WO2019053436A1 - Sous-échantillonnage spatio-temporel de signaux vidéo numériques - Google Patents

Sous-échantillonnage spatio-temporel de signaux vidéo numériques Download PDF

Info

Publication number
WO2019053436A1
WO2019053436A1 PCT/GB2018/052604 GB2018052604W WO2019053436A1 WO 2019053436 A1 WO2019053436 A1 WO 2019053436A1 GB 2018052604 W GB2018052604 W GB 2018052604W WO 2019053436 A1 WO2019053436 A1 WO 2019053436A1
Authority
WO
WIPO (PCT)
Prior art keywords
stream
video frames
sampling
sampled
video
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/GB2018/052604
Other languages
English (en)
Inventor
Fan Zhang
Mariana AFONSO
David Bull
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Bristol
Original Assignee
University of Bristol
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by University of Bristol filed Critical University of Bristol
Publication of WO2019053436A1 publication Critical patent/WO2019053436A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/587—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal sub-sampling or interpolation, e.g. decimation or subsequent interpolation of pictures in a video sequence
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/154—Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/177—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a group of pictures [GOP]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/179—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a scene or a shot
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/59—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution

Definitions

  • the present inventors have devised a new method of processing video frames that can significantly reduce the memory required to store video frames and the bandwidth required to transmit video frames in a substantially visually lossless manner.
  • a computer implemented video processing method comprising : receiving an incoming stream of video frames; temporally down-sampling a first stream of video frames; temporally up-sampling the first stream of video frames; performing a first assessment at a first assessment module comprising assessing the perceptual quality of the temporally up- sampled video frames with respect to a corresponding portion of the incoming stream of video frames using a first perceptual quality metric; outputting a first output comprising the temporally down-sampled video frames if the outcome of the first perceptual quality assessment is a first outcome and outputting a non-temporally down-sampled stream of video frames if the outcome of the first perceptual quality assessment is a second outcome; spatially down-sampling a second stream of video frames; spatially up-sampling the second stream of video frames; performing a second assessment at a second assessment module, the second assessment comprising either: assessing the perceptual quality of the up
  • the method according to the first aspect of the invention provides an efficient resolution adaptation framework which dynamically applies spatial and temporal down- sampling to frames exhibiting relatively low spatial detail and lower motion.
  • visually-lossless spatiotemporal resolution adaptation is achieved by assessing the perceptual quality of a down-sampled video sequence using distinct first and second perceptual quality metrics.
  • this method is integrated with conventional video compression algorithms, it is possible to trade off the relationship between resolutions and coding quantisation parameters to optimise rate quality performance.
  • a modified Convolutional Neural Network (CNN)-based single image super resolution method can be employed to reconstruct full spatial resolution content.
  • CNN Convolutional Neural Network
  • a method according to an embodiment of the invention can provide one or more of the following advantages: • It uses advanced perceptual quality metrics to selectively apply spatially and/or temporally down-sampled videos within the coding loop.
  • FIG. 1 is a diagram of a computer implemented method according to an embodiment of the invention
  • FIG. 2 is a diagram of a system configured to perform method of FIG. 1 ;
  • FIG. 3 is diagram of a computer implemented method according to a second embodiment of the invention.
  • FIG. 4 is a diagram of a system configured to perform method of FIG. 3;
  • FIG. 5 is a graph illustrating the percentage of frames saved on the BVI-HFR database
  • FIG. 6 shows rate-PSNR curves for the method according to an embodiment of the invention and its HEVC anchor for nine different sets of video frames
  • FIG. 7 shows rate-VMAF curves for the method and its HEVC anchor for the nine different sets of video frames of FIG. 6.
  • Embodiments of the invention relate to a computer implemented video processing method comprising content-adaptive spatiotemporal down-sampling based on a measure of quality degradation due to down-sampling
  • FIG. 1 is a flow chart illustrating a computer implemented video processing method 10 according to an embodiment of the invention.
  • FIG. 2 is a system diagram of a system arranged to perform the method. The method is a video acquisition method.
  • an incoming stream of video frames is received from an image capture device C such as a digital video camera or recorder.
  • the incoming stream can comprise raw video frames captured by camera sensors.
  • a first stream of video frames which in the illustrated embodiment is a portion of the incoming stream of video frames, are temporally down-sampled at module TDS.
  • the first stream can comprise 2, 4 or 8 frames for example (i.e down- sampling is by a factor of 2, 4, or 8).
  • Temporal down-sampling can for example be achieved using frame averaging, which emulates a hold-type display (e.g. LCD).
  • Other filter operations such as frame dropping, weighted averaging or non-linear filtering could however be applied here.
  • the down-sampled video frames are temporally up-sampled at module TUS.
  • Temporal up-sampling can be achieved using various algorithms. In the illustrated embodiment temporal up-sampling is achieved using nearest-neighbour interpolation such as that described in R. C. Gonzalez and R. E. Woods, Digital Image Processing, 2nd ed. Prentice Hall, 2002.
