WO2018036481A1 - 一种检测场景切换帧的方法、装置和系统 - Google Patents

一种检测场景切换帧的方法、装置和系统 Download PDF

Info

Publication number
WO2018036481A1
WO2018036481A1 PCT/CN2017/098483 CN2017098483W WO2018036481A1 WO 2018036481 A1 WO2018036481 A1 WO 2018036481A1 CN 2017098483 W CN2017098483 W CN 2017098483W WO 2018036481 A1 WO2018036481 A1 WO 2018036481A1
Authority
WO
WIPO (PCT)
Prior art keywords
frame
threshold
scene switching
gop
frames
Prior art date
Application number
PCT/CN2017/098483
Other languages
English (en)
French (fr)
Inventor
熊婕
杨友庆
黄一宏
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP17842903.1A priority Critical patent/EP3499460A1/en
Priority to KR1020197007484A priority patent/KR20190039265A/ko
Priority to JP2019510927A priority patent/JP2019528643A/ja
Publication of WO2018036481A1 publication Critical patent/WO2018036481A1/zh
Priority to US16/284,664 priority patent/US10917643B2/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—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
    • H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
    • H04N19/159—Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
    • 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
    • H04N19/142—Detection of scene cut or scene change
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
    • 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
    • H04N19/103—Selection of coding mode or of prediction mode
    • H04N19/114—Adapting the group of pictures [GOP] structure, e.g. number of B-frames between two anchor frames
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/146—Data rate or code amount at the encoder output
    • 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
    • H04N19/48—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using compressed domain processing techniques other than decoding, e.g. modification of transform coefficients, variable length coding [VLC] data or run-length data
    • 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
    • H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
    • H04N19/87—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving scene cut or scene change detection in combination with video compression

