:4:4 to YCbCr 4:2:0), or performing remapping of the input image components to obtain a more compression-resistant signal distribution (e.g., using histogram equalization on one of the color components). Metadata may be associated with pre-processing and appended to the bitstream.
[0087] In encoder 200, the image is encoded by encoder elements as described below. The image to be encoded is divided (202) and processed in units, for example, coding units (CUs). Each unit is encoded using, for example, intra-frame or inter-frame modes. When a unit is encoded in intra-frame mode, it performs intra-frame prediction (260). In inter-frame mode, motion estimation (275) and compensation (270) are performed. The encoder decides (205) whether to use intra-frame or inter-frame mode to encode the unit and indicates the intra-frame / inter-frame decision by, for example, a prediction mode flag. The prediction residual is calculated by, for example, subtracting (210) the predicted block from the original image block.
[0088] The prediction residual is then transformed (225) and quantized (230). The quantized transform coefficients, along with the motion vector and other syntax elements, are entropy-encoded (245) to output a bitstream. The encoder can skip the transform and apply quantization directly to the untransformed residual signal. The encoder can bypass both the transform and quantization, i.e., encode the residual directly without applying the transform or quantization process.
[0089] The encoder decodes the encoded blocks to provide a reference for further prediction. The quantized transform coefficients are dequantized (240) and inverse transformed (250) to decode the prediction residuals. The decoded prediction residuals and the predicted blocks are combined (255) to reconstruct the image blocks. A loop filter (265) is applied to the reconstructed image to perform, for example, deblocking / SAO (Sample Adaptive Offset) filtering to reduce coding artifacts. The filtered image is stored in the reference image buffer (280).
[0090] Figure 3 This is an illustration of an example video decoder. In the exemplary decoder 300, the bitstream is decoded by decoder elements as described below. The video decoder 300 generally performs the same as... Figure 2 The encoding process described herein is reciprocal to the decoding process. Encoder 200 also generally performs video decoding as part of the video data encoding.
[0091] Specifically, the decoder's input includes a video bitstream, which can be generated by the video encoder 200. First, the bitstream is entropy-decoded (330) to obtain transform coefficients, motion vectors, and other encoded information. Image segmentation information indicates how the image is segmented. Therefore, the decoder can segment (335) the image based on the decoded image segmentation information. The transform coefficients are dequantized (340) and inverse transformed (350) to decode the prediction residuals. The decoded prediction residuals and the predicted blocks are combined (355) to reconstruct image blocks. The predicted blocks can be obtained from intra-frame prediction (360) or motion-compensated prediction (i.e., inter-frame prediction) (375) (370). A loop filter (365) is applied to the reconstructed image. The filtered image is stored in a reference image buffer (380).
[0092] The decoded image can undergo further post-decoding processing (385), such as inverse color transformation (e.g., conversion from YCbCr4:2:0 to RGB 4:4:4) or inverse remapping of the inverse of the remapping process performed in pre-encoding processing (201). Post-decoding processing can utilize metadata derived in pre-encoding processing and signaled in the bitstream. In the example, the decoded image (e.g., after applying a loop filter (365) and / or after post-decoding processing (385) if post-decoding processing is used) can be sent to a display device for presentation to the user.
[0093] Figure 4 This is an illustration showing examples of systems that can implement the various aspects and examples described herein. System 400 can be embodied as a device including the various components described below and configured to perform one or more aspects described in this document. Examples of such devices include, but are not limited to, various electronic devices such as personal computers, laptop computers, smartphones, tablet computers, digital multimedia set-top boxes, digital television receivers, personal video recording systems, networked home appliances, and servers. The elements of system 400 can be embodied individually or in combination in a single integrated circuit (IC), multiple ICs, and / or discrete devices. For example, in at least one example, the processing elements and encoder / decoder elements of system 400 are distributed across multiple ICs and / or discrete devices. In many examples, system 400 is communicatively coupled to one or more other systems or other electronic devices via, for example, a communication bus or through dedicated input and / or output ports. In many examples, system 400 is configured to implement one or more aspects described in this document.
[0094] System 400 includes at least one processor 410 configured to execute instructions loaded therein for implementing various aspects, such as those described herein. Processor 410 may include embedded memory, input / output interfaces, and various other circuit systems known in the art. System 400 includes at least one memory 420 (e.g., a volatile memory device and / or a non-volatile memory device). System 400 includes a storage device 440 that may include non-volatile memory and / or volatile memory, including but not limited to electrically erasable programmable read-only memory (EEPROM), read-only memory (ROM), programmable read-only memory (PROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, disk drives, and / or optical disk drives. As a non-limiting example, storage device 440 may include internal storage devices, attached storage devices (including removable and non-removable storage devices), and / or network-accessible storage devices.
[0095] System 400 includes an encoder / decoder module 430 configured to, for example, process data to provide encoded or decoded video, and the encoder / decoder module 430 may include its own processor and memory. The encoder / decoder module 430 represents a module that can be included in a device to perform encoding and / or decoding functions. As is known, a device may include one or both of an encoding module and a decoding module. Furthermore, the encoder / decoder module 430 may be implemented as a separate element of system 400, or may be incorporated into processor 410 as a combination of hardware and software, as is known to those skilled in the art.