  • the perceptual quality of the temporally up-sampled video frames is assessed at a first assessment module AM I with respect to a corresponding portion of the incoming stream of video frames using a first perceptual quality metric. The purpose is to determine whether the down-sampling has adversely affected the quality of the video frames by assessing the perceptual impact of temporal artefacts in the down-sampled video frames.
  • the first perceptual quality metric comprises a low complexity frame rate dependent video quality metric (FRQM) as described in F. Zhang, A. Mackin, and D. R. Bull, "A new frame rate dependant quality model based on temporal wavelet decomposition and spatiotemporal pooling," in Proc. IEEE Int Conf. on Image Processing. IEEE, 2017. FRQM is employed to assess the perceptual quality difference between temporally down-sampled videos (different sample factors may be applied) and its full frame rate original. If the resulting FRQM score is higher than a predetermined threshold (e.g. 48 out of 100), those video frames with lower frame rates are stored in place of their original versions. Otherwise, full frame rate video frames are retained.
  • a predetermined threshold e.g. 48 out of 100
  • FRQM Due to the flexibility of FRQM, it is possible to apply this comparison and decision making within a time window comprising a small number of frames (e.g. 2, 4 or 8). If multiple re-sampling factors can all meet the above criterion, then the highest re-sampling ratio will be selected.
  • the method outputs a first output which can either be the temporally down- sampled video frames or the first stream of video frames.
  • the temporally down- sampled video frames are used for the first output if the perceptual quality assessment is for example greater than a first threshold value.
  • the non-temporally down-sampled first stream of video frames is output if the perceptual quality assessment is for example less than the threshold value.
  • the method determines whether the first stream of video frames was suitable for temporal down-sampling and, if so, it uses the temporally down-sampled video frames.
  • a second stream of video frames which in this embodiment are the video frames output at step 20, are spatially down-sampled at module SDS.
  • the second stream can consist of a single video frame at a time.
  • Spatial down-sampling can be achieved using various interpolation filters. In the illustrated embodiment spatial down-sampling is achieved using a Lanczos3 filter as described in K. Turkowski, "Filters for common resampling-tasks," Graphics gems, pp. 147-165, 1990.
  • the down-sampled video frame is spatially up-sampled at module SUS. Spatial up-sampling can be achieved using various super-resolution algorithms.
  • spatial up-sampling is achieved using a modified CNN-based super-resolution approach (VDSR) such as that described in J. Kim, J. K. L, and K. M. L, "Accurate image super-resolution using very deep convolutional networks", in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 1646-1654.
  • VDSR CNN-based super-resolution approach
  • This super-resolution model can be re-trained using compressed content from the HomTex database (M. Afonso, A. Katsenou, F. Zhang, D. Agrafiotis, and D.
  • the perceptual quality of the up-sampled second stream of video frame is assessed at a second assessment module AM2 with respect to a corresponding portion of the incoming stream of video frames using a second perceptual quality metric.
  • the purpose is to determine whether the down-sampling has adversely affected the quality of the video frame.
  • the second perceptual quality metric comprises a perception-based hybrid video quality metric (PVM) as described in F. Zhang and D. Bull, "A perception- based hybrid model for video quality assessment,” IEEE Trans, on Circuits and Systems for Video Technology, vol. 26, no. 6, pp. 1017-1028, 2016.
  • PVM perception-based hybrid video quality metric
  • the method outputs a second output comprising the spatially down-sampled video frames if the perceptual quality assessment is for example greater than a second threshold value and outputting a non-spatially down-sampled stream of video frames if the perceptual quality assessment is less than a threshold value.
  • a pre-determined threshold e.g. 53 out of 100
  • the down- sampling is considered as visually lossless, and the frame at lower resolution is kept. Otherwise, the frame with original spatial resolution (after temporal down-sampling) is stored.
  • various spatial sample factors e.g. 1, 1.5, 2 and 4
  • the down-sampling using the highest factor can be applied.
  • a processed video stream is constructed using the first and second outputs to create a third output.
  • the temporal output serves as the input for the spatial processing
  • the second output is the processed video stream.
  • the steps of temporally and spatially down-sampling can comprise generating resolution flag bits which are provided to the first and second assessment modules.
  • the third output can therefore comprise the stream of processed video frames and the generated resolution flag bits.
  • the processed video stream can be stored in a computer memory device M.
  • the stored, processed video stream can be spatially up-sampled at module SUS2.
  • Spatial up-sampling can be achieved using a modified CNN-based super- resolution approach (VDSR) such as that described in J. Kim, J. K. L, and K. M. L, "Accurate image super-resolution using very deep convolutional networks", in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 1646-1654.
  • VDSR CNN-based super- resolution approach
  • the spatially up-sampled video frames can be temporally up-sampled at module TUS2.
  • Temporal up-sampling can for example be achieved by repeating frames.
  • modules TDS, TUS, AM I, SDS, SUS and AM2 together define a selective down- sampling system SI which can reduce the memory consumption in raw data storage through spatial and/or temporal down-sampling,
  • the re-sampled video content is assessed by perceptual quality metrics, which ensure the adaptation process is visually lossless.
  • Modules SUS2 and TUS2 together define an up-sampling system S2 for up-sampling a video stream processed by the selective down-sampling system SI .
  • the selective down-sampling system SI, memory M and up-sampling system S2 can be implemented on separate computing devices, or the same computing device.
  • FIG. 3 is a flow chart illustrating a computer implemented video processing method 50 according to a further embodiment of the invention.