Definitions

  • the present invention relates to the field of video technologies, and in particular, to a method, device, and system for detecting a scene switching frame.
  • IPTV Internet Protocol Television
  • OTT Internet Protocol Television
  • a video consists of a sequence of consecutive video frames, usually including more than one scene.
  • a video consists of 4 scenes, scene 1 and scene 3 correspond to the video of the football field, and scene 2 and scene 4 correspond to the scene.
  • Video of the auditorium is a video of the auditorium.
  • the video coding impairment caused by encoding the video is not only related to the video coding type, frame rate, resolution, and code rate, but also related to the complexity of the scene. Therefore, when evaluating video coding damage, it is necessary to detect the scene first. The location where the switch occurred. Therefore, scene change detection must be very accurate when evaluating video quality.
  • each video frame is encoded into a different type of frame, such as an I frame, a P frame, or a B frame.
  • the I frame is an intra prediction coded frame, and only the data in the frame is referenced during encoding
  • the P frame is a forward predictive coded frame, that is, a unidirectional difference frame, and the frame is recorded with an I frame (or P)
  • the B frame is a bidirectionally predicted interpolated coded frame, that is, a bidirectional difference frame, and the difference between the current frame and the previous frame and the subsequent frame is recorded.
  • the IPTV monitoring scheme provided by the standard ITU-T P1201.2 provides a method for detecting a scene switching frame.
  • the prior art only detects a scene switching frame in an I frame, but in reality, many scene switching frames are P frames. Therefore, detecting a scene switching frame according to the prior art may result in a missed detection.
  • the embodiment of the invention provides a method and a device for detecting a scene switching frame, which are used to solve the problem of the missed scene switching frame existing in the prior art.
  • a first aspect provides a method for detecting a scene switching frame, where the video includes N image groups GOP, and N is an integer greater than or equal to 2.
  • the method includes:
  • the relative size is greater than or equal to the first threshold, and with In the case where the relative size is greater than or equal to the second threshold, it is determined that Pmax is a scene switching frame, wherein The median or average value of the size of multiple I frames between frames where Pmax is closest to Pmax before Pmax, a median or average value of sizes of a plurality of P frames in the Kth GOP; the first threshold is greater than 0 and less than 1, and the second threshold is greater than 1.
  • a scene switching frame in the P frame when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • I threshold is the first threshold
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video.
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • Pm, ..., P-1 represents the P frame before Pmax in the Kth GOP
  • P1 ..., Pn represents the P frame after Pmax in the Kth GOP
  • F is used to find Pm, ..., P-1, The median or average value of the size of P1,...,Pn;
  • n min(num_after_P_frames,max_num)
  • num_before_P_frames is the number of P frames before Pmax in the Kth GOP
  • num_after_P_frames is the number of P frames after Pmax in the Kth GOP
  • max_num represents the preset number of frames to be considered.
  • a method for implementing video quality assessment where the video includes N image groups GOP, N is an integer greater than or equal to 2, and the method includes:
  • the relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to the second threshold, and there is no B frame in the Kth GOP, or is determined with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to a second threshold, and the Kth GOP has a B frame, and with
  • Pmax is a scene switching frame, wherein The median or mean value of the size of multiple I frames between the scene switching frames closest to Pmax before Pmax and Pmax, Is the median or average value of the sizes of the plurality of P frames in the Kth GOP, a median or average of the sizes of all B frames between the scene switching frames closest to Pmax before Pmax and Pmax; the first threshold is greater than 0 and less than 1, the second threshold is greater than 1, the third The threshold is greater than 1.
  • a scene switching frame in a P frame when detecting a scene switching frame in a video, a scene switching frame in a P frame can be detected, thereby effectively reducing a problem of a missed scene switching frame, and in detecting a P frame.
  • the size of the I frame and the P frame is considered, and the size of the B frame is also considered, thereby further improving the accuracy of detecting the scene switching frame in the P frame.
  • the first threshold is calculated.
  • I threshold is the first threshold
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video.
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • Pm, ..., P-1 represents the P frame before Pmax in the Kth GOP
  • P1 ..., Pn represents the P frame after Pmax in the Kth GOP
  • F is used to find Pm, ..., P-1, The median or average value of the size of P1,...,Pn;
  • n min(num_after_P_frames,max_num)
  • num_before_P_frames is the number of P frames before Pmax in the Kth GOP
  • num_after_P_frames For the number of P frames after Pmax in the Kth GOP
  • max_num indicates the preset number of frames to be considered.
  • B threshold is the third threshold
  • P median is the median or average value of the sizes of all P frames in the video
  • B median is the median or average of the sizes of all B frames in the video.
  • B median is the median or average of the sizes of all B frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the video is determined to be a scene switching frame.
  • the median or average value of the size of other P frames other than the P frame as the new P median and according to the formula or A new B theeshold is calculated, and the new B threshold is used to determine whether the Pmax of the next GOP is a scene switching frame.
  • the third threshold Bthreshold can be updated in real time, and the influence of the P frame that has been determined to be the scene switching frame is excluded in time, and therefore, the scene switching frame in the P frame can be further improved. Accuracy.
  • a third aspect provides a detecting apparatus for detecting a scene switching frame in a video, where the video includes N GOPs, N is an integer greater than or equal to 2, and the detecting apparatus includes: a first determining unit and a second Determining unit;
  • the first determining unit is configured to determine a maximum P frame Pmax among all P frames of the Kth GOP, where the size of Pmax is Where K is a variable and the values are sequentially from M to N, 1 ⁇ M ⁇ N;
  • the second determining unit is configured to determine with The relative size is greater than or equal to the first threshold, and with In the case where the relative size is greater than or equal to the second threshold, it is determined that Pmax is a scene switching frame, wherein The median or average value of the size of multiple I frames between frames where Pmax is closest to Pmax before Pmax, a median or average value of sizes of a plurality of P frames in the Kth GOP; the first threshold is greater than 0 and less than 1, and the second threshold is greater than 1.
  • the scene switching frame in the video when the scene switching frame in the video is detected, the scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • the second determining unit may be specifically according to a formula or Calculation with Relative size
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • I threshold is the first threshold
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video.
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the second determining unit may be specifically according to a formula Calculation with Relative size formula Can be reflected simply and effectively with The relative size relationship.
  • the second determining unit may be specifically calculated according to the following formula
  • P_m,..., P-1 represents the P frame before Pmax in the Kth GOP
  • P1,..., Pn represents the P frame after Pmax in the Kth GOP
  • F is used to find P_m,..., P-1, The median or average value of the size of P1,...,Pn;
  • n min(num_after_P_frames,max_num)
  • num_before_P_frames is the number of P frames before Pmax in the Kth GOP
  • num_after_P_frames is the number of P frames after Pmax in the Kth GOP
  • max_num represents the preset number of frames to be considered.
  • a fourth aspect provides a detecting apparatus for detecting a scene switching frame in a video, where the video includes N image groups GOP, N is an integer greater than or equal to 2, and the detecting apparatus includes: a first determining unit and a second Determining unit;
  • the first determining unit is configured to determine a maximum P frame Pmax among all P frames of the Kth GOP, where the size of Pmax is Where K is a variable and the values are sequentially from M to N, 1 ⁇ M ⁇ N;
  • the second determining unit is configured to determine with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to the second threshold, and there is no B frame in the Kth GOP, or is determined with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to a second threshold, and the Kth GOP has a B frame, and with In the case where the relative size is greater than or equal to the third threshold, it is determined that Pmax is a scene switching frame, wherein The median or mean value of the size of multiple I frames between the scene switching frames closest to Pmax before Pmax and Pmax, Is the median or average value of the sizes of the plurality of P frames in the Kth GOP, a median or average of the sizes of all B frames between the scene switching frames closest to Pmax before Pmax and Pmax; the first threshold is greater than 0 and less than 1, the second threshold is greater than 1, the third The threshold is greater than 1.
  • a scene switching frame in a P frame when detecting a scene switching frame in a video, a scene switching frame in a P frame can be detected, thereby effectively reducing a problem of a missed scene switching frame, and detecting a P frame.
  • the scene switching frame not only the size of the I frame and the P frame but also the size of the B frame is considered, thereby further improving the accuracy of detecting the scene switching frame in the P frame.
  • the second determining unit may be specifically according to a formula or Calculation with Relative size
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • I threshold is the first threshold
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video.
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the second determining unit may be specifically according to a formula Calculation with Relative size formula Can be reflected simply and effectively with The relative size relationship.
  • the second determining unit may be specifically calculated according to the following formula
  • P_m,..., P-1 represents the P frame before Pmax in the Kth GOP
  • P1,..., Pn represents the P frame after Pmax in the Kth GOP
  • F is used to find P_m,..., P-1, The median or average value of the size of P1,...,Pn;
  • n min(num_after_P_frames,max_num)
  • num_before_P_frames is the number of P frames before Pmax in the Kth GOP
  • num_after_P_frames is the number of P frames after Pmax in the Kth GOP
  • max_num represents the preset number of frames to be considered.
  • the second determining unit may be specifically according to a formula or Calculation with Relative size
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • the second determining unit is according to a formula Calculation
  • the second determining unit can be based on a formula The first threshold is calculated.
  • I threshold is the first threshold
  • I median is the median or average of the sizes of all I frames in the video
  • B median is the median or average of the sizes of all B frames in the video.
  • I median is the median or average of the sizes of all I frames in the video
  • B median is the median or average of the sizes of all B frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the video is determined to be a scene switching frame.
  • the median or average value of the size of other P frames other than the P frame as the new P median and according to the formula or A new B threshold is calculated, and the new B threshold is used to determine whether the Pmax of the next GOP is a scene switching frame.
  • the third threshold Bthreshold can be updated in real time, and the influence of the P frame that has been determined to be the scene switching frame is eliminated in time, and therefore, the scene switching frame in the P frame can be further improved. Accuracy.
  • a fifth aspect provides a detecting apparatus for detecting a scene switching frame in a video, the detecting apparatus comprising a processor and a memory;
  • the memory is configured to store computer operation instructions
  • the processor configured to execute computer operating instructions stored in the memory, such that the detecting device performs any one of the first aspect or the first aspect, the second aspect, or the second aspect of the present invention Possible implementations provide the methods.
  • a scene switching frame in the P frame when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • a detecting device comprising a media unit and a detecting device
  • the media unit is configured to acquire the video and transmit the video to the detecting device;
  • the detecting device configured to acquire the video from the media unit, and perform any possible implementation manner of the third or third aspect of the present invention, the fourth aspect, or any possible implementation manner of the fourth aspect
  • the operation performed by the detecting device provided by any of the possible implementations of the fifth aspect or the fifth aspect.
  • a scene switching frame in the P frame when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • a seventh aspect a system for implementing video quality assessment, the system comprising a video server, a transmission device, and a video terminal, wherein a video stream sent by the video server is transmitted to the video terminal via the transmission device;
  • the transmission device or the video terminal may specifically include any possible implementation manner of the third or third aspect of the present invention, any possible implementation manner of the fourth aspect or the fourth aspect, the fifth aspect or the fifth a detection device provided by any of the possible implementations of the aspect; or
  • the system further includes a first detecting device, which may be any possible implementation manner of the third or third aspect of the present invention, any possible implementation of the fourth aspect or the fourth aspect.
  • the detecting device provided by any of the possible implementations of the fifth aspect or the fifth aspect, wherein the transmitting device 2020 or the video terminal 2030 is connected to the first detecting device, and the first detecting device passes The transmission device or the video terminal connected to the first detecting device acquires the video stream.
  • a scene switching frame in the P frame when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • FIGS. 1A and 1B are schematic diagrams showing the networking structure of a video system 100 according to Embodiment 1 of the present invention.
  • FIGS. 2A, 2B and 2C are schematic diagrams of a GOP provided by Embodiment 1 of the present invention.
  • 3A, 3B, and 3C are schematic flowcharts of a method provided by Embodiment A of Embodiment 1 of the present invention.
  • 4A, 4B, and 4C are schematic flowcharts of a method provided by Embodiment B of Embodiment 1 of the present invention.
  • 5A, 5B, and 5C are schematic diagrams of examples of GOPs provided by Embodiment 1 of the present invention.
  • FIG. 6 is a schematic structural diagram of a detecting apparatus 200 according to Embodiment 2 of the present invention.
  • FIG. 7 is a schematic structural diagram of a detecting apparatus 1000 according to Embodiment 3 of the present invention.
  • FIG. 8 is a schematic structural diagram of a detecting apparatus 400 according to Embodiment 4 of the present invention.
  • 9A-9C are schematic diagrams showing the structure of a system 2000 according to Embodiment 5 of the present invention.
  • FIG. 1A is a schematic diagram of a network structure of a video system 100 according to an embodiment of the present invention.
  • the video system 100 includes a video server 110, one or more transmission devices 120, and a video terminal 130.
  • the video stream sent by the video server 110 is used.
  • the transmission device 120 transmits to the video terminal 130.
  • the video system 100 may specifically be an IPTV system as shown in FIG. 1B.
  • the video server 110 is specifically a video head end (video HE)
  • the transmission device 120 specifically includes a core router (Core Router, CR), a broadband network gateway (BNG), and an optical line terminal. (Optical Line Terminal, OLT) and other network devices
  • the video terminal 130 is specifically a Set Top Box (STB).
  • STB Set Top Box
  • the video stream when the video stream is transmitted from the video server to the video terminal, the video stream may be lost, delayed, jittered, out of order, etc. due to changes in the network state. unusual phenomenon. These anomalies may cause defects such as blooming, jamming, etc. on the video screen played on the screen of the video terminal, resulting in a decrease in the video viewing experience of the user. Therefore, the user's video experience needs to be monitored through video quality assessment.
  • the video coding impairment caused by encoding the video is not only related to the video coding type, frame rate, resolution, and code rate, but also related to the complexity of the scene. Therefore, when evaluating video coding damage, it is necessary to detect the scene first. The location where the switch occurred.
  • the decoder of the video terminal when a packet loss occurs during video transmission, usually uses the video content of the corresponding area in the previous frame of the damaged frame as the content of the damaged area in the damaged frame to perform the damaged frame.
  • Error compensation so the smaller the content gap between the damaged frame and the previous frame, the better the compensation effect, and when the damaged frame is a scene switching frame, since the scene switching frame is almost completely different from the content of the previous frame, at this time The compensation effect is the worst. Therefore, when affecting the video quality caused by packet loss, it is necessary to consider whether the damaged frame is a scene switching frame.
  • the encoded video frame sequence is composed of a plurality of Group of Picture (GOP), as shown in FIG. 2A, each GOP starts with an I frame, followed by some P frames and B frames, to the next I. The previous frame of the frame ends.
  • the I frame is an intra-coded frame
  • the P frame is a forward reference frame
  • the B frame is a bidirectional reference frame.
  • the intra prediction encoding is often used in encoding (the macroblocks in the scene switching frame mostly use intra prediction encoding)
  • the sixth frame in GOP1 is a scene switching frame
  • the fourth frame is subjected to the fourth frame.
  • a video with a duration of T (such as 10 seconds) includes a video frame 1 at time t1 and a video frame 2 at time t2, if t1 is less than t2, such as t1 is 1. 30 milliseconds in seconds, t2 is 5 seconds and 40 milliseconds, and video frame 1 is considered to be in front of video frame 2.
  • a detection device for implementing scene switching frame detection may be deployed in the video system.
  • the detecting device may be deployed on any device (such as the transmitting device 120 and the video terminal 130) through which the video stream passes, or may be connected to any device through which the video stream passes and obtain the video stream by mirroring.
  • FIGS. 1A and 1B are schematic flowcharts of a method according to Embodiment 1 of the present invention.
  • the method of Embodiment 1 of the present invention can be applied to the video system 100 shown in FIGS. 1A and 1B, and is executed by a detecting apparatus.
  • a scene switching frame in a video (hereinafter referred to as a video to be inspected) is detected.
  • the to-be-checked video may be read from a video file or may be obtained from the acquired video stream.
  • the video to be detected may be a complete video or a video segment in a video.
  • a measurement time window is usually set to detect the video segment in the measurement time window. For example, when detecting a video, the length of the measurement time window is set to 10 seconds. The video segment of 0-10 seconds of the video is detected as the video to be inspected, and then the video segment of 10-20 seconds is detected as the video to be inspected, and so on.
  • the detecting module may first determine the type (such as I frame, P frame, B frame) and size of each video frame in the to-be-detected video.
  • the video stream is acquired in real time, and the information of the video frame is extracted from the packet message of the video stream corresponding to the measurement time window (for example, 10-20 seconds), and the size (in bytes) of the video frame is calculated.
  • the process of calculating the size of the video frame is to cycle all the packet messages in the measurement time window, first find the start identifier of the current video frame from the header of the packet message, and then load the length of the packet including the start identifier and The load length of subsequent packet messages is accumulated until the start identifier of the next video frame is found, and the summation is the size of the current video frame.
  • a specific implementation of calculating the size of a video frame can be found in the standard ITU-T P1201.2.
  • the type of the video frame may be determined according to the field random_access_indicator in the packet header of the packet message. For I frames, the type of the frame can be determined by random_access_indicator whether encrypted or not. For non-I frames, when the video is not encrypted, the frame type can be directly obtained from the frame header of the video frame. When the video is encrypted, or when the frame header of the video frame is lost, the frame time or the display timestamp of the frame may be firstly used.
  • the Present Time Stamp estimates the GOP mode, which is generally P B B P B B or P B B B P B B. The difference between the current PTS value and the previous PTS value can describe the mode of the GOP. Once the GOP mode is determined, the mode of all lost or encrypted video frames can be determined.
  • a specific implementation of determining the type of video frame can be found in the standard ITU-T P1201.2.
  • the video to be tested can be divided into several GOPs.
  • a video to be inspected usually includes a plurality of GOPs, assuming N, and N is an integer greater than or equal to 2, as exemplified in FIG. 5A.
  • the video frame filled with the solid color shadow is an I frame
  • the video frame filled with the slanted shadow is a P frame
  • the video frame without the fill color is a B frame.
  • a video frame is an image.
  • each video frame is encoded into different types of frames, such as an I frame, a P frame, and a B frame.
  • the I frame is an intra prediction coded frame, and only the data in the frame is referenced during encoding, so the complete image data is included
  • the P frame is a forward predictive coded frame, that is, a one-way difference frame, and the frame is recorded before and The difference of an I frame (or P frame);
  • the B frame is a bidirectionally predicted interpolated coded frame, that is, a bidirectional difference frame, recorded This is the difference between this frame and the previous frame and the next frame.
  • I frames are generally larger than P frames, and P frames are generally larger than B frames.
  • the I frame size is 2-5 times the size of the P frame, and the P frame size is 2-5 times the size of the B frame.
  • the scene switching frame has a large difference from the content of the previous frame, even if the scene switching frame is encoded as a P frame, the macroblocks in the scene switching frame are mostly referred to by reference to other macroblocks in the frame. Encoding, so it will be bigger. If the size of the P frame exceeds half of the I frame, the P frame is likely to be a scene switching frame. Therefore, when detecting a scene switching frame of the P frame type, the relative size relationship between the P frame and the I frame can be referred to.
  • the content of the video in the video changes relatively fast, such as a fierce soccer game such as shooting a relatively intense scene
  • the correlation between the before and after video frames is small
  • the size of the P frame is also relatively large, and even exceeds half the size of the I frame.
  • the size difference between adjacent P frames of the scene switching frame is not large, but the size difference between the P frame of the scene switching frame and the P frame of the adjacent scene switching frame is relatively large. Therefore, when detecting a scene switching frame of a P frame type, the ratio of the sizes between adjacent P frames can also be referred to.
  • the embodiment A of the embodiment 1 of the present invention is described in detail below with reference to FIG. 3A to detect a scene switching frame in the video to be inspected.
  • the video includes N GOPs, and N is an integer greater than or equal to 2.
  • the video to be inspected may be a video in the video file, or a video in the video stream, such as a video in the video stream sent by the video server to the video terminal. Accordingly, the detecting device may be built in. Any of the devices (such as the transmission device 120 and the video terminal 130) through which the video stream passes may be bypassed to any device through which the video stream passes, and the video stream is obtained by mirroring.