[0096] Program code to be loaded onto processor 410 or encoder / decoder 430 to execute the various aspects described in this document may be stored in storage device 440 and subsequently loaded onto memory 420 for execution by processor 410. According to various examples, one or more of processor 410, memory 420, storage device 440, and encoder / decoder module 430 may store one or more of a variety of items during the execution of the processes described in this document. Such stored items may include, but are not limited to, input video, decoded video or portions thereof, bitstreams, matrices, variables, and intermediate or final results from processing equations, formulas, operations, and operational logic.
[0097] In some examples, the memory within processor 410 and / or encoder / decoder module 430 is used to store instructions and provide working memory for the processing required during encoding or decoding. However, in other examples, external memory (e.g., the processing device may be processor 410 or encoder / decoder module 430) is used for one or more of these functions. External memory may be memory 420 and / or storage device 440, such as volatile memory and / or non-volatile flash memory. In several examples, external non-volatile flash memory is used to store, for example, the operating system of a television. In at least one example, fast external volatile memory such as RAM is used as working memory for video encoding and decoding operations.
[0098] Inputs to the components of system 400 can be provided by a variety of input devices as indicated in box 445. Such input devices include, but are not limited to, (i) a radio frequency (RF) section that receives, for example, RF signals transmitted over the air by a broadcaster, (ii) component input terminals (or a set of COMP input terminals), (iii) a universal serial bus (USB) input terminal, and / or (iv) a high-definition multimedia interface (HDMI) input terminal. Other examples ( Figure 4 (Not shown in the image) Includes composite video.
[0099] In various examples, the input device of block 445 has associated corresponding input processing elements, as known in the art. For example, the RF section may be associated with elements suitable for: (i) selecting a desired frequency (also known as selecting a signal, or band-limiting a signal to a frequency band), (ii) down-converting the selected signal, (iii) band-limiting it again to a narrower frequency band to (e.g.) select a signal band that may be referred to as a channel in some examples, (iv) demodulating the down-converted and band-limited signal, (v) performing error correction, and / or (vi) demultiplexing to select a desired data packet stream. The RF section of various examples includes one or more elements performing these functions, such as a frequency selector, signal selector, band limiter, channel selector, filter, downconverter, demodulator, error corrector, and demultiplexer. The RF section may include a tuner that performs multiple of these functions, including, for example, down-converting a received signal to a lower frequency (e.g., intermediate frequency or near-baseband frequency) or to baseband. In one set-top box example, the RF section and its associated input processing elements receive RF signals transmitted via a wired (e.g., cable) medium and perform frequency selection by filtering, down-converting, and re-filtering to the desired frequency band. Various examples rearrange the order of the above (and other) components, remove some of these components, and / or add other components that perform similar or different functions. Adding components may include inserting components between existing components, such as, for example, inserting amplifiers and analog-to-digital converters. In many examples, the RF section includes an antenna.
[0100] USB and / or HDMI terminals may include corresponding interface processors for connecting system 400 to other electronic devices across USB and / or HDMI connections. It should be understood that various aspects of input processing (e.g., Reed-Solomon error correction) may be implemented, for example, within a separate input processing IC or within processor 410 as needed. Similarly, aspects of USB or HDMI interface processing may be implemented, as needed, within a separate interface IC or within processor 410. The demodulated, error-corrected, and demultiplexed stream is provided to various processing elements, including, for example, processor 410 and an encoder / decoder 430 operating in combination with memory and storage elements, to process the data stream as needed for presentation on an output device.
[0101] Various components of system 400 can be provided within an integrated housing. Within the integrated housing, various components can be interconnected and transmit data therebetween using suitable connection means 425 (e.g., internal buses known in the art, including inter-IC (I2C) buses, wiring, and printed circuit boards).
[0102] System 400 includes a communication interface 450 that enables communication with other devices via a communication channel 460. The communication interface 450 may include, but is not limited to, a transceiver configured to transmit and receive data via the communication channel 460. The communication interface 450 may include, but is not limited to, a modem or network interface card (NIC), and the communication channel 460 may be implemented, for example, within a wired and / or wireless medium.
[0103] In several examples, a wireless network (such as a Wi-Fi network, e.g., IEEE 802.11 (IEEE stands for Institute of Electrical and Electronics Engineers)) is used to provide data to system 400 via streaming or otherwise. In these examples, the Wi-Fi signal is received via communication channel 460 and communication interface 450, which are adapted for Wi-Fi communication. Communication channel 460 in these examples is typically connected to an access point or router that provides access to external networks, including the Internet, to allow streaming applications and other over-the-top services to communicate. Other examples use a set-top box to provide streaming data to system 400, delivering data via an HDMI connection in input box 445. Still other examples use an RF connection in input box 445 to provide streaming data to system 400. As mentioned above, several examples provide data in a non-streaming manner. Furthermore, several examples use wireless networks other than Wi-Fi, such as cellular networks or Bluetooth® networks.
[0104] System 400 can provide output signals to a variety of output devices, including a display 475, a speaker 485, and other peripheral devices 495. Various examples of the display 475 include one or more of, for example, a touchscreen display, an organic light-emitting diode (OLED) display, a curved display, and / or a foldable display. The display 475 can be used in televisions, tablets, laptops, cellular phones (mobile phones), or other devices. The display 475 can also be integrated with other components (e.g., as in a smartphone) or stand alone (e.g., an external monitor for a laptop computer). Other peripheral devices 495 include, in various examples, one or more of a standalone digital video disc (or digital universal disc, both terms), a disc player, a stereo system, and / or a lighting system. Various examples utilize one or more peripheral devices 495 that provide functionality based on the output of system 400. For example, a disc player performs the function of playing the output of system 400.