  • FIG. 4 is a system diagram of a system arranged to perform the method 50.
  • the method 50 utilised a model which predicts optimum QP and resolution values.
  • the model is parameterised ( , ⁇ , ⁇ in eqn. (1) below) by training on representative content.
  • the rate-quality performance of a video coding algorithm is key to achieving the best quality at any given bit rate or alternatively, minimising the bit rate for a given quality target.
  • the method utilises a Quantisation-Resolution Optimisation (QRO) module S3, which adaptively selects optimal spatial and temporal resolutions based on the quantisation parameters used in compression. Through the optimum trade-off between resolution and quantisation, the compression efficiency can be significantly improved.
  • QRO Quantisation-Resolution Optimisation
  • Steps 12 to 24 of the method 50 can be identical to steps 12 to 24 of the method 10 of FIG. 1 except that the incoming stream of full resolution video frames are provided by a source S such as a digital video camera or digital video storage device. This could be uncompressed video or previously compressed video content (transcoding).
  • modules TDS, TUS, AM I, SDS and SUS form part of the QRO module S3.
  • Full resolution video frames are first sent to the QRO module S3, where the temporal resolution is firstly optimised according to the quantisation levels.
  • Temporal down-sampling and up-sampling are applied as described with reference to FIGS. 1 and 2.
  • a quality metric which can detect temporal artefacts (e.g. blur) is then employed with a threshold (e.g.
  • each frame of the output from AM I is down-sampled to a lower spatial resolution, and then up-sampled to its original resolution using the same sample factor.
  • sample rate factor of two is used, although other ratios can be used.
  • OSRD Optimum Spatial Resolution Determination module
  • PSNR S Peak Signal to Noise Ratio
  • maxNCC normalised cross correlation between the current frame and the next consecutive frame
  • Equation (2) If resolution down-sampling has been applied, the quantisation parameter is also updated through equation (2) below in order to achieve a quality which is similar to using original resolution and QP. Equation (2) :
  • ⁇ 3 ⁇ 4 .5K . is the updated quantisation parameter used for compressing lower resolution content.
  • K has a constant value of 6, which is based on the training results using the HomTex and the Harmonic video databases.
  • step 54 the low resolution frames are sent to the host encoder for compression instead of the original frames, and the generated bitstream is combined with resolution flag bits.
  • modules TDS, TUS, AM I, SDS, SUS and OSRD together define the QRO module S3 which outputs spatially and temporally down-sampled video frames (according to the spatial and temporal decision processes) along with resolution flag bits and an updated quantisation parameter (QPnew) which are sent to the encoder for compression.
  • QRO module S3 which outputs spatially and temporally down-sampled video frames (according to the spatial and temporal decision processes) along with resolution flag bits and an updated quantisation parameter (QPnew) which are sent to the encoder for compression.
  • the QR optimised bitstream can be transmitted through various channels to the consumers for example.
  • the transmitted bitstream is first decoded as video frames at the various spatial and temporal transmitted resolutions.
  • step 60 these can then be further up-sampled to full resolution for display according to the resolution flag bits at modules SUS2 and TUS2 for display on a display device W.
  • modules SUS2 and TUS2 together define an up-sampling system S4 for up- sampling decoded version of the video stream processed by the selective down- sampling system S3.
  • the selective down-sampling system S2, encoder E can be implemented on separate computing devices, or the same computing device.
  • the decoder D and up- sampling system S2 can be implemented on separate computing devices, or the same computing device.
  • equation (1) and (2) are examples of many possible decision equations. Equally, the parameters may vary if other training databases or configurations are employed. As with the method 10, above-described order of spatiotemporal re-sampling can be reversed or combined in a single operation for region/volume processing rather than whole frame processing.
  • the resolution adaptation method 10 of FIG. 1 was firstly tested for visually-lossless video acquisition applications, on the BVI-HFR database.
  • the number of frames saved for all 120fps sequences are shown in FIG. 5.
  • temporal resolution adaptation was enabled.
  • the reconstructed (after temporally up-sampling) video exhibits equivalent perceptual quality as the original, according to the results of a double stimulus based subjective experiment. It should also be noted that this figure is highly content dependent.
  • FIGS. 6 and 7 the method of FIG. 4 was evaluated for video compression on a dataset containing nine lower frame rate sequences (50/60 Hz), which have been specified for the IEEE ICIP 2017 Video Compression Grand Challenge ("Grand challenge at ICIP 2017: Video compression technology," 2017. [Online]. Available: http://www. provision-itn.eu/grand-challenge-videocompression-icip2017.htm). These comprise both HD and 4K content from the JVET (Joint Video Exploration Team) UHD test set and the BVI Texture database. For evaluation and demonstration, the system utilised the HEVC reference codec (HM 16.14). It can be observed in FIGS.
  • HM 16.14 the HEVC reference codec
  • the resolution adaptation framework dynamically applies spatial and temporal down-sampling to frames exhibiting relatively low spatial detail and lower motion.
  • visually-lossless spatiotemporal resolution adaptation is achieved by assessing the perceptual quality of a down-sampled video sequence using two perceptual quality metrics such as FRQM and PVM.
  • two perceptual quality metrics such as FRQM and PVM.
  • the resolution adaptation approach as well as providing a basis for compression over delivery networks, can also be integrated into future camera architectures, enabling much more efficient data acquisition, and storage of studio-quality content.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