  • the method provided in Embodiment A of Embodiment 1 of the present invention includes:
  • the detecting device performs an operation on each GOP in order to determine whether the largest P frame Pmax among all P frames in each GOP is a scene switching frame, where M is greater than the Mth GOP of the N GOPs, where M is greater than Equal to 1 and less than or equal to N.
  • the detection can be started from the first GOP, that is, M is equal to 1. It is also possible to start detection from a GOP after the first GOP.
  • the video to be inspected is a video starting from the beginning of a video stream. Since the frame size in the first two GOPs is generally less informative, usually from the first Three GOPs start detecting, then M equals 3.
  • Step 102 Determine a maximum P frame Pmax among all P frames of the Kth GOP, where the size of Pmax is Where K is a variable and the values are from M to N, 1 ⁇ M ⁇ N.
  • Step 103 In the determination with The relative size is greater than or equal to the first threshold, and with In the case where the relative size is greater than or equal to the second threshold, it is determined that Pmax is a scene switching frame, wherein The median or average value of the size of multiple I frames between frames where Pmax is closest to Pmax before Pmax, a median or average value of sizes of a plurality of P frames in the Kth GOP; the first threshold is greater than 0 and less than 1, and the second threshold is greater than 1.
  • the second P frame in the Kth GOP is Pmax
  • the I frame of the Kth GOP is not a scene switching frame
  • the K-2th GOP The third P frame (hereinafter referred to as P'max) is a scene switching frame
  • the scene switching frame closest to Pmax before Pmax is P'max.
  • the I frame of the first GOP may be determined as a scene switching frame, and when there is no scene switching frame between the Pmax of the Kth GOP and the I frame of the first GOP, Pmax The scene switching frame closest to Pmax is the I frame of the first GOP. If the video to be inspected is a video segment in a video and the video segment of the scene switching frame has been detected before the video to be detected (hereinafter referred to as the previous video segment), the scene switching frame closest to Pmax before Pmax may be located. In the previous video segment.
  • the median or average value of the size of all or part of the I frame between frames can be switched using the scene closest to Pmax before Pmax and Pmax.
  • the second P frame in the Kth GOP is Pmax
  • the scene switching frame closest to Pmax before Pmax is P'max
  • two I frames Ik and I are included between Pmax and P'max.
  • K-1 Refers to the average of the sizes of I k and I k-1 .
  • a preferred embodiment is as follows:
  • Pm, ..., P-1 represents the P frame before Pmax in the Kth GOP
  • P1 ..., Pn represents the P frame after Pmax in the Kth GOP
  • F is used to find Pm, ..., P-1, The median or average value of the size of P1,...,Pn;
  • n min(num_after_P_frames,max_num)
  • num_before_P_frames is the number of P frames before Pmax in the Kth GOP
  • num_after_P_frames is the number of P frames after Pmax in the Kth GOP
  • max_num represents the preset number of frames to be considered. As shown in FIG. 5C, num_before_P_frames is equal to 7, num_after_P_frames is equal to 4, and max_num is set to 6, where m is equal to 6, and n is equal to 4.
  • the first threshold may be preset, and different GOPs in the same to-be-tested video may adopt the same first threshold. As according to the formula Calculation Setting the first threshold to 0.53, as according to the formula Calculation The first threshold is set to 0.47.
  • the first threshold may also be calculated and dynamically adjusted to further improve the accuracy of the detection.
  • the process of calculating and dynamically adjusting is as follows:
  • the first threshold is calculated according to the following formula:
  • I median is a median or average value of sizes of all I frames in the to-be-detected video
  • P median is a median or average value of sizes of all P frames in the to-be-detected video.
  • the video to be detected includes 9 P frames, and the sizes are 3 5 3 6 4 7 3 5 4, which are arranged in order of size to be 3 3 3 4 4 5 5 6 7 and then P.
  • the median is 4;
  • the video to be tested includes 3 P frames, each of which is 15 12 18 , and is arranged in order of size 12 15 18 , and I median is 15.
  • I median is the median or average of the sizes of all I frames in the video
  • P median is the median or average of the sizes of all P frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the determined video may be removed from the to-be-detected video.
  • the median or average value of the size of other P frames other than the P frame as the new P median and the formula or A new I threshold is calculated, and the new I threshold is used to determine whether the Pmax of the subsequent GOP (such as the next GOP of the Kth GOP) is a scene switching frame.
  • the first threshold I threshold can be updated in real time, the influence of the P frame that has been determined to be the scene switching frame is eliminated in time, and therefore, the accuracy of detecting the scene switching frame in the P frame can be further improved.
  • the specific implementation manner may be: if the first threshold is according to a formula Obtained with Relative size for: If the first threshold is according to a formula Obtained with Relative size for:
  • the second threshold may be preset, and different GOPs in the same to-be-tested video may adopt the same second threshold, such as 1.51.
  • Embodiment A of Embodiment 1 of the present invention may further include step 101.
  • Step 101 Detect a scene switching frame from an I frame of the Mth to the Nth GOP of the N GOPs.
  • step 101 when determining whether the I frame of the Kth GOP is a scene switching frame, specifically, according to the ratio of the size of the I frame of the Kth GOP to the size of the I frame of the K-1th GOP, The ratio of the average value of the sizes of all P frames in the K-1th GOP to the average frame size of all P frames in the Kth GOP, and the size of all B frames in the K-1 GOP. The ratio of the average value to the average frame size of all B frames in the Kth GOP determines whether the I frame of the Kth GOP is a scene switching frame.
  • the specific implementation is as follows:
  • condition (1) and condition (2) are further determined; otherwise, it is determined that the I frame of the Kth GOP is not a scene switching frame.
  • r P is less than the threshold third threshold or r P is greater than the fourth threshold
  • condition (1) and condition (2) are satisfied, it is determined that the I frame of the Kth GOP is a scene switching frame, and otherwise, the I frame of the Kth GOP is determined not to be a scene switching frame.
  • the I frame of the first GOP may be directly determined as a scene switching frame, and when the K is not equal to 1, the above method is used to determine whether the I frame of the Kth GOP is a scene switching frame.
  • implementation manner A may be implemented in implementation manner J: first, step 101 is performed, and then steps 102-103 are performed, that is, first, the first frame of the Mth to the Nth GOP are detected.
  • the scene switching frame re-detects the scene switching frame in the P frame of the Mth GOP to the Nth GOP. For example, first determining the I frame of the GOP1 as a scene switching frame, and then determining whether the I frame of the GOPM (such as GOP1) to the GOPN is a scene switching frame, and then sequentially determining whether the Pmax of the GOPM (such as GOP1) to the GOPN is a scene switching. frame.
  • the implementation manner K may be implemented in the implementation manner: Step 101 is performed in steps 102-103, and the scene switching frame is sequentially detected in the order of the video frame, that is, from the Mth. (such as the first one)
  • the GOP starts to detect the scene switching frame in the current GOP by the GOP according to the order of the GOP, and firstly determines whether the I frame in the current GOP is the scene switching frame and then judges the current when detecting the scene switching frame in the current GOP. Whether Pmax in the GOP is a scene switching frame.
  • the I frame of the GOP1 is first determined as a scene switching frame, and then the Pmax of the GOP1, the Iframe of the GOP2, the Pmax of the GOP2, the Iframe of the GOP3, the Pmax of the GOP3, the I frame of the GOPN, the Pmax of the GOPN, and the Pmax of the GOPN are sequentially determined. Whether to switch frames for the scene. In short, when K is less than N, it is determined whether the I frame of the K+1th GOP is a scene switching frame after determining whether the Pmax of the Kth GOP is a scene switching frame.
  • Embodiment A since the probability that the scene is continuously switched in a short time is low, before determining whether an I frame is a scene switching frame, the I frame and the previous one may be calculated first.
  • the distance between the scene switching frames hereinafter referred to as the first distance
  • the first distance if the first distance is less than or equal to the distance threshold, it is determined that the I frame is not a scene switching frame, otherwise, according to the standard ITU-T P1201.2
  • a method is provided to determine whether the I frame is a scene switching frame.
  • the specific implementation manner is: when K is smaller than N, after determining whether the Pmax of the Kth GOP is a scene switching frame, determining an I frame of the K+1th GOP (hereinafter referred to as a current I frame) and the Determining that the current I frame is not a scene switching frame when the distance between the scene switching frames closest to the current I frame before the current I frame is less than or equal to the distance threshold, otherwise, it may be further provided according to the standard ITU-T P1201.2. The method determines whether the current I frame is a scene switching frame.
  • the distance between two video frames referred to in the present invention refers to how many video frames differ between two video frames, and the distance between the Xth video frame and the Yth video frame is YN, two phases The distance between adjacent video frames is 1.
  • the distance threshold may be preset, and different GOPs in the same to-be-tested video may adopt the same distance threshold.
  • the distance threshold may also be determined and dynamically adjusted to further improve the accuracy of the detection.
  • the process of determining and dynamically adjusting the distance threshold is as follows:
  • an initial distance threshold is determined, and the determined initial distance threshold may be one of the following three lengths:
  • the length of the GOP referred to in the present invention refers to the number of video frames included in the GOP.
  • the to-be-detected video includes 8 GOPs, and the length is 10 6 8 7 8 7 9 8 , and the initial distance threshold determined according to the mode (1) is 10; the initial distance threshold determined according to the mode (2) is 8; The initial distance threshold determined according to mode (3) is 8, because the number of GOPs having a length of 8 is the largest.
  • the length of each GOP is the same, and the initial distance threshold calculated by the above three methods is the same.
  • the distance threshold is updated to the second distance.
  • Embodiment A of Embodiment 1 of the present invention when detecting a scene switching frame in a video, the size of the largest P frame Pmax among all P frames of the GOP of the video is detected. a relative magnitude relationship between the median or average values of the sizes of the plurality of I frames between the scene switching frames closest to Pmax before Pmax and Pmax, and Determining whether Pmax is a scene switching frame to detect a scene switching frame in a P frame, thereby effectively reducing a problem of a missed scene switching frame by determining a relative size relationship between a median or an average value of a plurality of P frames in the GOP.
  • the P frame is generally larger than the B frame, and usually the P frame size is 2-5 times the size of the B frame. Since the scene switching frame has a large difference from the content of the previous frame, even if the scene switching frame is encoded as a P frame, the macroblocks in the scene switching frame mostly use intra prediction encoding, so it is relatively large. If the P frame size is less than 2 times of the B frame, the P frame may be a scene switching frame. Therefore, when detecting a P frame type scene switching frame, the relative size relationship between the P frame and the B frame may be referred to.
  • the embodiment B of the embodiment 1 of the present invention is described in detail below with reference to FIG. 4A to detect a scene switching frame in the video to be inspected.
  • the video includes N GOPs, and N is an integer greater than or equal to 2.
  • the method provided in Embodiment B of Embodiment 1 of the present invention includes:
  • Step 202 Same as step 102, and details are not described herein again.
  • Step 203 in determining with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to the second threshold, and there is no B frame in the Kth GOP, or is determined with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to a second threshold, and the Kth GOP has a B frame, and with
  • Pmax is a scene switching frame, wherein The mean value of the size of multiple I frames between the scene switching frames closest to Pmax before Pmax and Pmax, Is the median or average value of the sizes of the plurality of P frames in the Kth GOP, a median or average of the sizes of all B frames between the scene switching frames closest to Pmax before Pmax and Pmax; the first threshold is greater than 0 and less than 1, the second threshold is greater than 1, the third The threshold is greater than 1.
  • the third threshold may be preset, and different GOPs in the same to-be-tested video may adopt the same third threshold. As according to the formula Calculation When the first threshold is set to 2.87, as according to the formula Calculation The first threshold is set to 1.87.
  • the third threshold may also be calculated and dynamically adjusted to further improve the accuracy of the detection.
  • the process of calculating and dynamically adjusting is as follows:
  • the third threshold is calculated according to the following formula:
  • B threshold is the third threshold
  • P median is the median or average value of the sizes of all P frames in the video
  • Bmedian is the median or average of the sizes of all B frames in the to-be-detected video. value.
  • B median is the median or average of the sizes of all B frames in the video, so, by formula or The effective threshold can be accurately calculated to accurately detect the scene switching frame in the P frame.
  • the determined video may be removed from the to-be-detected video.
  • the median or average value of the size of other P frames other than the P frame as the new P median and according to the formula or A new B threshold is calculated, and the new B threshpld is used to determine whether the Pmax of the next GOP is a scene switching frame.
  • the third threshold Bthreshold can be updated in real time, the influence of the P frame that has been determined to be the scene switching frame is eliminated in time, and therefore, the accuracy of detecting the scene switching frame in the P frame can be further improved.
  • Embodiment B of Embodiment 1 of the present invention may further include step 201, which is the same as step 101, and is not described again.
  • Embodiment B may also be implemented in an implementation similar to the two implementations of Embodiment A (Embodiment J and Implementation K), as shown in FIGS. 4B and 4C, respectively.
  • Embodiment B of Embodiment 1 of the present invention when detecting a scene switching frame in a video, the size of the largest P frame Pmax among all P frames of the GOP of the video is detected.
  • the relative magnitude relationship between the median or average values of the sizes of all B frames between the scene switching frames closest to Pmax before Pmax and Pmax is determined whether Pmax is a scene switching frame to detect a scene switching frame in the P frame, thereby being effective
  • Pmax is a scene switching frame to detect a scene switching frame in the P frame, thereby being effective
  • the problem of the missed scene switching frame is reduced, and when detecting the scene switching frame in the P frame, not only the size of the I frame and the P frame but also the size of the B frame is considered, thereby further improving
  • Embodiment 2 of the present invention provides a detecting apparatus 200 for detecting a scene switching frame in a video, where the video includes N GOPs, and N is an integer greater than or equal to 2.
  • the detecting device 200 includes a first determining unit 210 and a second determining unit 220.
  • the first embodiment of the second embodiment corresponds to the embodiment A of the embodiment 1, as follows:
  • the first determining unit 210 is configured to determine a maximum P frame Pmax among all P frames of the Kth GOP, where the size of Pmax is Where K is a variable and the values are from M to N, 1 ⁇ M ⁇ N.
  • the second determining unit 220 is configured to determine with The relative size is greater than or equal to the first threshold, and with In the case where the relative size is greater than or equal to the second threshold, it is determined that Pmax is a scene switching frame, wherein The median or average value of the size of multiple I frames between frames where Pmax is closest to Pmax before Pmax, a median or average value of sizes of a plurality of P frames in the Kth GOP; the first threshold is greater than 0 and less than 1, and the second threshold is greater than 1.
  • the first determining unit 210 may be specifically configured to perform the step 102 of the method described in Embodiment A of the foregoing Embodiment 1, where the second determining unit 220 may be specifically configured to execute the foregoing Embodiment 1 Step 103 of the method described in A.
  • the second determining unit 220 may be further configured to detect a scene switching frame from an I frame of the Mth to the Nth GOPs of the Nth GOP, and may be used to perform the implementation of Embodiment 1 above. Step 101 of the method described in mode A.
  • Embodiment 2 corresponds to Embodiment B of Embodiment 1, as follows:
  • the first determining unit 210 is configured to determine a maximum P frame Pmax among all P frames of the Kth GOP, where the size of Pmax is Where K is a variable and the values are from M to N, 1 ⁇ M ⁇ N.
  • the second determining unit 220 is configured to determine with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to the second threshold, and there is no B frame in the Kth GOP, or is determined with The relative size is greater than or equal to the first threshold, with The relative size is greater than or equal to a second threshold, and the Kth GOP has a B frame, and with In the case where the relative size is greater than or equal to the third threshold, it is determined that Pmax is a scene switching frame, wherein The median or mean value of the size of multiple I frames between the scene switching frames closest to Pmax before Pmax and Pmax, Is the median or average value of the sizes of the plurality of P frames in the Kth GOP, a median or average of the sizes of all B frames between the scene switching frames closest to Pmax before Pmax and Pmax; the first threshold is greater than 0 and less than 1, the second threshold is greater than 1, the third The threshold is greater than 1.
  • the first determining unit 210 may be specifically configured to perform the step 202 of the method described in Embodiment B of the foregoing Embodiment 1, where the second determining unit 220 may be specifically configured to execute the foregoing Embodiment 1 Step 203 of the method described in B.
  • the second determining unit 220 may be further configured to detect a scene switching frame from an I frame of the Mth to the Nth GOPs of the Nth GOP, and may be used to perform the implementation of Embodiment 1 above. Step 201 of the method described in mode B.
  • Embodiment 2 of the present invention when detecting a scene switching frame in a video, not only the scene switching frame in the I frame but also the size of the largest P frame Pmax among all the P frames of the GOP of the video is detected. a relative magnitude relationship between the median or average values of the sizes of the plurality of I frames between the scene switching frames closest to Pmax before Pmax and Pmax, and Determining whether Pmax is a scene switching frame with respect to a relative size relationship between a median or an average value of a plurality of P frames in the GOP, so as to implement detection of a scene switching frame in a P frame, thereby effectively reducing a missed scene switching frame. problem.
  • Embodiment B of Embodiment 2 of the present invention when detecting a scene switching frame in a P frame, not only the size of the I frame and the P frame but also the size of the B frame is considered, thereby further improving the detection of the P frame.
  • the accuracy of the scene switching frame when detecting a scene switching frame in a P frame, not only the size of the I frame and the P frame but also the size of the B frame is considered, thereby further improving the detection of the P frame. The accuracy of the scene switching frame.
  • Embodiment 3 of the present invention provides a detecting apparatus 1000.
  • the detecting apparatus 1000 includes a processor 1010 and a memory 1020, wherein the processor 1010 and the memory 1020 are completed by a bus. Communication with each other.
  • the memory 1020 is configured to store computer operation instructions.
  • the memory 1020 may include a high speed RAM memory and may also include a non-volatile memory such as at least one disk memory.
  • the processor 1010 is configured to execute computer operation instructions stored in the memory 1020.
  • the processor 1010 may be a central processing unit (CPU) and is a computer core unit.
  • the processor 1010 executes the computer operation instruction such that the detecting device 1000 performs the method described in Embodiment 1 above.
  • Embodiment 3 of the present invention when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • Embodiment 4 of the present invention provides a detecting apparatus 400.
  • the detecting device 400 includes a media unit 4010 and a detecting device 4020.
  • the media unit 4010 is configured to acquire the video (hereinafter referred to as a video to be tested) and transmit the video to the detecting device 4020.
  • the media unit 4010 may specifically read the to-be-detected video from the video file, or may acquire the to-be-detected video from the media stream sent by the received video server.
  • the video to be inspected may be a complete video, or may be a video segment in a video. If it is a video segment, the media unit 4010 may display the video in which the video segment is located (ie, include the video segment).
  • the video is transmitted to the detecting device 4020 and the video segment in the received video is detected by the detecting device 4020 to detect a scene switching frame in the to-be-detected video.
  • the detecting device 4020 may specifically be the detecting device 200 provided in Embodiment 2 or the detecting device 1000 provided in Embodiment 3, for acquiring the to-be-detected video from the media unit 4010 and executing the detecting device provided in Embodiment 2 200 or the operation performed by the detecting device 1000 provided in Embodiment 3.
  • the detecting device 4020 may further evaluate the quality of the video in which the to-be-detected video or the to-be-detected video is located according to the detected scene switching frame.
  • Embodiment 4 of the present invention when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • Embodiment 5 of the present invention provides a system 2000 for implementing video quality assessment.
  • system 2000 includes a video server 2010, a transmission device 2020, and a video terminal 2030, wherein a video stream transmitted by the video server 2010 is transmitted to the video terminal 2030 via the transmission device 2020.
  • the transmission device 2020 or the video terminal 2030 may specifically include the detection device 200 provided in Embodiment 2 or the detection device 1000 provided in Embodiment 3.
  • the transmission device 2020 and the video terminal 2030 may each include the detection device 200 provided in Embodiment 2 or the detection device 1000 provided in Embodiment 3.
  • the transmission device 2020 or the video terminal 2030 may specifically be the detection device 400 provided in Embodiment 4.
  • the system further includes a detecting device 2040, as shown in FIGS. 9B and 9C, the detecting device 2040 may specifically be the detecting device 200 provided in Embodiment 2 or the detecting device provided in Embodiment 3. 1000, the transmission device 2020 or the video terminal 2030 is connected to the detecting device 2040, and the detecting device 2040 acquires the said device 2020 or the video terminal 2030 connected to the detecting device 2040 Video stream.
  • the transmission device 2020 and the video terminal 2030 may be respectively connected to one detecting device 2040.
  • Embodiment 5 of the present invention when detecting a scene switching frame in a video, a scene switching frame in the P frame can be detected, thereby effectively reducing the problem of the missed scene switching frame.
  • the disclosed systems, devices, and methods may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • there may be another division manner for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the functions may be stored in a computer readable storage medium if implemented in the form of a software functional unit and sold or used as a standalone product.
  • the technical solution of the present invention which is essential or contributes to the prior art, or a part of the technical solution, may be embodied in the form of a software product, which is stored in a storage medium, including
  • the instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