[0105] In various examples, signaling such as AV.Link, Consumer Electronics Control (CEC), or other communication protocols enabling device-to-device control with or without user intervention is used to transmit control signals between system 400 and display 475, speaker 485, or other peripheral devices 495. Output devices can be communicatively coupled to system 400 via dedicated connections through corresponding interfaces 470, 480, and 490. Alternatively, output devices can be connected to system 400 via communication interface 450 using communication channel 460. Display 475 and speaker 485 can be integrated into a single unit within an electronic device, such as a television. In various examples, display interface 470 includes a display driver, such as a timing controller (TCon) chip.
[0106] Display 475 and speaker 485 may alternatively be separate from one or more other components, for example, if the RF section of input 445 is part of a separate set-top box. In various examples where display 475 and speaker 485 are external components, the output signal may be provided via dedicated output connections, including, for example, an HDMI port, a USB port, or a COMP output.
[0107] The example can be executed by computer software implemented by processor 410, or by hardware, or by a combination of hardware and software. As a non-limiting example, the example can be implemented by one or more integrated circuits. Memory 420 can be of any type suitable for the technical environment and can be implemented using any suitable data storage technology, such as optical memory devices, magnetic memory devices, semiconductor-based memory devices, fixed memory, and removable memory, as a non-limiting example. Processor 410 can be of any type suitable for the technical environment and can encompass one or more of microprocessors, general-purpose computers, special-purpose computers, and processors based on multi-core architectures, as a non-limiting example.
[0108] Various implementations involve decoding. As used herein, “decoding” can encompass, for example, all or part of a process performed on a received encoded sequence to produce a final output suitable for display. In many examples, such processes include one or more processes typically performed by a decoder, such as entropy decoding, inverse quantization, inverse transform, and differential decoding. In many examples, such processes also or alternatively include processes performed by a decoder of various embodiments described herein, such as obtaining a quantized latent variable associated with a neural network (NN) model, obtaining a quantization offset associated with the latent variable, reconstructing the latent variable associated with the NN model based on the quantization offset, etc.
[0109] As further examples, in one example, "decoding" refers only to entropy decoding; in another example, "decoding" refers only to differential decoding; and in yet another example, "decoding" refers to a combination of entropy decoding and differential decoding. Whether the phrase "decoding process" is intended to specifically refer to a subset of operations or to refer more broadly to a wider decoding process will be clear based on the specific context of the description and is considered well understood by those skilled in the art.
[0110] Various implementations involve encoding. In a manner similar to the above discussion of “decoding,” the term “encoding,” as used herein, can encompass, for example, all or part of the process performed on an input video sequence to produce an encoded bitstream. In various examples, such processes include one or more processes typically performed by an encoder, such as partitioning, differential coding, transform, quantization, and entropy coding. In various examples, such processes also or alternatively include processes performed by an encoder according to various embodiments described herein, such as obtaining a quantized latent variable associated with a neural network (NN) model, obtaining a quantization offset associated with the latent variable, encoding the latent variable associated with the NN model based on the quantization offset, etc.
[0111] As further examples, in one example, "encoding" refers only to entropy encoding; in another example, "encoding" refers only to differential encoding; and in yet another example, "encoding" refers to a combination of differential and entropy encoding. Whether the phrase "encoding process" is intended to specifically refer to a subset of operations or to refer more broadly to a wider encoding process will be clear based on the specific context of the description and is considered well understood by those skilled in the art.
[0112] Note that the grammatical elements used in this article are descriptive terms. Therefore, the use of other grammatical element names is not excluded.
[0113] When the accompanying drawings are presented as flowcharts, it should be understood that a block diagram of the corresponding apparatus is also provided. Similarly, when the accompanying drawings are presented as block diagrams, it should be understood that a flowchart of the corresponding method / process is also provided.
[0114] The embodiments and aspects described herein can be implemented, for example, in methods or processes, apparatuses, software programs, data streams, or signals. Even if discussed only in the context of a single implementation (e.g., discussed only as a method), implementations of the discussed features can be implemented in other forms (e.g., apparatuses or programs). Apparatuses can be implemented, for example, in suitable hardware, software, and firmware. Methods can be implemented, for example, in a processor, which generally refers to a processing device, including, for example, a computer, microprocessor, integrated circuit, or programmable logic device. Processors also include communication devices, such as, for example, computers, cellular phones, portable / personal digital assistants (“PDAs”), and other devices that facilitate information communication between end users.
[0115] References to “an example” or “an example” or “an implementation” or “an implementation”, and other variations thereof, mean that a particular feature, structure, characteristic, etc., described in connection with that example is included in at least one example. Therefore, the phrases “in an example” or “in a sample” or “in an implementation” or “in an implementation” appearing in multiple places throughout this application, and any other variations thereof, do not necessarily refer to the same example.
[0116] Furthermore, this application may involve "determining" various types of information. Determining information may include, for example, one or more of estimated information, calculated information, predicted information, or information retrieved from memory. Acquiring may include receiving, retrieving, constructing, generating, and / or determining.
[0117] Furthermore, this application may involve "accessing" various types of information. Accessing information may include, for example, receiving information, retrieving information (e.g., from memory), storing information, moving information, copying information, calculating information, determining information, predicting information, or estimating information, or more of these.