Un procédé de traitement vidéo mis en œuvre par ordinateur comprend un sous-échantillonnage spatio-temporel adaptatif au contenu sur la base d'une mesure de dégradation de qualité due à un sous-échantillonnage.
PCT/GB2018/052604 2017-09-14 2018-09-13 Sous-échantillonnage spatio-temporel de signaux vidéo numériques Ceased WO2019053436A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
GBGB1714791.9A GB201714791D0 (en) 2017-09-14 2017-09-14 Video processing method
GB1714791.9 2017-09-14

Publications (1)

Publication Number Publication Date
WO2019053436A1 true WO2019053436A1 (fr) 2019-03-21

Family

ID=60159628

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/GB2018/052604 Ceased WO2019053436A1 (fr) 2017-09-14 2018-09-13 Sous-échantillonnage spatio-temporel de signaux vidéo numériques

Country Status (2)

Country Link
GB (1) GB201714791D0 (fr)
WO (1) WO2019053436A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115225961A (zh) * 2022-04-22 2022-10-21 上海赛连信息科技有限公司 一种无参考网络视频质量评价方法和装置
EP4167571A1 (fr) * 2021-10-13 2023-04-19 Disney Enterprises, Inc. Échantillonneur avec compensation de distorsion de débit de codec

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4985767A (en) * 1988-02-23 1991-01-15 U. S. Philips Corporation Spatio-temporal sub-sampling of digital video signals representing a succession of interlaced or sequential images, transmission of high-definition television images, and emission and reception stages for such a system
EP0731608A2 (fr) * 1995-03-10 1996-09-11 Sharp Kabushiki Kaisha Codeur et décodeur d'image avec sélection d'une région
US20060188014A1 (en) * 2005-02-23 2006-08-24 Civanlar M R Video coding and adaptation by semantics-driven resolution control for transport and storage
WO2009014156A1 (fr) * 2007-07-20 2009-01-29 Fujifilm Corporation Appareil de traitement d'image, procédé et programme de traitement d'image
US20160269733A1 (en) * 2015-03-10 2016-09-15 Apple Inc. Adaptive Chroma Downsampling and Color Space Conversion Techniques