本发明公开了一种检测视频中的场景切换帧的方法、装置和系统。在检测视频中的场景切换帧时，根据所述视频的GOP的所有P帧中最大的P帧Pmax的大小Pｋ max与Pmax和Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值的相对大小关系、以及Pｋ max与该GOP中的多个P帧的大小的中值或平均值的相对大小关系判断Pmax是否为场景切换帧，以检测P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。

Description

一种检测场景切换帧的方法、装置和系统 技术领域
本发明涉及视频技术领域，尤其涉及一种检测场景切换帧的方法、装置及系统。
背景技术
随着通信技术的发展，IPTV(Internet Protocol Television，因特网协议电视)、OTT等视频业务，已经进入大规模商用阶段。为了保证视频业务的质量，必须对视频质量进行评估，以便及时采取相应的措施进行调整，以保证视频业务的正常运行。因此，如何对视频质量进行准确的评估，便成为一个迫切需要解决的重要问题。
一段视频由多个连续的视频帧序列组成，通常包括不止一个场景，如一段视频包括4个场景，场景1和场景3对应的是拍摄的足球场的视频，场景2和场景4对应的是拍摄的观众席的视频。
在对视频质量进行评估时，需要先检测出场景发生切换的位置，即场景切换帧的位置，进而基于场景评估视频质量。如，对视频进行编码带来的视频编码损伤不仅和视频编码类型、帧率、分辨率和码率相关，还和场景的复杂度有关，因此，在评估视频编码损伤时，需要先检测出场景发生切换的位置。所以，在对视频质量进行评估时，场景切换检测必须非常准确。
在对视频中的各视频帧进行编码时会将各视频帧编码成不同类型的帧，如I帧、P帧、B帧。其中，I帧为帧内预测编码帧，在编码时只参考帧内的数据；P帧为前向预测编码帧，即单向差别帧，记录的是本帧跟之前的一个I帧(或P帧)的差别；B帧为双向预测内插编码帧，即双向差别帧，记录的是本帧与前一帧和后一帧的差别。
标准ITU-T P1201.2所提供的IPTV监控方案提供了一种检测场景切换帧的方法，然而，该现有技术只检测I帧中的场景切换帧，但实际上很多场景切换帧为P帧，因此，根据该现有技术检测场景切换帧会导致漏检。
发明内容
本发明实施例提供一种检测场景切换帧的方法和装置，用于解决现有技术中存在的漏检场景切换帧的问题。
第一方面，提供了一种检测场景切换帧的方法，所述视频包括N个图像组GOP，N为大于等于2的整数，所述方法包括：
确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000001
其中K为变量且取值依次从M至N，1≤M≤N；
在确定
Figure PCTCN2017098483-appb-000002
和
Figure PCTCN2017098483-appb-000003
的相对大小大于或等于第一阈值，且
Figure PCTCN2017098483-appb-000004
和
Figure PCTCN2017098483-appb-000005
的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000006
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000007
为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
根据本发明第一方面提供的方法，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
在第一方面的第一种可能的实现方式中，
Figure PCTCN2017098483-appb-000008
和
Figure PCTCN2017098483-appb-000009
的相对大小
Figure PCTCN2017098483-appb-000010
具体可以根据公式
Figure PCTCN2017098483-appb-000011
或者
Figure PCTCN2017098483-appb-000012
计算。
公式
Figure PCTCN2017098483-appb-000013
或
Figure PCTCN2017098483-appb-000014
可以简单有效地反映出
Figure PCTCN2017098483-appb-000015
和
Figure PCTCN2017098483-appb-000016
的相对大小关系。
在一种实现方式中，根据公式
Figure PCTCN2017098483-appb-000017
计算
Figure PCTCN2017098483-appb-000018
相应地，可以根据公式
Figure PCTCN2017098483-appb-000019
计算所述第一阈值。
在另一种实现方式中，根据公式
Figure PCTCN2017098483-appb-000020
计算
Figure PCTCN2017098483-appb-000021
相应地，可以根据公式
Figure PCTCN2017098483-appb-000022
计算所述第一阈值。
其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000023
或
Figure PCTCN2017098483-appb-000024
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第一方面或第一方面的第一种可能的实现方式，在第二种可能的实现方式中，
Figure PCTCN2017098483-appb-000025
和
Figure PCTCN2017098483-appb-000026
的相对大小
Figure PCTCN2017098483-appb-000027
具体可以根据公式
Figure PCTCN2017098483-appb-000028
计算。公式
Figure PCTCN2017098483-appb-000029
可以简单有效地反映出
Figure PCTCN2017098483-appb-000030
和
Figure PCTCN2017098483-appb-000031
的相对大小关系。
具体可以根据如下公式计算
Figure PCTCN2017098483-appb-000032
Figure PCTCN2017098483-appb-000033
其中，P-m，…，P-1表示第K个GOP中Pmax之前的P帧，P1，…，Pn表示第K个GOP中Pmax之后的P帧，F用于求P-m，…，P-1，P1，…，Pn的大小的中值或平均值；
m＝min(num_before_P_frames，max_num)
n＝min(num_after_P_frames，max_num)
其中，num_before_P_frames为第K个GOP中Pmax之前的P帧数，num_after_P_frames为第K个GOP中Pmax之后的P帧数，max_num表示预设的需要考虑的帧数。
通过该公式计算
Figure PCTCN2017098483-appb-000034
考虑了所述第一视频帧所在GOP中离所述第一视频帧最近的一些视频帧，从而进一步提升了检测P帧中的场景切换帧的准确率。
第二方面，提供了一种实现视频质量评估的方法，所述视频包括N个图像组GOP，N为大于等于2的整数，所述方法包括：
确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000035
其中K为变量且取值依次从M至N，1≤M≤N；
在确定
Figure PCTCN2017098483-appb-000036
和
Figure PCTCN2017098483-appb-000037
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000038
和
Figure PCTCN2017098483-appb-000039
的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在确定
Figure PCTCN2017098483-appb-000040
和
Figure PCTCN2017098483-appb-000041
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000042
和
Figure PCTCN2017098483-appb-000043
的相对大小大于或等于第二阈值，所述第K个GOP中有B帧，且
Figure PCTCN2017098483-appb-000044
和
Figure PCTCN2017098483-appb-000045
的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000046
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或均值，
Figure PCTCN2017098483-appb-000047
为所述第K个GOP中的多个P帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000048
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
根据本发明第二方面提供的方法，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题，并且，在检测P帧中的场景切换帧时，不仅考虑了I帧和P帧的大小，还考虑了B帧的大小，从而进一步提升了检测P帧中的场景切换帧的准确率。
在第二方面的第一种可能的实现方式中，
Figure PCTCN2017098483-appb-000049
和
Figure PCTCN2017098483-appb-000050
的相对大小
Figure PCTCN2017098483-appb-000051
具体可以根据公式
Figure PCTCN2017098483-appb-000052
或者
Figure PCTCN2017098483-appb-000053
计算。
公式
Figure PCTCN2017098483-appb-000054
或
Figure PCTCN2017098483-appb-000055
可以简单有效地反映出
Figure PCTCN2017098483-appb-000056
和
Figure PCTCN2017098483-appb-000057
的相对大小关系。
在一种实现方式中，可以根据公式
Figure PCTCN2017098483-appb-000058
计算
Figure PCTCN2017098483-appb-000059
相应地，可以根据公式
Figure PCTCN2017098483-appb-000060
计算所述第一阈值。
在另一种实现方式中，可以根据公式
Figure PCTCN2017098483-appb-000061
计算
Figure PCTCN2017098483-appb-000062
相应地，可以根据公式
Figure PCTCN2017098483-appb-000063
计算所述第一阈值。
其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000064
或
Figure PCTCN2017098483-appb-000065
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第二方面或第二方面的第一种可能的实现方式，在第二种可能的实现方式中，
Figure PCTCN2017098483-appb-000066
和
Figure PCTCN2017098483-appb-000067
的相对大小
Figure PCTCN2017098483-appb-000068
具体可以根据公式
Figure PCTCN2017098483-appb-000069
计算。公式
Figure PCTCN2017098483-appb-000070
可以简单有效地反映出
Figure PCTCN2017098483-appb-000071
和
Figure PCTCN2017098483-appb-000072
的相对大小关系。
具体可以根据如下公式计算
Figure PCTCN2017098483-appb-000073
Figure PCTCN2017098483-appb-000074
其中，P-m，…，P-1表示第K个GOP中Pmax之前的P帧，P1，…，Pn表示第K个GOP中Pmax之后的P帧，F用于求P-m，…，P-1，P1，…，Pn的大小的中值或平均值；
m＝min(num_before_P_frames，max_num)
n＝min(num_after_P_frames，max_num)
其中，num_before_P_frames为第K个GOP中Pmax之前的P帧数，num_after_P_frames 为第K个GOP中Pmax之后的P帧数，max_num表示预设的需要考虑的帧数。
通过该公式计算
Figure PCTCN2017098483-appb-000075
考虑了所述第一视频帧所在GOP中离所述第一视频帧最近的一些视频帧，从而进一步提升了检测P帧中的场景切换帧的准确率。
结合第二方面或第二方面的第一种或第二种可能的实现方式，在第三种可能的实现方式中，
Figure PCTCN2017098483-appb-000076
和
Figure PCTCN2017098483-appb-000077
的相对大小
Figure PCTCN2017098483-appb-000078
具体可以根据公式
Figure PCTCN2017098483-appb-000079
或者
Figure PCTCN2017098483-appb-000080
计算。
公式
Figure PCTCN2017098483-appb-000081
或者
Figure PCTCN2017098483-appb-000082
可以简单有效地反映出
Figure PCTCN2017098483-appb-000083
和
Figure PCTCN2017098483-appb-000084
的相对大小关系。
在一种实现方式中，根据公式
Figure PCTCN2017098483-appb-000085
计算
Figure PCTCN2017098483-appb-000086
相应地，可以根据公式
Figure PCTCN2017098483-appb-000087
计算所述第三阈值。
在另一种实现方式中，根据公式
Figure PCTCN2017098483-appb-000088
计算
Figure PCTCN2017098483-appb-000089
相应地，可以根据公式
Figure PCTCN2017098483-appb-000090
计算所述第三阈值。
其中，Bthreshold为所述第三阈值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值。
因为Pmedian为所述视频中的所有P帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000091
或
Figure PCTCN2017098483-appb-000092
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第二方面的第三种可能的实现方式，在第四种可能的实现方式中，在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P 帧的大小的中值或平均值作为新的Pmedian，并根据公式
Figure PCTCN2017098483-appb-000093
或
Figure PCTCN2017098483-appb-000094
计算新的Btheeshold，所述新的Bthreshold用于判断下一个GOP的Pmax是否为场景切换帧。
根据第二方面的第四种可能的实现方式，第三阈值Bthreshold可以实时更新，及时排除了已经确定为场景切换帧的P帧的影响，因此，可以进一步提升检测P帧中的场景切换帧的准确率。
第三方面，提供了一种用于检测视频中的场景切换帧的检测装置，所述视频包括N个GOP，N为大于等于2的整数，所述检测装置包括：第一确定单元和第二确定单元；
所述第一确定单元，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000095
其中K为变量且取值依次从M至N，1≤M≤N；
所述第二确定单元，用于在确定
Figure PCTCN2017098483-appb-000096
和
Figure PCTCN2017098483-appb-000097
的相对大小大于或等于第一阈值，且
Figure PCTCN2017098483-appb-000098
和
Figure PCTCN2017098483-appb-000099
的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000100
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000101
为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
根据本发明第三方面提供的检测装置，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
在第三方面的第一种可能的实现方式中，所述第二确定单元，具体可以根据公式
Figure PCTCN2017098483-appb-000102
或者
Figure PCTCN2017098483-appb-000103
计算
Figure PCTCN2017098483-appb-000104
和
Figure PCTCN2017098483-appb-000105
的相对大小
Figure PCTCN2017098483-appb-000106
公式
Figure PCTCN2017098483-appb-000107
或
Figure PCTCN2017098483-appb-000108
可以简单有效地反映出
Figure PCTCN2017098483-appb-000109
和
Figure PCTCN2017098483-appb-000110
的相对大小关系。
在一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000111
计算
Figure PCTCN2017098483-appb-000112
相应地，所述第 二确定单元可以根据公式
Figure PCTCN2017098483-appb-000113
计算所述第一阈值。
在另一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000114
计算
Figure PCTCN2017098483-appb-000115
相应地，所述第二确定单元可以根据公式
Figure PCTCN2017098483-appb-000116
计算所述第一阈值。
其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000117
或
Figure PCTCN2017098483-appb-000118
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第三方面或第三方面的第一种可能的实现方式，在第二种可能的实现方式中，所述第二确定单元具体可以根据公式
Figure PCTCN2017098483-appb-000119
计算
Figure PCTCN2017098483-appb-000120
和
Figure PCTCN2017098483-appb-000121
的相对大小
Figure PCTCN2017098483-appb-000122
公式
Figure PCTCN2017098483-appb-000123
可以简单有效地反映出
Figure PCTCN2017098483-appb-000124
和
Figure PCTCN2017098483-appb-000125
的相对大小关系。
所述第二确定单元具体可以根据如下公式计算
Figure PCTCN2017098483-appb-000126
Figure PCTCN2017098483-appb-000127
其中，P_m，…，P-1表示第K个GOP中Pmax之前的P帧，P1，…，Pn表示第K个GOP中Pmax之后的P帧，F用于求P_m，…，P-1，P1，…，Pn的大小的中值或平均值；
m＝min(num_before_P_frames，max_num)
n＝min(num_after_P_frames，max_num)
其中，num_before_P_frames为第K个GOP中Pmax之前的P帧数，num_after_P_frames为第K个GOP中Pmax之后的P帧数，max_num表示预设的需要考虑的帧数。
通过该公式计算
Figure PCTCN2017098483-appb-000128
考虑了所述第一视频帧所在GOP中离所述第一视频帧最近的一些视频帧，从而进一步提升了检测P帧中的场景切换帧的准确率。
第四方面，提供了一种检测视频中的场景切换帧的检测装置，所述视频包括N个图像组GOP，N为大于等于2的整数，所述检测装置包括：第一确定单元和第二确定单元；
所述第一确定单元，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000129
其中K为变量且取值依次从M至N，1≤M≤N；
所述第二确定单元，用于在确定
Figure PCTCN2017098483-appb-000130
和
Figure PCTCN2017098483-appb-000131
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000132
和
Figure PCTCN2017098483-appb-000133
的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在确定
Figure PCTCN2017098483-appb-000134
和
Figure PCTCN2017098483-appb-000135
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000136
和
Figure PCTCN2017098483-appb-000137
的相对大小大于或等于第二阈值，所述第K个GOP中有B帧，且
Figure PCTCN2017098483-appb-000138
和
Figure PCTCN2017098483-appb-000139
的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000140
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或均值，
Figure PCTCN2017098483-appb-000141
为所述第K个GOP中的多个P帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000142
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
根据本发明第四方面提供的检测装置，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题，并且，在检测P帧中的场景切换帧时，不仅考虑了I帧和P帧的大小，还考虑了B帧的大小，从而进一步提升了检测P帧中的场景切换帧的准确率。
在第四方面的第一种可能的实现方式中，所述第二确定单元，具体可以根据公式
Figure PCTCN2017098483-appb-000143
或者
Figure PCTCN2017098483-appb-000144
计算
Figure PCTCN2017098483-appb-000145
和
Figure PCTCN2017098483-appb-000146
的相对大小
Figure PCTCN2017098483-appb-000147
公式
Figure PCTCN2017098483-appb-000148
或
Figure PCTCN2017098483-appb-000149
可以简单有效地反映出