[0118] Furthermore, this application may relate to "receiving" various types of information. "Receiving," like "accessing," is intended as a broad term. Receiving information may include, for example, access information or retrieval information (e.g., from memory) or more of these. Moreover, "receiving" generally involves, in some way during operation, information such as, for example, storing information, processing information, transmitting information, moving information, copying information, erasing information, calculating information, determining information, predicting information, or estimating information.
[0119] It should be understood that the use of any of the following: “ / ”, “and / or”, and “at least one,” such as in the cases of “A / B,” “A and / or B,” and “at least one of A and B,” is intended to cover selecting only the first listed option (A), or only the second listed option (B), or selecting both options (A and B). As a further example, in the cases of “A, B, and / or C” and “at least one of A, B, and C,” such wording is intended to cover selecting only the first listed option (A), or only the second listed option (B), or only the third listed option (C), or only the first and second listed options (A and B), or only the first and third listed options (A and C), or only the second and third listed options (B and C), or selecting all three options (A, B, and C). This can be extended with respect to the number of items listed, as will be apparent to those skilled in the art and related fields.
[0120] Furthermore, as used herein, the term "signal" refers, among other things, to instructing the corresponding decoder to do something. In this way, in the examples, the same parameters are used on both the encoder and decoder sides. Therefore, for example, the encoder can transmit (explicit signaling) a specific parameter to the decoder so that the decoder can use the same specific parameter. Conversely, if the decoder already has that specific parameter as well as other parameters, signaling can be used without transmission (implicit signaling) to simply allow the decoder to know and select that specific parameter. Bit savings are achieved in various examples by avoiding the transmission of any actual function. It should be understood that signaling can be implemented in various ways. For example, one or more syntax elements, flags, etc., are used in various examples to signal information to the corresponding decoder. While the foregoing refers to the verb form of the word "signal," the word "signal" can also be used as a noun in this text.
[0121] As will be apparent to those skilled in the art, implementations can generate various signals formatted to carry information that can, for example, be stored or transmitted. The information may include, for example, instructions for performing a method, or data generated by one of the implementations. For example, the signal may be formatted to carry a bitstream of the example. Such a signal may be formatted, for example, as an electromagnetic wave (e.g., using the radio frequency portion of a spectrum) or a baseband signal. Formatting may include, for example, encoding a data stream and modulating a carrier wave with the encoded data stream. The information carried by the signal may be, for example, analog or digital information. The signal may be transmitted via a variety of different wired or wireless links as known. The signal may be stored on, or accessed or received from, a processor-readable medium.
[0122] This document describes numerous examples. Features of the examples may be provided individually or in any combination across multiple claim classes and types. Furthermore, examples may include one or more of the features, devices, or aspects described herein, individually or in any combination across multiple claim classes and types. For example, the features described herein may be implemented in a bitstream or signal that includes information generated as described herein. This information may allow a decoder to decode the bitstream, and the encoder, bitstream, and / or decoder may be implemented according to any of the described embodiments. For example, the features described herein may be implemented by creating and / or transmitting and / or receiving and / or decoding a bitstream or signal. For example, the features described herein may be implemented as a method, process, apparatus, medium storing instructions, medium storing data, or signal. For example, the features described herein may be implemented by a TV, set-top box, cellular phone, tablet computer, or other electronic device performing decoding. The TV, set-top box, cellular phone, tablet computer, or other electronic device may (e.g., using a monitor, screen, or other type of display) display a resulting image (e.g., an image from residual reconstruction of a video bitstream). The TV, set-top box, cellular phone, tablet computer, or other electronic device may receive a signal including an encoded image and perform decoding.
[0123] Neural network (NN) algorithms can be configured (e.g., applied to) image and / or video compression. NNs (such as deep NNs) can be applied to compression. For example, an NN can be configured to replace one or more modules associated with a video standard and / or replace the entire structure through an end-to-end NN-based learning process.
[0124] In the example, one or more of the following schemes can be configured: one or more encoder methods, methods built using one or more standards (e.g., on top of them), and / or one or more end-to-end NN-based codecs.
[0125] One or more encoder methods can use one or more NN-based algorithms to enhance or accelerate the encoder of a codec.
[0126] In one or more methods built using one or more standards (e.g., on top of them), one or more modules of the codec can be replaced by one or more NN-based methods, such as post-filters, prediction modules, etc.
[0127] In one or more end-to-end NN-based codecs and / or codec tools, one or more compression schemes (such as prediction, transform, quantization, and / or entropy coding) can be subverted.
[0128] This article may describe one or more end-to-end NN-based codecs and / or codec tools.
[0129] One or more encoding / decoding methods can apply prediction patterns, such as predefined prediction patterns, and / or transformations. One or more neural network-based methods can use (e.g., depend on) one or more parameters learned on a dataset (e.g., a large dataset) during the training phase. For example, one or more neural network-based methods can depend on one or more parameters learned on a large dataset by iteratively minimizing a loss function during the training phase. In the case of image and / or video compression, the loss function can be defined by rate-distortion cost. Rate can represent the estimated bit rate of the encoded bitstream (e.g., video data). Distortion can quantify the quality of the decoded video, for example, relative to the input (e.g., the original input). The quality of the decoded input image can be optimized, for example, based on a metric of mean squared error and / or an approximation of human-perceived visual quality.
[0130] Figure 5 An exemplary self-encoder architecture is illustrated. For example, Figure 5 An exemplary autoencoder based on a total decomposition prior can be illustrated for image compression. For example... Figure 5 The diagram illustrates an exemplary end-to-end compression system. The input to the encoder portion of the network is shown. It may include one or more of the following: an image or video frame; a portion of an image; a tensor representing a set of images; or a tensor representing a portion of a set of images (e.g., a cropped image).