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4985767A (en) * 1988-02-23 1991-01-15 U. S. Philips Corporation Spatio-temporal sub-sampling of digital video signals representing a succession of interlaced or sequential images, transmission of high-definition television images, and emission and reception stages for such a system
EP0731608A2 (fr) * 1995-03-10 1996-09-11 Sharp Kabushiki Kaisha Codeur et décodeur d'image avec sélection d'une région
US20060188014A1 (en) * 2005-02-23 2006-08-24 Civanlar M R Video coding and adaptation by semantics-driven resolution control for transport and storage
WO2009014156A1 (fr) * 2007-07-20 2009-01-29 Fujifilm Corporation Appareil de traitement d'image, procédé et programme de traitement d'image
US20160269733A1 (en) * 2015-03-10 2016-09-15 Apple Inc. Adaptive Chroma Downsampling and Color Space Conversion Techniques

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4167571A1 (fr) * 2021-10-13 2023-04-19 Disney Enterprises, Inc. Échantillonneur avec compensation de distorsion de débit de codec
JP2023058466A (ja) * 2021-10-13 2023-04-25 ディズニー エンタープライゼス インコーポレイテッド コーデックレート歪み補償ダウンサンプラ
US11765360B2 (en) 2021-10-13 2023-09-19 Disney Enterprises, Inc. Codec rate distortion compensating downsampler
JP7419478B2 (ja) 2021-10-13 2024-01-22 ディズニー エンタープライゼス インコーポレイテッド コーデックレート歪み補償ダウンサンプラ
CN115225961A (zh) * 2022-04-22 2022-10-21 上海赛连信息科技有限公司 一种无参考网络视频质量评价方法和装置
CN115225961B (zh) * 2022-04-22 2024-01-16 上海赛连信息科技有限公司 一种无参考网络视频质量评价方法和装置

Also Published As

Publication number Publication date
GB201714791D0 (en) 2017-11-01

Similar Documents

Publication Publication Date Title
US11810332B2 (en) Apparatus and method for performing artificial intelligence (AI) encoding and AI decoding on image
KR102500761B1 (ko) 영상의 ａｉ 부호화 및 ａｉ 복호화 방법, 및 장치
CN107211128B (zh) 自适应色度下采样和色彩空间转换技术
EP4365820A1 (fr) Réseau de super-résolution vidéo, et procédé et dispositif de traitement de codage, décodage et super-résolution vidéo
EP2774370B1 (fr) Décomposition de couche dans un codage vdr hiérarchique
KR20150010903A (ko) 모바일 단말 화면을 위한 3ｋ해상도를 갖는 디스플레이 영상 생성 방법 및 장치
US11601617B2 (en) Method for forming an output image sequence from an input image sequence, method for reconstructing an input image sequence from an output image sequence, associated devices, server equipment, client equipment and computer programs
US10616498B2 (en) High dynamic range video capture control for video transmission
KR102287942B1 (ko) 전처리를 이용한 영상의 ａｉ 부호화 및 ａｉ 복호화 방법, 및 장치
KR102312337B1 (ko) Ai 부호화 장치 및 그 동작방법, 및 ai 복호화 장치 및 그 동작방법
KR20150068402A (ko) 동영상 압축 방법
EP3938998B1 (fr) Élimination de bandes à l'aide d'une nouvelle métrique de bandes
US11297353B2 (en) No-reference banding artefact predictor
WO2019053436A1 (fr) Sous-échantillonnage spatio-temporel de signaux vidéo numériques
CN110582022B (zh) 视频编解码方法和装置及存储介质
US12058340B2 (en) Video compression technique using a machine learning system
US20080095464A1 (en) System and Method for Representing Motion Imagery Data
EP4672738A1 (fr) Traitement sélectif de trames dans un pipeline de codage à base de coups
EP3185554A1 (fr) Dispositifs de codage et de reconstruction vidéo avec quantification adaptative
Kuhmünch et al. Video-scaling algorithm based on human perception for spatiotemporal stimuli
CN110677653A (zh) 视频编解码方法和装置及存储介质
Masala et al. Content-based group-of-picture size control in distributed video coding
KR20210154700A (ko) 화상통화시 업스케일을 위한 적응적 인공지능 다운스케일 방법 및 장치
Matsuo et al. Video coding of 8K UHDTV by HEVC/H. 265 with spatio-gradational reduction and its restoration
WO2021057676A1 (fr) Procédé et appareil de codage de vidéo, procédé et appareil de décodage de vidéo, dispositif électronique et support de stockage lisible

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18773565

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18773565

Country of ref document: EP

Kind code of ref document: A1