Figure PCTCN2017098483-appb-000150
和
Figure PCTCN2017098483-appb-000151
的相对大小关系。
在一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000152
计算
Figure PCTCN2017098483-appb-000153
相应地，所述第二确定单元可以根据公式
Figure PCTCN2017098483-appb-000154
计算所述第一阈值。
在另一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000155
计算
Figure PCTCN2017098483-appb-000156
相应地，所述第二确定单元可以根据公式
Figure PCTCN2017098483-appb-000157
计算所述第一阈值。
其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000158
或
Figure PCTCN2017098483-appb-000159
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第四方面或第四方面的第一种可能的实现方式，在第二种可能的实现方式中，所述第二确定单元具体可以根据公式
Figure PCTCN2017098483-appb-000160
计算
Figure PCTCN2017098483-appb-000161
和
Figure PCTCN2017098483-appb-000162
的相对大小
Figure PCTCN2017098483-appb-000163
公式
Figure PCTCN2017098483-appb-000164
可以简单有效地反映出
Figure PCTCN2017098483-appb-000165
和
Figure PCTCN2017098483-appb-000166
的相对大小关系。
所述第二确定单元具体可以根据如下公式计算
Figure PCTCN2017098483-appb-000167
Figure PCTCN2017098483-appb-000168
其中，P_m，…，P-1表示第K个GOP中Pmax之前的P帧，P1，…，Pn表示第K个GOP中Pmax之后的P帧，F用于求P_m，…，P-1，P1，…，Pn的大小的中值或平均值；
m＝min(num_before_P_frames，max_num)
n＝min(num_after_P_frames，max_num)
其中，num_before_P_frames为第K个GOP中Pmax之前的P帧数，num_after_P_frames为第K个GOP中Pmax之后的P帧数，max_num表示预设的需要考虑的帧数。
通过该公式计算
Figure PCTCN2017098483-appb-000169
考虑了所述第一视频帧所在GOP中离所述第一视频帧最近的一些视频帧，从而进一步提升了检测P帧中的场景切换帧的准确率。
结合第四方面或第四方面的第一种或第二种可能的实现方式，在第三种可能的实现方式中，所述第二确定单元具体可以根据公式
Figure PCTCN2017098483-appb-000170
或者
Figure PCTCN2017098483-appb-000171
计算
Figure PCTCN2017098483-appb-000172
和
Figure PCTCN2017098483-appb-000173
的相对大小
Figure PCTCN2017098483-appb-000174
公式
Figure PCTCN2017098483-appb-000175
或者
Figure PCTCN2017098483-appb-000176
可以简单有效地反映出
Figure PCTCN2017098483-appb-000177
和
Figure PCTCN2017098483-appb-000178
的相对大小 关系。
在一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000179
计算
Figure PCTCN2017098483-appb-000180
相应地，所述第二确定单元可以根据公式
Figure PCTCN2017098483-appb-000181
计算所述第一阈值。
在另一种实现方式中，所述第二确定单元根据公式
Figure PCTCN2017098483-appb-000182
计算
Figure PCTCN2017098483-appb-000183
相应地，所述第二确定单元可以根据公式
Figure PCTCN2017098483-appb-000184
计算所述第一阈值。
其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000185
或
Figure PCTCN2017098483-appb-000186
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
结合第四方面的第三种可能的实现方式，在第四种可能的实现方式中，在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据公式
Figure PCTCN2017098483-appb-000187
或
Figure PCTCN2017098483-appb-000188
计算新的Bthreshold，所述新的Bthreshold用于判断下一个GOP的Pmax是否为场景切换帧。
根据第四方面的第四种可能的实现方式，第三阈值Bthreshold可以实时更新，及时排除了已经确定为场景切换帧的P帧的影响，因此，可以进一步提升检测P帧中的场景切换帧的准确率。
第五方面，提供了一种检测视频中的场景切换帧的检测装置，所述检测装置包括处理器和存储器；
所述存储器，用于存放计算机操作指令；
所述处理器，用于执行所述存储器中存放的计算机操作指令使得所述检测装置执行本发明第一方面或第一方面的任一可能的实现方式、第二方面或第二方面的任一可能的实现方式所提供的方法。
根据本发明的第五方面提供的检测装置，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
第六方面，提供了一种检测设备，所述检测设备包括媒体单元和检测装置；
所述媒体单元，用于获取所述视频，并将所述视频传输给所述检测装置；
所述检测装置，用于从所述媒体单元获取所述视频，并执行本发明第三方面或第三方面的任一可能的实现方式、第四方面或第四方面的任一可能的实现方式、第五方面或第五方面的任一可能的实现方式所提供的检测装置所执行的操作。
根据本发明的第六方面提供的检测设备，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
第七方面，提供了一种实现视频质量评估的系统，所述系统包括视频服务器、传输设备和视频终端，其中，所述视频服务器发送的视频流经由所述传输设备传输到所述视频终端；
所述传输设备或所述视频终端具体可以包括本发明第三方面或第三方面的任一可能的实现方式、第四方面或第四方面的任一可能的实现方式、第五方面或第五方面的任一可能的实现方式所提供的检测装置；或者，
所述系统还包括第一检测装置，所述第一检测装置具体可以为本发明第三方面或第三方面的任一可能的实现方式、第四方面或第四方面的任一可能的实现方式、第五方面或第五方面的任一可能的实现方式所提供的检测装置，所述传输设备2020或所述视频终端2030与所述第一检测装置相连，并且所述第一检测装置通过与第一检测装置连接的所述传输设备或所述视频终端获取所述视频流。
根据本发明的第七方面提供的系统，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
附图说明
为了更清楚地说明本发明实施例的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1A和1B是本发明实施例1提供的视频系统100的组网结构示意图；
图2A、2B和2C是本发明实施例1提供的GOP示意图；
图3A、3B和3C是本发明实施例1的实施方式A提供的方法流程示意图；
图4A、4B和4C是本发明实施例1的实施方式B提供的方法流程示意图；
图5A、5B和5C是本发明实施例1提供的GOP示例的示意图；
图6是本发明实施例2提供的检测装置200的结构示意图；
图7是本发明实施例3提供的检测装置1000的结构示意图；
图8是本发明实施例4提供的检测设备400的结构示意图；
图9A-9C是本发明实施例5提供的系统2000的结构示意图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
图1A为本发明实施例提供的一种视频系统100的组网结构示意图，视频系统100包括视频服务器110、一个或多个传输设备120和视频终端130，其中，视频服务器110发送的视频流经由所述传输设备120传输到视频终端130。
视频系统100具体可以是如图1B所示的IPTV系统。在该IPTV系统中，视频服务器110具体为视频头端(video headEnd，video HE)，传输设备120具体包括核心路由器(Core Router，CR)、宽带网络网关(Broadband Network Gateway，BNG)、光线路终端(Optical Line Terminal，OLT)等网络设备，视频终端130具体为机顶盒(Set Top Box，STB)。
在如图1A、1B所示的视频系统中，当视频流从视频服务器传送到视频终端时，由于所经过的网络状态的变化，会导致视频流出现丢包、时延、抖动、乱序等异常现象。这些异常现象会造成视频终端屏幕上所播放的视频画面出现花屏、卡顿等缺陷，导致用户的视频观看体验下降。因此，需要通过视频质量评估对用户的视频体验进行监控。
在对视频质量进行评估时，往往需要先检测出场景发生切换的位置，即场景切换帧的位置，进而基于场景评估视频质量。
如，对视频进行编码带来的视频编码损伤不仅和视频编码类型、帧率、分辨率和码率相关，还和场景的复杂度有关，因此，在评估视频编码损伤时，需要先检测出场景发生切换的位置。
又如，当视频传输过程中发生丢包时，视频终端的解码器通常会将受损帧的前一帧中对应区域的视频内容作为受损帧中受损区域的内容来对受损帧进行错误补偿，因此受损帧与前一帧的内容差距越小，补偿效果越好，而当受损帧为场景切换帧时，由于场景切换帧与其前一帧的内容几乎完全不同，此时的补偿效果是最差的，因此，在对丢包所导致的视频质量造成的影响时，需要考虑受损帧是否为场景切换帧。
再如，编码后的视频帧序列由多个图像组(Group of Picture，GOP)组成，如图2A所示，每个GOP由I帧开始，随后是一些P帧和B帧，到下一个I帧的前一帧结束。I帧为帧内编码帧，P帧为前向参考帧，B帧为双向参考帧。当一个GOP中的某一帧发生丢包时时，丢包导致的解码错误会在后面的视频帧中持续扩散，通常认为扩散直至该GOP的最后一帧。如图2B所示，GOP1中第四帧受损时，通常认为错误会一直扩散至该GOP的最后一帧。但是当GOP中有场景切换帧时，由于场景切换帧与其前面的帧的内容几乎完全不同，编码时多采用帧内预测编码(该场景切换帧中的宏块大多采用帧内预测编码)，因此，如果该GOP内的且位于该场景切换帧之前的某一帧受损，错误扩散会在场景切换帧截止，如图2C所示，GOP1中第六帧为场景切换帧，当第四帧受损时，错误扩散至第六帧就截止了。所以，在对丢包所导致的视频质量造成的影响时，需要检测场景切换帧。
需要说明的是，在本发明实施例中所说的各个帧之间的前后顺序，指的是所指代的帧 在视频中的时间上的前后，如，一个时长为T(如10秒)的视频包括一个时刻t1处的视频帧1和一个时刻t2处的视频帧2，如果t1小于t2，如t1为1秒30毫秒，t2为5秒40毫秒，则认为视频帧1在视频帧2的前面。
在具体实现时，可以在视频系统中部署用于实现场景切换帧检测的检测装置。该检测装置可以部署在视频流所经过的任一设备(如传输设备120、视频终端130)上，也可以旁挂到视频流所经过的任一设备上并通过镜像的方式获取视频流。
图3A和3B是本发明实施例1提供的方法流程示意图，本发明实施例1的方法可以应用于图1A、1B所示的视频系统100，由检测装置执行。
在本发明实施例1中，对视频(后续称为待检视频)中的场景切换帧进行检测。所述待检视频可以是从视频文件中读取的，也可以是从获取的视频流中获取的。所述待检视频具体可以是一个完整的视频，也可以是一个视频中的一个视频段。对于较长的视频，通常会设置测量时间窗口，对该测量时间窗口中的视频段进行检测，如，在对一个视频进行检测时，将测量时间窗口的长度设置为10秒，则先对该视频的0-10秒的视频段作为待检视频进行检测，再对10-20秒的视频段作为待检视频进行检测，依次类推。
在检测之前，检测模块可以先确定所述待检视频中的各视频帧的类型(如I帧、P帧、B帧)和大小。
如，实时获取视频流，从对应测量时间窗口(如10-20秒)的视频流的分组报文中提取视频帧的信息，计算视频帧的大小(单位为字节)。具体计算视频帧的大小的过程是循环该测量时间窗口内所有的分组报文，先从分组报文的头中找到当前视频帧的开始标识，然后将包括该开始标识的报文的负载长度以及后续的各分组报文的负载长度累加起来，直至发现下一个视频帧的开始标识，累加和便是当前视频帧的大小。计算视频帧的大小的具体实现方式可参见标准ITU-T P1201.2。
然后确定测量时间窗口内的所有视频帧的类型。具体可以根据分组报文的报文头中的字段random_access_indicator确定视频帧的类型。针对I帧，无论加密与否，都可以通过random_access_indicator确定帧的类型。针对非I帧，当视频没有加密时，可直接从视频帧的帧头中获取帧类型，当视频加密时，或视频帧的帧头丢失时，可先根据帧大小或帧的显示时间戳(Present Time Stamp，PTS)估计GOP模式，GOP模式一般是P B B P B B或P B B B P B B B。当前PTS值和上一个PTS值之间的差值能够描述GOP的模式。一旦GOP模式确定了，所有丢失或被加密的视频帧的模式即可被确定。确定视频帧的类型的具体实现方式可参见标准ITU-T P1201.2。
通过上述两步，便可将待检视频分成若干个GOP。一个待检视频通常包括多个GOP，假定为N个，N为大于等于2的整数，如图5A所示的示例。在如图5A所示的示例中，纯色阴影填充的视频帧为I帧，斜线阴影填充的视频帧为P帧，无填充色的视频帧为B帧。
一个视频帧为一幅图像，在对视频中的各视频帧进行编码时会将各视频帧编码成不同类型的帧，如I帧、P帧、B帧。其中，I帧为帧内预测编码帧，在编码时只参考帧内的数据，因此包含完整的图像数据；P帧为前向预测编码帧，即单向差别帧，记录的是本帧跟之前的一个I帧(或P帧)的差别；B帧为双向预测内插编码帧，即双向差别帧，记录的 是本帧与前一帧和后一帧的差别。
I帧普遍大于P帧，P帧普遍大于B帧。通常I帧大小是P帧大小的2-5倍，P帧大小是B帧大小的2-5倍。
由于场景切换帧与其前一帧的内容差异较大，因此场景切换帧即便被编码为P帧，该场景切换帧中的宏块也大多采用帧内预测编码方式参考本帧内的其它宏块进行编码，所以，会比较大。如果P帧的大小超过I帧的一半，那么该P帧很可能就是场景切换帧，因此，在检测P帧类型的场景切换帧时，P帧与I帧的相对大小关系可以参考。
然而，当视频中的画面内容变化比较快时，如拍摄的激烈的足球赛事等运动比较剧烈的场面，由于前后视频帧之间的相关性小，在对非场景切换帧进行编码时，即便参考前一视频帧将其编码为P帧，压缩率也会比较低，此时P帧的大小也会比较大，甚至会超过I帧大小的一半。在这种情况下，不是场景切换帧的相邻P帧之间的大小差距并不大，而是场景切换帧的P帧和其相邻的不是场景切换帧的P帧的大小差距则比较大，因此，在检测P帧类型的场景切换帧时，相邻P帧之间大小的比值也可以参考。
基于上述分析，下面结合图3A对本发明实施例1的实施方式A详加阐述，以检测待检视频中的场景切换帧，所述视频包括N个GOP，N为大于等于2的整数。
所述待检视频，可以是视频文件中的一段视频，也可以是视频流中的一段视频，如视频服务器发送给视频终端的视频流中的一段视频，相应地，所述检测装置，可以内置于所述视频流所经过的任一设备(如传输设备120、视频终端130)上，也可以旁挂到所述视频流所经过的任一设备上并通过镜像的方式获取所述视频流。
如图3A所示，本发明实施例1的实施方式A提供的方法包括：
所述检测装置从所述N个GOP中的第M个GOP开始依次对每个GOP执行如下操作以判断每个GOP中的所有P帧中最大的P帧Pmax是否为场景切换帧，其中M大于等于1且小于等于N。
具体可以从第一个GOP开始检测，即M等于1。也可以从第一个GOP后面的某个GOP开始检测，例如，待检视频是一个视频流中开始的一段视频，由于前两个GOP中的帧大小一般不太具有可参考性，通常从第三个GOP开始检测，则M等于3。
步骤102：确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000189
其中K为变量且取值依次从M至N，1≤M≤N。
步骤103：在确定
Figure PCTCN2017098483-appb-000190
和
Figure PCTCN2017098483-appb-000191
的相对大小大于或等于第一阈值，且
Figure PCTCN2017098483-appb-000192
和
Figure PCTCN2017098483-appb-000193
的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000194
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000195
为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
另外，在确定
Figure PCTCN2017098483-appb-000196
和
Figure PCTCN2017098483-appb-000197
的相对大小小于第一阈值，或
Figure PCTCN2017098483-appb-000198
和
Figure PCTCN2017098483-appb-000199
的相对大小小于第 二阈值的情况下，确定Pmax不为场景切换帧。
如图5B所示，第K个GOP中的第二个P帧为Pmax，第K个GOP的I帧不是场景切换帧，第K-1个GOP中没有场景切换帧，第K-2个GOP的第三个P帧(后续称为P’max)为场景切换帧，则Pmax之前离Pmax最近的场景切换帧为P’max。
在具体实现时，在步骤102之前还可以将第一个GOP的I帧确定为场景切换帧，则当第K个GOP的Pmax和第一个GOP的I帧之间没有场景切换帧时，Pmax之前离Pmax最近的场景切换帧即为第一个GOP的I帧。如果所述待检视频是一个视频中的视频片段且待检视频之前有已经检测过场景切换帧的视频片段(后续称为前一视频片段)，则Pmax之前离Pmax最近的场景切换帧可能位于所述前一视频片段中。
在计算
Figure PCTCN2017098483-appb-000200
时，可以利用Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有或部分I帧的大小的中值或平均值。如图5B所示，第K个GOP中的第二个P帧为Pmax，Pmax之前离Pmax最近的场景切换帧为P’max，Pmax和P’max之间包含两个I帧Ik和Ik-1，
Figure PCTCN2017098483-appb-000201
指的就是Ik和Ik-1的大小的平均值。
Figure PCTCN2017098483-appb-000202
可以为所述第K个GOP中的所有或部分P帧的大小的中值或平均值。一种优选的实施方式如下：
根据如下公式计算
Figure PCTCN2017098483-appb-000203
Figure PCTCN2017098483-appb-000204
其中，P-m，…，P-1表示第K个GOP中Pmax之前的P帧，P1，…，Pn表示第K个GOP中Pmax之后的P帧，F用于求P-m，…，P-1，P1，…，Pn的大小的中值或平均值；
m＝min(num_before_P_frames，max_num)
n＝min(num_after_P_frames，max_num)
其中，num_before_P_frames为第K个GOP中Pmax之前的P帧数，num_after_P_frames为第K个GOP中Pmax之后的P帧数，max_num表示预设的需要考虑的帧数。如图5C所示，num_before_P_frames等于7，num_after_P_frames等于4，max_num设置为6，此时m等于6，n等于4。
通过该公式计算
Figure PCTCN2017098483-appb-000205
考虑了所述第一视频帧所在GOP中离所述第一视频帧最近的一 些视频帧，从而进一步提升了检测P帧中的场景切换帧的准确率。
在具体实现时，可以根据如下公式计算
Figure PCTCN2017098483-appb-000206
和
Figure PCTCN2017098483-appb-000207
的相对大小
Figure PCTCN2017098483-appb-000208
Figure PCTCN2017098483-appb-000209
或者，
Figure PCTCN2017098483-appb-000210
公式
Figure PCTCN2017098483-appb-000211
可以简单有效地反映出
Figure PCTCN2017098483-appb-000212
和
Figure PCTCN2017098483-appb-000213
的相对大小关系。
所述第一阈值可以预先设置，同一待检视频中不同GOP可以采用相同的第一阈值。如当根据公式
Figure PCTCN2017098483-appb-000214
计算
Figure PCTCN2017098483-appb-000215
时，将所述第一阈值设置为0.53，如当根据公式
Figure PCTCN2017098483-appb-000216
计算
Figure PCTCN2017098483-appb-000217
时，将所述第一阈值设置为0.47。
所述第一阈值也可以是计算出来并可以进行动态调整的，从而进一步提高检测的准确性。计算并动态调整的过程如下：
在对该待检视频的第M个GOP进行检测之前先根据如下公式计算第一阈值：
Figure PCTCN2017098483-appb-000218
或者，
Figure PCTCN2017098483-appb-000219
其中，Imedian为所述待检视频中的所有I帧的大小的中值或平均值，Pmedian为所述待检视频中的所有P帧的大小的中值或平均值。以中值为例，如，所述待检视频中包括9个P帧，大小分别为3 5 3 6 4 7 3 5 4，按照大小顺序排列为3 3 3 4 4 5 5 6 7，则Pmedian为4；所述待检视频中包括3个P帧，大小分别为15 12 18，按照大小顺序排列为12 15 18，则Imedian为15。
因为Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000220
或
Figure PCTCN2017098483-appb-000221
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
然后，在每次确定出一个新的类型为P帧的场景切换帧后，如确定第K个GOP的Pmax为场景切换帧后，可以将所述待检视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并公式
Figure PCTCN2017098483-appb-000222
或
Figure PCTCN2017098483-appb-000223
计算新的Ithreshold，并利用所述新的Ithreshold判断后续GOP(如所述第K个GOP的下一个GOP)的Pmax是否为场景切换帧。
由于第一阈值Ithreshold可以实时更新，及时排除了已经确定为场景切换帧的P帧的影响，因此，可以进一步提升检测P帧中的场景切换帧的准确率。
具体实现方式可以是，若所述第一阈值根据公式
Figure PCTCN2017098483-appb-000224
获得，所述
Figure PCTCN2017098483-appb-000225
和
Figure PCTCN2017098483-appb-000226
的相对大小
Figure PCTCN2017098483-appb-000227
为：
Figure PCTCN2017098483-appb-000228
若所述第一阈值根据公式