[0131] In the example, the input can have one or more components, such as monochrome, RGB, or YCbCr components.
[0132] enter It can be fed into the encoder network In this context, it is called an analytical transformation. This can be a series of convolutional layers with activation functions. The convolution can be, or can include, a mechanism for spatially downsampling the input. For example, the convolution can be selected to have a stride of 2 in the vertical and / or horizontal directions. The convolution can result in an output having half the size of the input in one or more (e.g., two) dimensions. The output of the convolution can be a tensor of shape C×H×W. H and W can be the spatial height and width, respectively. C can correspond to the adjustable number of channels. For RGB images, The first convolution takes a 3-channel tensor as input, with these channels corresponding to the color components. The encoder network can be viewed as a learned transformation. For example, this learned transformation can be lossy because the number of elements in the output latent tensor is typically less than the number of elements in the source C×H×W input. The output of the analysis (e.g., primarily in the form of a 3-way array, called a 3D tensor) can be referred to as a latent representation or a latent variable tensor. From a broader perspective, the set of latent variables can construct a latent space, which can be frequently used in the context of end-to-end compression based on neural networks.
[0133] Output It can be quantized. For example, a quantized output can produce a tensor. Tensor It can be entropy encoded into a binary stream (e.g., bitstream or video data) for storage and / or transmission.
[0134] At the decoder, the bitstream and / or video data can be entropy decoded (ED) to obtain... .
[0135] Decoder networks (such as) This can be referred to as a synthesis transformation. For example, a decoder network can generate a reconstructed input: The input for reconstruction can be based on a quantized latent representation. input (For example, the original) Approximate to ). It can be a series of upsampled convolutions, such as transposed convolutions or convolutions followed by upsampled filters. The decoder network can be viewed as learned inverse transforms and / or denoising and generative transforms.
[0136] Compression systems (such as image and / or video compression systems) may have one or more compression architectures based on decomposition priors and / or super-priors.
[0137] In the example, the performance of the compression system can be measured as a trade-off between the number of bits required for transmission and the quality of the decoded content. The compression model can be trained using loss functions, such as those following a Lagrange form: .
[0138] R It can represent rate and / or bit rate. D It can represent the distortion of the content after decoding.
[0139] As described in this article, Figure 5 An exemplary autoencoder-based compression architecture is illustrated. Figure 5The exemplary autoencoder-based compression architecture illustrated in the diagram may include an inference configuration (e.g., an actual inference configuration) comprising an encoder that produces a bitstream (e.g., video data). The resulting bitstream (e.g., video data) may be transmitted and / or decoded by a decoder. To achieve a bitrate of R (e.g., a small bitrate), R The bitstream (e.g., video data) used at the encoder and / or decoder by the lossless compression algorithm (such as arithmetic codec) may know (e.g., may need to know) the quantized latent tensor. The probability of.
[0140] Figure 6 The illustration shows an exemplary training of a fully decomposed prior model. Figure 6 As illustrated, entropy encoding / decoding can be non-differentiable. Rate term. R It can be used It is represented by entropy. The entropy can be estimated via a differentiable surrogate, for example, using... The learned probability model ,like Figure 6 As shown in the diagram. In so-called fully decomposed prior (e.g., probabilistic) models, latent variables within a channel can be assumed to be independent and / or identically distributed (i.i.d.). For example, latent tensors The joint probability distribution of variables within a channel can be decomposed on the same probability distribution of the variables (e.g., each variable) (e.g., the name of the model).
[0141] As for distortion, such NNs can be trained using one or more of the following types of losses: loss based on objective metrics and / or loss based on subjective metrics.
[0142] In the example, the loss based on an objective metric could be mean squared error (MSE), for example, Figure 6 The diagram in the middle In the example, loss based on objective metrics can be based on structural similarity (SSIM). The results may not be as perceptually good as the second type, but the fidelity to the signal (such as the original signal) (e.g., an image) can be higher.
[0143] In the example, the loss based on subjective (e.g., or subjectively via a proxy) metrics can be used during the training phase using a generative adversarial network (GAN) and / or a high-level visual metric via a proxy NN.
[0144] exist Figure 5 and / or Figure 6In the illustrated autoencoder-based compression system, the entropy encoder and / or decoder can rely on a fully factored prior (e.g., probabilistic) model. One or more separately trained entropy models can be configured and / or considered for each latent channel. Since the samples (e.g., each sample) are encoded using the same distribution, for example, assuming the samples are independent and / or identically distributed (i.i.d.), latent channels can be disregarded. Spatial and / or channel correlations within. In the context of... After processing, It may not be identically distributed (iid).
[0145] Figure 7 An exemplary self-encoder employing a priori approach is illustrated. Figure 7 As illustrated, the model may include one or more additional convolutional sequences. and . It can output one or more learned distribution parameters. For example, the one or more learned distribution parameters can be and / or can include one or more scales and / or means of a Gaussian or Laplace distribution, for example, for latent... Each (e.g.) element. Quantized tensor. (For example, called supersubmersible) can be made by Output. For example, a quantized tensor. It may need to be encoded and / or transmitted for the decoder to decode. Edge information. Transmitting the hypertonic tensor using a fully decomposed probabilistic model. It probably won't cost too much, because Corresponding to by Further downsampling And for potential Using a custom Gaussian probability model for each (e.g., each) element is far more efficient than transporting lightweight data. The burden.