Figure PCTCN2017098483-appb-000229
获得，所述
Figure PCTCN2017098483-appb-000230
和
Figure PCTCN2017098483-appb-000231
的相对大小
Figure PCTCN2017098483-appb-000232
为：
Figure PCTCN2017098483-appb-000233
在具体实现时，可以根据如下公式计算
Figure PCTCN2017098483-appb-000234
和
Figure PCTCN2017098483-appb-000235
的相对大小
Figure PCTCN2017098483-appb-000236
Figure PCTCN2017098483-appb-000237
公式
Figure PCTCN2017098483-appb-000238
可以简单有效地反映出
Figure PCTCN2017098483-appb-000239
和
Figure PCTCN2017098483-appb-000240
的相对大小关系。
所述第二阈值通常可以预先设置，同一待检视频中不同GOP可以采用相同的第二阈值，如设置为1.51。
本发明实施例1的实施方式A还可以包括步骤101。
步骤101：从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧。
在步骤101中，在判断第K个GOP的I帧是否为场景切换帧时，具体可以根据所述第K个GOP的I帧的大小与第K-1个GOP的I帧的大小的比值、第K-1个GOP内的所有P帧的大小的平均值与所述第K个GOP内的所有P帧的大小的平均帧的比值、以及第K-1个GOP内的所有B帧的大小的平均值与所述第K个GOP内的所有B帧的大小的平均帧的比值判断所述第K个GOP的I帧是否为场景切换帧。具体实现方式如下：
1、计算第K个GOP的I帧的大小与第K-1个GOP的I帧的大小的比值rI；
2、计算第K-1个GOP内的所有P帧的大小的平均值与第K个GOP内的所有P帧的大小的平均帧的比值rP；
3、计算第K-1个GOP的所有B帧的大小的平均值与第K个GOP内的所有B帧的大小的平均帧的比值rB；
4、如果比值rI大于第一阈值或小于第二阈值，则进一步判断下列条件(1)和条件(2)，否则，确定第K个GOP的I帧不是场景切换帧。
条件(1)rP小于阈值第三阈值或rP大于第四阈值
条件(2)rB小于阈值第五阈值或rB大于第六阈值
如果条件(1)和条件(2)均满足，确定第K个GOP的I帧为场景切换帧，否则，确定第K个GOP的I帧不是场景切换帧。
上述实现方式的具体细节可参见标准ITU-T P1201.2。
在具体实现时，可以直接将第一个GOP的I帧确定为场景切换帧，并在K不等于1时利用上述方法判断第K个GOP的I帧是否为场景切换帧。
在具体实现时，如图3B所示，可以采用实现方式J实现实施方式A：先执行步骤101再执行步骤102-103，即，先检测出第M个GOP至第N个GOP的I帧中的场景切换帧再检测第M个GOP至第N个GOP的P帧中的场景切换帧。如，先将GOP1的I帧确定为场景切换帧，再判断GOPM(如GOP1)至GOPN中的I帧是否为场景切换帧，然后依次判断GOPM(如GOP1)至GOPN中的Pmax是否为场景切换帧。
在具体实现时，如图3C所示，也可以采用实现方式K实现实施方式A：将步骤101贯穿到步骤102-103中，按照视频帧的顺序依次检测场景切换帧，即，从第M个(如第一个)GOP开始根据GOP的顺序逐个GOP检测当前GOP中的场景切换帧，并在检测当前GOP中的场景切换帧时先判断当前GOP中的I帧是否为场景切换帧再判断当前GOP中的Pmax是否为场景切换帧。如，先将GOP1的I帧确定为场景切换帧，然后依次判断GOP1的Pmax、GOP2的I帧、GOP2的Pmax、GOP3的I帧、GOP3的Pmax、……、GOPN的I帧、GOPN的Pmax是否为场景切换帧。简而言之，即是在K小于N时，在判断第K个GOP的Pmax是否为场景切换帧之后判断第K+1个GOP的I帧是否为场景切换帧。
如果采用实现方式K来实现实施方式A，由于场景在短时间内连续发生切换的几率较低，所以，在判断某个I帧是否为场景切换帧前，还可以先计算该I帧与前一个场景切换帧之间的距离(后续称为第一距离)，如果所述第一距离小于或等于距离阈值，则确定该I帧不是场景切换帧，否则，可以进一步根据标准ITU-T P1201.2提供的方法判断该I帧是否为场景切换帧。具体实现方式为：当K小于N时，在判断第K个GOP的Pmax是否为场景切换帧之后，在确定所述第K+1个GOP的I帧(后续称为当前I帧)和所述当前I帧之前离所述当前I帧最近的场景切换帧之间的距离小于或等于距离阈值时，确定所述当前I帧不是场景切换帧，否则，可以进一步根据标准ITU-T P1201.2提供的方法判断所述当前I帧是否为场景切换帧。
本发明所说的两个视频帧之间的距离，指的是两个视频帧之间相差多少个视频帧，第X个视频帧和第Y个视频帧之间的距离为Y-N，两个相邻视频帧之间的距离为1。
所述距离阈值可以预先设置，同一待检视频中不同GOP可以采用相同的距离阈值。
所述距离阈值也可以是确定出来并进行动态调整的，从而进一步提高检测的准确性。确定并动态调整距离阈值的过程如下：
在从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧之前，先确定初始距离阈值，确定的初始距离阈值可以为如下三个长度中的其中一个：
(1)所述N个GOP中长度最长的GOP的长度；
(2)所述N个GOP的长度的平均值；
(3)长度L，其中，所述N个GOP中长度为L的GOP的数量最多。
本发明所说的GOP的长度指的是GOP所包括的视频帧的数量。
如，所述待检视频包括8个GOP，长度依次为10 6 8 7 8 7 9 8，根据方式(1)确定的初始距离阈值为10；根据方式(2)确定的初始距离阈值为8；根据方式(3)确定的初始距离阈值为8，因为长度为8的GOP的数量最多。
当采用固定GOP长度编码时，每个GOP的长度是相同的，则通过上述三种方式算出的初始距离阈值是一样的。
然后，在确定出一个新的场景切换帧时，如果所述新的场景切换帧与所述新的场景切换帧之前离所述新的场景切换帧最近的场景切换帧的距离(后续称为第二距离)小于所述距离阈值，将所述距离阈值更新为所述第二距离。
根据本发明的实施例1的实施方式A，在检测视频中的场景切换帧时，根据所述视频的GOP的所有P帧中最大的P帧Pmax的大小
Figure PCTCN2017098483-appb-000241
与Pmax和Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值的相对大小关系、以及
Figure PCTCN2017098483-appb-000242
与该GOP中的多个P帧的大小的中值或平均值的相对大小关系判断Pmax是否为场景切换帧，以检测P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
另外，在视频编码时，P帧普遍大于B帧，通常P帧大小是B帧大小的2-5倍。由于场景切换帧与其前一帧的内容差异较大，因此场景切换帧即便被编码为P帧，该场景切换帧中的宏块也大多采用帧内预测编码，所以，会比较大。如果P帧大小小于B帧的2倍，那么该P帧有可能是场景切换帧，因此，在检测P帧类型的场景切换帧时，P帧与B帧的相对大小关系可以参考。
基于上述分析，下面结合图4A对本发明实施例1的实施方式B详加阐述，以检测待检视频中的场景切换帧，所述视频包括N个GOP，N为大于等于2的整数。
如图4A所示，本发明实施例1的实施方式B提供的方法包括：
步骤202：同步骤102，不再赘述。
步骤203：在确定
Figure PCTCN2017098483-appb-000243
和
Figure PCTCN2017098483-appb-000244
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000245
和
Figure PCTCN2017098483-appb-000246
的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在确定
Figure PCTCN2017098483-appb-000247
和
Figure PCTCN2017098483-appb-000248
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000249
和
Figure PCTCN2017098483-appb-000250
的相对大小大于或等于第二阈值，所述第K个 GOP中有B帧，且
Figure PCTCN2017098483-appb-000251
和
Figure PCTCN2017098483-appb-000252
的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000253
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的均值，
Figure PCTCN2017098483-appb-000254
为所述第K个GOP中的多个P帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000255
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
另外，还可以在确定
Figure PCTCN2017098483-appb-000256
和
Figure PCTCN2017098483-appb-000257
的相对大小小于第一阈值，或
Figure PCTCN2017098483-appb-000258
和
Figure PCTCN2017098483-appb-000259
的相对大小小于第二阈值、或
Figure PCTCN2017098483-appb-000260
和
Figure PCTCN2017098483-appb-000261
的相对大小小于第三阈值的情况下，确定Pmax不为场景切换帧。
其中，确定
Figure PCTCN2017098483-appb-000262
和
Figure PCTCN2017098483-appb-000263
的相对大小大于或等于第一阈值的方法、
Figure PCTCN2017098483-appb-000264
和
Figure PCTCN2017098483-appb-000265
的相对大小大于或等于第二阈值的方法，同步骤102，不再赘述。
在具体实现时，可以根据如下公式计算
Figure PCTCN2017098483-appb-000266
和
Figure PCTCN2017098483-appb-000267
的相对大小
Figure PCTCN2017098483-appb-000268
Figure PCTCN2017098483-appb-000269
或者，
Figure PCTCN2017098483-appb-000270
公式
Figure PCTCN2017098483-appb-000271
或
Figure PCTCN2017098483-appb-000272
可以简单有效地反映出
Figure PCTCN2017098483-appb-000273
和
Figure PCTCN2017098483-appb-000274
的相对大小关系。
所述第三阈值可以预先设置，同一待检视频中不同GOP可以采用相同的第三阈值。如当根据公式
Figure PCTCN2017098483-appb-000275
计算
Figure PCTCN2017098483-appb-000276
时，将所述第一阈值设置为2.87，如当根据公式
Figure PCTCN2017098483-appb-000277
计算
Figure PCTCN2017098483-appb-000278
时，将所述第一阈值设置为1.87。
所述第三阈值也可以是计算出来并进行动态调整的，从而进一步提高检测的准确性。计算并动态调整的过程如下：
在对该待检视频的第M个GOP进行检测之前先根据如下公式计算第三阈值：
Figure PCTCN2017098483-appb-000279
或者，
Figure PCTCN2017098483-appb-000280
其中，Bthreshold为所述第三阈值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，Bmedian为所述待检视频中的所有B帧的大小的中值或平均值。
因为Pmedian为所述视频中的所有P帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值，所以，通过公式
Figure PCTCN2017098483-appb-000281
或
Figure PCTCN2017098483-appb-000282
可以准确地计算出有效阈值，从而精确地检测出P帧中的场景切换帧。
然后，在每次确定出一个新的类型为P帧的场景切换帧后，如确定第K个GOP的Pmax为场景切换帧后，可以将所述待检视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据据公式
Figure PCTCN2017098483-appb-000283
或
Figure PCTCN2017098483-appb-000284
计算新的Bthreshold，所述新的Bthreshpld用于判断下一个GOP的Pmax是否为场景切换帧。
由于第三阈值Bthreshold可以实时更新，及时排除了已经确定为场景切换帧的P帧的影响，因此，可以进一步提升检测P帧中的场景切换帧的准确率。
本发明实施例1的实施方式B还可以包括步骤201，同步骤101，不再赘述。
在具体实现时，也可以采用与实施方式A的两种实现方式(实现方式J和实现方式K)类似的实现方式来实现实施方式B，分别如图4B和4C所示。
根据本发明的实施例1的实施方式B，在检测视频中的场景切换帧时，根据所述视频的GOP的所有P帧中最大的P帧Pmax的大小
Figure PCTCN2017098483-appb-000285
与Pmax和Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值的相对大小关系、
Figure PCTCN2017098483-appb-000286
与该GOP中的多个P 帧的大小的中值或平均值的相对大小关系、以及
Figure PCTCN2017098483-appb-000287
与Pmax和Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值的相对大小关系判断Pmax是否为场景切换帧，以检测P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题，并且，在检测P帧中的场景切换帧时，不仅考虑了I帧和P帧的大小，还考虑了B帧的大小，从而进一步提升了检测P帧中的场景切换帧的准确率。
根据本发明实施例1，本发明实施例2提供了一种用于检测视频中的场景切换帧的检测装置200，所述视频包括N个GOP，N为大于等于2的整数。如图6所示，检测装置200包括第一确定单元210和第二确定单元220。
实施例2的第一种实施方式对应实施例1的实施方式A，如下：
所述第一确定单元210，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000288
其中K为变量且取值依次从M至N，1≤M≤N。
所述第二确定单元220，用于在确定
Figure PCTCN2017098483-appb-000289
和
Figure PCTCN2017098483-appb-000290
的相对大小大于或等于第一阈值，且
Figure PCTCN2017098483-appb-000291
和
Figure PCTCN2017098483-appb-000292
的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000293
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000294
为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
具体地，所述第一确定单元210具体可以用于执行上述实施例1的实施方式A所述的方法的步骤102，所述第二确定单元220具体可以用于执行上述实施例1的实施方式A所述的方法的步骤103。
进一步地，所述第二确定单元220还可以用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧，具体可以用于执行上述实施例1的实施方式A所述的方法的步骤101。
实施例2的第二种实施方式对应实施例1的实施方式B，如下：
所述第一确定单元210，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
Figure PCTCN2017098483-appb-000295
其中K为变量且取值依次从M至N，1≤M≤N。
所述第二确定单元220，用于在确定
Figure PCTCN2017098483-appb-000296
和
Figure PCTCN2017098483-appb-000297
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000298
和
Figure PCTCN2017098483-appb-000299
的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在 确定
Figure PCTCN2017098483-appb-000300
和
Figure PCTCN2017098483-appb-000301
的相对大小大于或等于第一阈值，
Figure PCTCN2017098483-appb-000302
和
Figure PCTCN2017098483-appb-000303
的相对大小大于或等于第二阈值，所述第K个GOP中有B帧，且
Figure PCTCN2017098483-appb-000304
和
Figure PCTCN2017098483-appb-000305
的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
Figure PCTCN2017098483-appb-000306
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或均值，
Figure PCTCN2017098483-appb-000307
为所述第K个GOP中的多个P帧的大小的中值或平均值，
Figure PCTCN2017098483-appb-000308
为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
具体地，所述第一确定单元210具体可以用于执行上述实施例1的实施方式B所述的方法的步骤202，所述第二确定单元220具体可以用于执行上述实施例1的实施方式B所述的方法的步骤203。
进一步地，所述第二确定单元220还可以用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧，具体可以用于执行上述实施例1的实施方式B所述的方法的步骤201。
根据本发明的实施例2，在检测视频中的场景切换帧时，不仅检测I帧中的场景切换帧，还根据所述视频的GOP的所有P帧中最大的P帧Pmax的大小
Figure PCTCN2017098483-appb-000309
与Pmax和Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值的相对大小关系、以及
Figure PCTCN2017098483-appb-000310
与该GOP中的多个P帧的大小的中值或平均值的相对大小关系判断Pmax是否为场景切换帧，以实现检测P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。根据本发明的实施例2的实施方式B，在检测P帧中的场景切换帧时，不仅考虑了I帧和P帧的大小，还考虑了B帧的大小，从而进一步提升了检测P帧中的场景切换帧的准确率。
根据本发明实施例1，本发明实施例3提供了一种检测装置1000，如图7所示，检测装置1000包括处理器1010和存储器1020，其中，处理器1010和存储器1020之间通过总线完成相互间的通信。
存储器1020，用于存放计算机操作指令。存储器1020可能包含高速RAM存储器，也可能还包括非易失性存储器(non-volatile memory)，例如至少一个磁盘存储器。
处理器1010，用于执行存储器1020中存放的计算机操作指令。处理器1010具体可以是中央处理器(CPU，central processing unit)，是计算机核心单元。
其中，处理器1010执行所述计算机操作指令使得检测装置1000执行上述实施例1所述的方法。
根据本发明的实施例3，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
根据本发明实施例1-3，本发明实施例4提供了一种检测设备400。如图8所示，所述检测设备400包括媒体单元4010和检测装置4020。
所述媒体单元4010，用于获取所述视频(后续称为待检视频)，并将所述视频传输给所述检测装置4020。所述媒体单元4010具体可以从视频文件中读取所述待检视频，也可以是从接收的视频服务器发送的媒体流中获取所述待检视频。所述待检视频具体可以是一个完整的视频，也可以是一个视频中的一个视频段，如果是视频段，所述媒体单元4010可以将所述视频段所在的视频(即包括所述视频段的视频)传输给所述检测装置4020并由所述检测装置4020对接收的视频中的视频段进行检测以检测出所述待检视频中的场景切换帧。
所述检测装置4020，具体可以为实施例2提供的检测装置200或实施例3提供的检测装置1000，用于从所述媒体单元4010获取所述待检视频并执行实施例2提供的检测装置200或实施例3提供的检测装置1000所执行的操作。
所述检测装置4020还可以进一步根据检测出的场景切换帧对所述待检视频或所述待检视频所在的视频的质量进行评估。
根据本发明的实施例4，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
根据本发明实施例1-3，本发明实施例5提供了一种实现视频质量评估的系统2000。如图9A所示，系统2000包括视频服务器2010、传输设备2020和视频终端2030，其中，所述视频服务器2010发送的视频流经由所述传输设备2020传输到所述视频终端2030。
在一种具体实现方式中，所述传输设备2020或所述视频终端2030具体可以包括实施例2提供的检测装置200或实施例3提供的检测装置1000。在具体实现时，所述传输设备2020和所述视频终端2030可以均包括实施例2提供的检测装置200或实施例3提供的检测装置1000。所述传输设备2020或所述视频终端2030具体可以为实施例4提供的检测设备400。
在另一种具体实现方式中，所述系统还包括检测装置2040，如图9B和9C所示，所述检测装置2040具体可以为实施例2提供的检测装置200或实施例3提供的检测装置1000，所述传输设备2020或所述视频终端2030与所述检测装置2040相连，并且所述检测装置2040通过与所述检测装置2040连接的所述传输设备2020或所述视频终端2030获取所述视频流。在具体实现时，所述传输设备2020和所述视频终端2030可以分别与一个检测装置2040相连。
根据本发明的实施例5，在检测视频中的场景切换帧时，可以检测出P帧中的场景切换帧，从而有效减少了漏检场景切换帧的问题。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本发明的范围。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装 置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请所提供的几个实施例中，应该理解到，所揭露的系统、装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。
所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本发明各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应所述以权利要求的保护范围为准。

Claims (28)

  1. 一种检测视频中的场景切换帧的方法，其特征在于，所述视频包括N个图像组GOP，N为大于等于2的整数，所述方法包括：
    确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
    Figure PCTCN2017098483-appb-100001
    其中K为变量且取值依次从M至N，1≤M≤N；
    在确定
    Figure PCTCN2017098483-appb-100002
    和
    Figure PCTCN2017098483-appb-100003
    的相对大小大于或等于第一阈值，且
    Figure PCTCN2017098483-appb-100004
    和
    Figure PCTCN2017098483-appb-100005
    的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
    Figure PCTCN2017098483-appb-100006
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
    Figure PCTCN2017098483-appb-100007
    为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
  2. 如权利要求1所述的方法，其特征在于，所述方法还包括：
    根据公式
    Figure PCTCN2017098483-appb-100008
    计算所述第一阈值；或者，
    根据公式
    Figure PCTCN2017098483-appb-100009
    计算所述第一阈值；
    其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
  3. 如权利要求2所述的方法，其特征在于，
    若所述第一阈值根据公式
    Figure PCTCN2017098483-appb-100010
    获得，所述
    Figure PCTCN2017098483-appb-100011
    和
    Figure PCTCN2017098483-appb-100012
    的相对大小
    Figure PCTCN2017098483-appb-100013
    为：
    Figure PCTCN2017098483-appb-100014
    若所述第一阈值根据公式
    Figure PCTCN2017098483-appb-100015
    获得，所述
    Figure PCTCN2017098483-appb-100016
    和
    Figure PCTCN2017098483-appb-100017
    的相对大小
    Figure PCTCN2017098483-appb-100018
    为：
    Figure PCTCN2017098483-appb-100019
  4. 如权利要求2或3所述的方法，其特征在于，所述方法还包括：
    在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据公式
    Figure PCTCN2017098483-appb-100020
    或
    Figure PCTCN2017098483-appb-100021
    计算新的Ithreshold，所述新的Ithreshold用于判断下 一个GOP的Pmax是否为场景切换帧。
  5. 如权利要求1-4任一所述的方法，其特征在于，所述方法还包括：从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧；
    其中，在K小于N时，在判断第K个GOP的Pmax是否为场景切换帧之后判断第K+1个GOP的I帧是否为场景切换帧；
    所述从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧具体包括：在确定第一距离小于或等于距离阈值时，确定所述第K+1个GOP的I帧不是场景切换帧，所述第一距离为所述第K+1个GOP的I帧和所述第K+1个GOP的I帧之前离所述第K+1个GOP的I帧最近的场景切换帧之间的距离。
  6. 如权利要求5所述的方法，其特征在于，所述方法还包括：
    设置所述距离阈值，所述距离阈值为如下三个长度中的其中一个：
    所述N个GOP中长度最长的GOP的长度；
    所述N个GOP的长度的中值或平均值；
    长度L，其中，所述N个GOP中长度为L的GOP的数量最多。
  7. 如权利要求5或6所述的方法，其特征在于，在确定出一个新的场景切换帧时，如果第二距离小于所述距离阈值，将所述距离阈值更新为所述第二距离，其中，所述第二距离具体为所述新的场景切换帧与所述新的场景切换帧之前离所述新的场景切换帧最近的场景切换帧之间的距离。
  8. 一种检测视频中的场景切换帧的方法，其特征在于，所述视频包括N个图像组GOP，N为大于等于2的整数，所述方法包括：
    确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
    Figure PCTCN2017098483-appb-100022
    其中K为变量且取值依次从M至N，1≤M≤N；
    在确定
    Figure PCTCN2017098483-appb-100023
    和
    Figure PCTCN2017098483-appb-100024
    的相对大小大于或等于第一阈值，
    Figure PCTCN2017098483-appb-100025
    和
    Figure PCTCN2017098483-appb-100026
    的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在确定
    Figure PCTCN2017098483-appb-100027
    和
    Figure PCTCN2017098483-appb-100028
    的相对大小大于或等于第一阈值，
    Figure PCTCN2017098483-appb-100029
    和
    Figure PCTCN2017098483-appb-100030
    的相对大小大于或等于第二阈值，所述第K个GOP中有B帧，且
    Figure PCTCN2017098483-appb-100031
    和
    Figure PCTCN2017098483-appb-100032
    的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
    Figure PCTCN2017098483-appb-100033
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或均 值，
    Figure PCTCN2017098483-appb-100034
    为所述第K个GOP中的多个P帧的大小的中值或平均值，
    Figure PCTCN2017098483-appb-100035
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；
    所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
  9. 如权利要求8所述的方法，其特征在于，所述方法还包括：
    根据公式
    Figure PCTCN2017098483-appb-100036
    计算所述第一阈值；或者，
    根据公式
    Figure PCTCN2017098483-appb-100037
    计算所述第一阈值；
    其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
  10. 如权利要求8或9所述的方法，其特征在于，所述方法还包括
    根据公式
    Figure PCTCN2017098483-appb-100038
    计算所述第三阈值；或者，
    根据公式
    Figure PCTCN2017098483-appb-100039
    计算所述第三阈值；
    其中，Bthreshold为所述第三阈值，Pmedian为所述视频中的所有P帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值。
  11. 如权利要求8-10任一所述的方法，其特征在于，所述方法还包括：
    在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据公式
    Figure PCTCN2017098483-appb-100040
    或
    Figure PCTCN2017098483-appb-100041
    计算新的Ithreshold，所述新的Ithreshold用于判断下一个GOP的Pmax是否为场景切换帧。
  12. 如权利要求8-11任一所述的方法，其特征在于，所述方法还包括：从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧；
    其中，在K小于N时，在判断第K个GOP的Pmax是否为场景切换帧之后判断第K+1个GOP的I帧是否为场景切换帧；
    所述从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧具体包括：在 确定第一距离小于或等于距离阈值时，确定所述第K+1个GOP的I帧不是场景切换帧，所述第一距离为所述第K+1个GOP的I帧和所述第K+1个GOP的I帧之前离所述第K+1个GOP的I帧最近的场景切换帧之间的距离。
  13. 如权利要求12所述的方法，其特征在于，所述方法还包括：
    设置所述距离阈值，所述距离阈值为如下三个长度中的其中一个：
    所述N个GOP中长度最长的GOP的长度；
    所述N个GOP的长度的中值或平均值；
    长度L，其中，所述N个GOP中长度为L的GOP的数量最多。
  14. 如权利要求13所述的方法，其特征在于，在确定出一个新的场景切换帧时，如果第二距离小于所述距离阈值，将所述距离阈值更新为所述第二距离，其中，所述第二距离具体为所述新的场景切换帧与所述新的场景切换帧之前离所述新的场景切换帧最近的场景切换帧之间的距离。
  15. 一种用于检测视频中的场景切换帧的检测装置，其特征在于，所述视频包括N个GOP，N为大于等于2的整数，所述检测装置包括：第一确定单元和第二确定单元；
    所述第一确定单元，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
    Figure PCTCN2017098483-appb-100042
    其中K为变量且取值依次从M至N，1≤M≤N；
    所述第二确定单元，用于在确定
    Figure PCTCN2017098483-appb-100043
    和
    Figure PCTCN2017098483-appb-100044
    的相对大小大于或等于第一阈值，且
    Figure PCTCN2017098483-appb-100045
    和
    Figure PCTCN2017098483-appb-100046
    的相对大小大于或等于第二阈值的情况下，确定Pmax为场景切换帧，其中
    Figure PCTCN2017098483-appb-100047
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或平均值，
    Figure PCTCN2017098483-appb-100048
    为所述第K个GOP中的多个P帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1。
  16. 如权利要求15所述的检测装置，其特征在于，
    所述第二确定单元，还用于根据公式
    Figure PCTCN2017098483-appb-100049
    计算所述第一阈值；或者，根据公式
    Figure PCTCN2017098483-appb-100050
    计算所述第一阈值；
    其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值， Pmedian为所述视频中的所有P帧的大小的中值或平均值。
  17. 如权利要求16所述的检测装置，其特征在于
    若所述第一阈值根据公式
    Figure PCTCN2017098483-appb-100051
    获得，所述
    Figure PCTCN2017098483-appb-100052
    和
    Figure PCTCN2017098483-appb-100053
    的相对大小
    Figure PCTCN2017098483-appb-100054
    为：
    Figure PCTCN2017098483-appb-100055
    若所述第一阈值根据公式
    Figure PCTCN2017098483-appb-100056
    获得，所述
    Figure PCTCN2017098483-appb-100057
    和
    Figure PCTCN2017098483-appb-100058
    的相对大小
    Figure PCTCN2017098483-appb-100059
    为：
    Figure PCTCN2017098483-appb-100060
  18. 如权利要求16或17所述的方法，其特征在于，
    所述第二确定单元，还用于在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据公式
    Figure PCTCN2017098483-appb-100061
    或
    Figure PCTCN2017098483-appb-100062
    计算新的Ithreshold，所述新的Ithreshold用于判断下一个GOP的Pmax是否为场景切换帧。
  19. 如权利要求15-18任一所述的检测装置，其特征在于，所述第二确定单元，还用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧；
    其中，在K小于N时，所述第二确定单元在判断第K个GOP的Pmax是否为场景切换帧之后判断第K+1个GOP的I帧是否为场景切换帧；
    所述第二确定单元用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧，具体包括：用于在确定第一距离小于或等于距离阈值时，确定所述第K+1个GOP的I帧不是场景切换帧，所述第一距离为所述第K+1个GOP的I帧和所述第K+1个GOP的I帧之前离所述第K+1个GOP的I帧最近的场景切换帧之间的距离。
  20. 如权利要求19所述的检测装置，其特征在于，所述第二确定单元，还用于设置所述距离阈值，所述距离阈值为如下三个长度中的其中一个：
    所述N个GOP中长度最长的GOP的长度；
    所述N个GOP的长度的中值或平均值；
    长度L，其中，所述N个GOP中长度为L的GOP的数量最多。
  21. 如权利要求19或20所述的检测装置，其特征在于，所述第二确定单元，还用于：在确定出一个新的场景切换帧时，如果第二距离小于所述距离阈值，将所述距离阈值更新为所述第二距离，其中，所述第二距离具体为所述新的场景切换帧与所述新的场景切换帧之前离所述新的场景切换帧最近的场景切换帧之间的距离。
  22. 一种检测视频中的场景切换帧的检测装置，其特征在于，所述视频包括N个图像组GOP，N为大于等于2的整数，所述检测装置包括：第一确定单元和第二确定单元；
    所述第一确定单元，用于确定第K个GOP的所有P帧中最大的P帧Pmax，其中Pmax的大小为
    Figure PCTCN2017098483-appb-100063
    其中K为变量且取值依次从M至N，1≤M≤N；
    所述第二确定单元，用于在确定
    Figure PCTCN2017098483-appb-100064
    和
    Figure PCTCN2017098483-appb-100065
    的相对大小大于或等于第一阈值，
    Figure PCTCN2017098483-appb-100066
    和
    Figure PCTCN2017098483-appb-100067
    的相对大小大于或等于第二阈值，且所述第K个GOP中没有B帧的情况下，或者在确定
    Figure PCTCN2017098483-appb-100068
    和
    Figure PCTCN2017098483-appb-100069
    的相对大小大于或等于第一阈值，
    Figure PCTCN2017098483-appb-100070
    和
    Figure PCTCN2017098483-appb-100071
    的相对大小大于或等于第二阈值，所述第K个GOP中有B帧，且
    Figure PCTCN2017098483-appb-100072
    和
    Figure PCTCN2017098483-appb-100073
    的相对大小大于或等于第三阈值的情况下，确定Pmax为场景切换帧，其中
    Figure PCTCN2017098483-appb-100074
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的多个I帧的大小的中值或均值，
    Figure PCTCN2017098483-appb-100075
    为所述第K个GOP中的多个P帧的大小的中值或平均值，
    Figure PCTCN2017098483-appb-100076
    为Pmax与Pmax之前离Pmax最近的场景切换帧之间的所有B帧的大小的中值或平均值；所述第一阈值大于0且小于1，所述第二阈值大于1，所述第三阈值大于1。
  23. 如权利要求22所述的检测装置，其特征在于，所述第二确定单元，还用于：
    根据公式
    Figure PCTCN2017098483-appb-100077
    计算所述第一阈值；或者，
    根据公式
    Figure PCTCN2017098483-appb-100078
    计算所述第一阈值；
    其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Pmedian为所述视频中的所有P帧的大小的中值或平均值。
  24. 如权利要求22或23所述的检测装置，其特征在于，所述第二确定单元，还用于：
    根据公式
    Figure PCTCN2017098483-appb-100079
    计算所述第三阈值；或者，
    根据公式
    Figure PCTCN2017098483-appb-100080
    计算所述第三阈值；
    其中，Ithreshold为所述第一阈值，Imedian为所述视频中的所有I帧的大小的中值或平均值，Bmedian为所述视频中的所有B帧的大小的中值或平均值。
  25. 如权利要求22-24任一所述的检测装置，其特征在于，所述第二确定单元，还用于在确定第K个GOP的Pmax为场景切换帧后，将所述视频中除去已确定为场景切换帧的P帧之外的其它P帧的大小的中值或平均值作为新的Pmedian，并根据公式
    Figure PCTCN2017098483-appb-100081
    或
    Figure PCTCN2017098483-appb-100082
    计算新的Ithreshold，所述新的Ithreshold用于判断下一个GOP的Pmax是否为场景切换帧。
  26. 如权利要求22-25任一所述的检测装置，其特征在于，所述第二确定单元，还用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧；
    其中，在K小于N时，在判断第K个GOP的Pmax是否为场景切换帧之后判断第K+1个GOP的I帧是否为场景切换帧；
    所述第二确定单元用于从所述N个GOP中的第M个至第N个GOP的I帧中检测场景切换帧，具体包括：用于在确定第一距离小于或等于距离阈值时，确定所述第K+1个GOP的I帧不是场景切换帧，所述第一距离为所述第K+1个GOP的I帧和所述第K+1个GOP的I帧之前离所述第K+1个GOP的I帧最近的场景切换帧之间的距离。
  27. 如权利要求26所述的检测装置，其特征在于，所述第二确定单元，还用于设置所述距离阈值，所述距离阈值为如下三个长度中的其中一个：
    所述N个GOP中长度最长的GOP的长度；
    所述N个GOP的长度的中值或平均值；
    长度L，其中，所述N个GOP中长度为L的GOP的数量最多。
  28. 如权利要求27所述的检测装置，其特征在于，所述第二确定单元，还用于：在确定出一个新的场景切换帧时，如果第二距离小于所述距离阈值，将所述距离阈值更新为所述第二距离，其中，所述第二距离具体为所述新的场景切换帧与所述新的场景切换帧之前离所述新的场景切换帧最近的场景切换帧之间的距离。
PCT/CN2017/098483 2016-08-23 2017-08-22 一种检测场景切换帧的方法、装置和系统 WO2018036481A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP17842903.1A EP3499460A1 (en) 2016-08-23 2017-08-22 Method, device, and system for detecting scene change frame
KR1020197007484A KR20190039265A (ko) 2016-08-23 2017-08-22 장면 변화 프레임을 검출하기 위한 방법, 디바이스, 및 시스템
JP2019510927A JP2019528643A (ja) 2016-08-23 2017-08-22 シーンチェンジフレームを検出するための方法および装置ならびにシステム
US16/284,664 US10917643B2 (en) 2016-08-23 2019-02-25 Method and apparatus for detecting scene change frame and system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610708531.5A CN107770538B (zh) 2016-08-23 2016-08-23 一种检测场景切换帧的方法、装置和系统
CN201610708531.5 2016-08-23