[0146] Figure 7 This can be illustrated as one or more operations and / or elements necessary for encoding and / or decoding an image. For example, by Synthesis distribution information (e.g., scale and / or mean) can appear at the encoder and / or decoder. As in video coding, the encoder can include one or more parts of the decoder to generate exactly the same metadata that the decoder will decode (e.g., to process the rest of the bitstream, such as video data). One or more bit operations (e.g., one or more precise bit operations) can be performed between the encoder and decoder to make the system work. Small differences in the generated parameters can lead to variations in the output parameters. The arithmetic decoder crashes (e.g., completely crashes).
[0147] Variable bit rate compression can be configured using an end-to-end learned image and / or video compression architecture.
[0148] Changing the compression ratio of encoded content can be used in lossy image and / or lossy video compression systems (e.g., it may be a requirement). In one example, the encoder might want to increase the compression ratio, for example, to reduce the bit rate of the compressed bitstream (e.g., video data) to suit the transmission rate requirements of the communication medium. In another example, the encoder might want to decrease the compression ratio, for example, to allow the decoder to reconstruct the image and / or video content with better visual quality (e.g., lower distortion). The desired compression ratio and / or compressed bit rate can be set before compression begins and / or during compression. For example, in video compression applications, each frame of content might need to be compressed at a different compression ratio (e.g., slightly different compression ratios) to prevent large fluctuations in bit rate and / or distortion across consecutive frames.
[0149] Image and / or video compression systems (e.g., traditional image and / or video compression systems that are not based on neural networks) may change scalar parameters (e.g., a single scalar parameter) and / or quantization step size, such as the coefficients of one or more (e.g., all) transforms (e.g., other representations) within the compression system. One or more other parameters within the compression system may remain unchanged (e.g., may not need to be changed). The methods described herein can provide a simple and / or computationally efficient algorithm for changing the compression ratio. For example, the methods described herein can provide an optimal or near-optimal trade-off / variation between bit rate and distortion when the compression ratio changes. This optimal or near-optimal trade-off / variation described herein can be attributed to the transform used, such as an orthogonal transform with one or more properties (e.g., one or more specific properties).
[0150] In the examples, while changing the compression ratio or bit rate can be performed using simple and / or sophisticated methods for image and / or video compression systems (e.g., conventional image and / or video compression systems), end-to-end learning image and / or video compression systems may not be configured as described herein (e.g., changing the compression ratio or bit rate). Changing the compression ratio can be performed by modifying one or more (e.g., all) weights of the neural network (NN) in the system. Changing the compression ratio for the NN in the system may require (e.g., necessitates) the use of different weights in the cost function. To retrain the entire system.
[0151] As described herein, changing the compression ratio may be necessary in a compression system, and / or an end-to-end learning image compression system may use a set of neural network weights (e.g., parameters) within the system. Using this set of neural network weights (e.g., parameters) within the system may not be a feasible solution for changing the compression ratio. For example, changing the compression ratio, as described herein, may require storing and / or loading one or more sets of neural network weights to modify the compression ratio and / or bit rate. If fine-tuning of the bit rate is desired (e.g., features with approximately 50 quantization step values / selected features of one or more video compression standards), storing one or more sets of neural network weights may be necessary and could be cumbersome (e.g., impractical). For end-to-end learning image and / or video compression systems, a set (e.g., a single set) of neural network weights (e.g., a model) can be configured and / or used. If a set (e.g., a single set) of neural network weights (e.g., a model) is configured and / or used, one or more additional parameters can be modified to adjust the compression ratio and / or bit rate.
[0152] As described in this paper, a single set of NN weights (e.g., the model) is used in end-to-end learning image or video compression systems to change the compression ratio (e.g., to achieve variable rate compression). Quantization algorithms, such as those utilizing reconstruction offsets, can be used.
[0153] Variable bit rate compression can be implemented by configuring and / or using one or more algorithms to leverage NN models (e.g., a single NN model).
[0154] In the example, a variable-rate image compression framework based on a conditional autoencoder can be configured, for example, instead of one or more convolutional layers. For instance, a conditional autoencoder may include one or more conditional convolutional layers (e.g., adjusting parameters at a cost). (Conditional). The quantization stride can be varied. For example, two parameters can be configured to change the bit rate. Conditional convolutional layers can be computationally complex. For example, a conditional convolutional layer may be more computationally complex than one or more standard convolutional layers, and may involve complex design and / or algorithms.
[0155] In the example, a variable-rate image compression framework based on so-called gain units can be configured. A gain unit can be a multiplication factor for each latent tensor channel. For example, latent tensor channels can be multiplied by their own factors before and / or after rounding operations. These two factors may not necessarily be reciprocals of each other, as described in this article: (1) (2) In equation (1), It can represent the analytical function of the learned compression system. Output latent variables (e.g., in horizontal position) and vertical position The passage (for example, such as) Figure 7 (As shown in the diagram). It can be for channels Gain and / or factor applied before rounding. This can be an operation that rounds a value to the nearest integer. This can be a result quantification index. In equation (2), this can be achieved by multiplying again by the gain and / or factor. (For example, it can be due to (e.g., each) channel) (and differences) are based on quantization indexes Quantified values of latent variables The reconstruction.
[0156] To roughly adjust the rate, we can learn for (e.g., each) latent channels (e.g., each) and / or A predefined set of six factors is used. For finer rate adjustment, two predefined factors can be obtained (e.g., selected) and interpolated (e.g., using exponential interpolation) to obtain gain units for the desired compression ratio and / or bit rate. The variable rate image compression framework based on one or more gain units as described herein can have a simpler design and / or algorithm. For example, the variable rate image compression framework based on one or more gain units as described herein can have a simpler design and / or algorithm than the conditional autoencoder-based variable rate image compression framework described herein.
[0157] In the example, the variable rate image compression framework can be based on changing parameters (e.g., a single parameter) and / or quantization step size (e.g., which is used to quantize latent tensor variables, such as each latent tensor variable).
[0158] (3) (4) Equations (3) and (4) described herein can be example cases (e.g., special cases) of equations (1) and (2) described herein. For example, if equations (1) and (2) are used and Then we can obtain equations (3) and (4). Note that... Can be independent of channel index To adjust the rate, the quantization step size can be changed. As described herein, the use of equations (3) and (4) can be similar to the methods used in image and / or video compression. The methods described herein (e.g., using equations (3) and (4)) can provide simple (e.g., more concise) designs and / or algorithms.
[0159] Figure 8 An exemplary comparison of the compression performance of several variable-rate methods (e.g., image compression for end-to-end learning) is illustrated.
[0160] like Figure 8 As shown, the baseline (e.g., Balle ICLR2018) curve uses a method that changes the compression ratio; in other words, it is retrained eight times, each time for a specific target bit rate. The Cui (e.g., CVPR2021) curve can use a method based on equations (1) and (2). The Tong (e.g., ICIP2023) curve can be based on equations (3) and (4). Figure 8 As shown, the curves for Cui CVPR2021 and Tong ICIP2023 may be lower than those for the default method at higher bit rates, which may indicate that their compression performance deteriorates at higher bit rates.
[0161] As described in this article, parameters (such as quantization offset) can be used in the quantization process within end-to-end learning image systems and / or video compression systems. Configuration can be based on changes in latent variables. A variable-rate end-to-end learning framework for image and / or video compression with parameters within the quantization operation. For example, an exemplary parameter could be the scalar quantization step size. For example, in equations (3) and / or (4) used in image or video compression systems. An exemplary parameter could be the accompanying quantization offset. Quantization can be implemented in two steps as follows: (5) (6) In equation (5), It can represent the compression system learned (e.g., such as...) Figure 7 Analysis function (as shown in the figure) Output latent variables (e.g., in horizontal position) and vertical position The passage middle). It can be the quantization step size, and This can be an operation that rounds a value to the nearest integer. It can be a result quantization index. In equation (6), it can be based on the quantization index. To reconstruct the quantized values of latent variables . It can indicate the quantization offset.
[0162] Quantization offset It can be used to reconstruct the quantified values of latent variables. (For example, as in equation (6)).
[0163] While the quantization operation utilizing offsets can be summarized by two equations, the exemplary implementation can use more equations and / or operations. For example, for each channel... and / or spatial location Calculate quantization offset Additional calculations may be required. For example, for all Latent variables with a value of 0 (e.g., The quantization offset can be zero.
[0164] Figure 9 The illustration depicts an exemplary hyperprior architecture autoencoder utilizing quantization offset. (As shown...) Figure 9 As shown, an exemplary hyper-prior architecture autoencoder (e.g., as...) Figure 7 (As illustrated) can be updated, for example, to reflect the quantization operation utilizing the quantization offset. Modifications including the quantization offset can occur at the decoder and / or encoder. For example, for other end-to-end learning image or video compression system architectures (such as...) Figure 6 The architecture illustrated can use quantization offsets on the encoder side (e.g., and / or decoder side).
[0165] Quantization offset can be used in end-to-end learning image and video compression systems. The functionality of quantization offset can be derived from a synthesis function. Implicitly learn and / or absorb into the composition function In (for example, when using a specific loss function) (This applies when training an end-to-end learning image or video compression system for a specific bit rate). Using explicit quantization offsets (e.g., as described in this paper) can improve the performance of variable bit rate compression.
[0166] Figure 10 The diagram illustrates example compression performance comparisons (e.g., three methods within end-to-end image compression). Solid curves demonstrate the compression performance of variable bit rate compression methods (e.g., the default variable bit rate compression method) by using (e.g., training) eight different models. Dashed (e.g., or dashed) curves demonstrate compression performance when only the quantization step size is changed. The performance of a single-model variable bit-rate compression method is shown. The dotted line curve illustrates the variation in quantization step size. Using quantization offset (e.g., appropriate quantization offset) A single-model variable bit rate compression method. For example... Figure 10 As illustrated, introducing a quantization offset can improve compression performance, for example, in a single-model variable bit rate compression system.
[0167] Quantization offset For each channel Horizontal index and / or vertical index They can be different, or have a quantization offset. It is possible (for example, only) for different underwater passages It may differ, or for all hidden passages and spatial index They can be the same.
[0168] Quantization offset for each latent variable It can be determined / computed (e.g., using a neural network, or any function) and / or retrieved (e.g., obtained from a lookup table).
[0169] Quantization offset for each latent variable It can be determined / computed (e.g., by some arithmetic and / or geometric interpolation) and / or the value (e.g., from a lookup table) can be subjected to some further processing.
[0170] Quantization offset for each latent variable It can be determined / computed (e.g., using a neural network and / or some arbitrary function with one or more inputs that can carry relevant information such as the estimated standard deviation of the latent variable distribution).
[0171] Various training strategies can be implemented for the framework described in this paper.
[0172] Exemplary training strategies may include training an end-to-end image or video compression system (e.g., the entire end-to-end image or video compression system, such as...). Figure 7 (One of the illustrations) and quantization offset (For example, training together from the beginning). For instance, a training strategy could include training an end-to-end image or video compression system (which includes subsystems). , , , (e.g., all subsystems), quantization step size (and / or additional subsystems / parameters) and quantization offset (For example, other parameters and / or tables from which the quantization offset can be derived).
[0173] Exemplary training strategies may include further training or fine-tuning an end-to-end pre-trained image or video compression system (e.g., an end-to-end pre-trained image or video compression system with a high cost function). Values (such as) Figure 10The top right corner point of the solid curve shown in the figure was trained (and there was no quantization offset in the quantization operation), for example, by modifying the quantization operation to include (e.g., also include) the quantization offset. The quantization offset (or parameters and / or lookup tables from which the quantization offset can be derived) and / or system parameters can be updated / learned in this fine-tuning phase (e.g., post-training).
[0174] Exemplary training strategies may include employing an end-to-end pre-trained image or video compression system (e.g., an end-to-end pre-trained image or video compression system with high cost function). Values (such as) Figure 10 The top right corner point of the solid curve shown in the figure was trained (and there was no quantization offset in the quantization operation), and the quantization operation was modified to include (e.g., also include) the quantization offset. The quantization offset within the system (or parameters and / or lookup tables from which the quantization offset can be derived) can be trained in the fine-tuning phase (e.g., post-training), while the system parameters can remain unchanged (e.g., remain unchanged).
[0175] Although features and elements have been described above in specific combinations, those skilled in the art will understand that each feature or element can be used alone or in any combination with other features and elements. Furthermore, the methods described herein can be implemented in a computer program, software, or firmware incorporated in a computer-readable medium for execution by a computer or processor. Examples of computer-readable media include electronic signals (transmitted via a wired or wireless connection) and computer-readable storage media. Examples of computer-readable storage media include, but are not limited to, read-only memory (ROM), random access memory (RAM), registers, cache memory, semiconductor memory devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, and optical media such as CD-ROMs and digital multifunction discs (DVDs). The processor associated with the software can be used to implement a radio frequency transceiver for use in a WTRU, UE, terminal, base station, RNC, or any host computer.
Claims
1. A video decoding device, comprising: The processor is configured as follows: Obtain latent variables associated with neural network (NN) models; Obtain the quantization offset associated with the latent variable; and The quantized latent variables associated with the NN model are reconstructed based on the quantization offset.
2. A video encoding device, comprising: The processor is configured as follows: Obtain latent variables associated with neural network (NN) models; Obtain the quantization offset associated with the latent variable; and The quantized latent variables associated with the NN model are reconstructed based on the quantization offset.
3. The device as claimed in any one of claims 1 or 2, wherein the processor is further configured to: The quantized latent variables associated with the NN model are further reconstructed based on the quantization step size.
4. The device as claimed in any one of claims 1 to 3, wherein the processor is further configured to: The quantization index is obtained based on the latent variables and the quantization step size; and The quantized latent variables associated with the NN model are further reconstructed based on the quantization index.
5. The device of claim 4, wherein the quantization index is an integer value.
6. The device of any one of claims 1 to 5, wherein the NN is trained without the quantization offset.
7. The device of any one of claims 1 to 6, wherein the quantization offset associated with the latent variable is obtained based on a lookup table.
8. The device of any one of claims 1 to 6, wherein the quantization offset associated with the latent variable is obtained using a second neural network.
9. The device of any one of claims 1 to 6, wherein the quantization offset associated with the latent variable is obtained based on the estimated standard deviation of the distribution associated with the latent variable.
10. A video decoding method, comprising: Obtain latent variables associated with neural network (NN) models; Obtain the quantization offset associated with the latent variable; as well as The quantized latent variables associated with the NN model are reconstructed based on the quantization offset.
11. A video encoding method, comprising: Obtain latent variables associated with neural network (NN) models; Obtain the quantization offset associated with the latent variable; as well as The quantized latent variables associated with the NN model are reconstructed based on the quantization offset.
12. The method of any one of claims 10 or 11, wherein the method further comprises: The quantized latent variables associated with the NN model are further reconstructed based on the quantization step size.
13. The method of any one of claims 10 to 12, wherein the method further comprises: The quantization index is obtained based on the latent variables and the quantization step size; as well as The quantized latent variables associated with the NN model are further reconstructed based on the quantization index.
14. The method of claim 13, wherein the quantization index is an integer value.
15. The method of any one of claims 10 to 14, wherein the NN is trained without the quantization offset.
16. The method of any one of claims 10 to 15, wherein the quantization offset associated with the latent variable is obtained based on a lookup table.
17. The method of any one of claims 10 to 15, wherein the quantization offset associated with the latent variable is obtained using a second neural network.
18. The method of any one of claims 10 to 15, wherein the quantization offset associated with the latent variable is obtained based on the estimated standard deviation of the distribution associated with the latent variable.
19. A computer program product stored on a non-transitory computer-readable medium and comprising program code instructions for implementing the steps of the method according to at least one of claims 10 to 18 when executed by a processor.
20. Video data comprising information representing the quantized latent variables decoded according to any one of claims 10 to 18.