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/284,664 Continuation US10917643B2 (en) 2016-08-23 2019-02-25 Method and apparatus for detecting scene change frame and system

Publications (1)

Publication Number Publication Date
WO2018036481A1 true WO2018036481A1 (zh) 2018-03-01

Family

ID=61245503

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/098483 WO2018036481A1 (zh) 2016-08-23 2017-08-22 一种检测场景切换帧的方法、装置和系统

Country Status (6)

Country Link
US (1) US10917643B2 (zh)
EP (1) EP3499460A1 (zh)
JP (1) JP2019528643A (zh)
KR (1) KR20190039265A (zh)
CN (1) CN107770538B (zh)
WO (1) WO2018036481A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111629261A (zh) * 2019-02-28 2020-09-04 阿里巴巴集团控股有限公司 信息处理方法、装置、电子设备及计算机可读存储介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114143606A (zh) 2018-08-22 2022-03-04 华为技术有限公司 一种实现视频流切换的方法、装置和系统
CN110891182B (zh) * 2018-09-11 2022-04-12 华为技术有限公司 一种实现视频流切换的方法、装置和系统
CN109168001B (zh) * 2018-09-27 2021-02-12 苏州科达科技股份有限公司 视频场景变化的检测方法、装置及视频采集设备
CN109361923B (zh) * 2018-12-04 2022-05-31 深圳市梦网视讯有限公司 一种基于运动分析的滑动时间窗场景切换检测方法和系统
CN112019850B (zh) * 2020-08-27 2022-08-23 广州市百果园信息技术有限公司 基于场景切换的图像组划分方法、视频编码方法及装置
CN112351278B (zh) * 2020-11-04 2023-07-07 北京金山云网络技术有限公司 一种视频的编码方法和装置，视频的解码方法和装置
US11743474B2 (en) 2021-08-27 2023-08-29 Meta Platforms, Inc. Shot-change detection using container level information

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1758759A (zh) * 2005-11-18 2006-04-12 清华大学 一种视频场景切换检测方法
CN101072342A (zh) * 2006-07-01 2007-11-14 腾讯科技（深圳）有限公司 一种场景切换的检测方法及其检测系统
CN102630013A (zh) * 2012-04-01 2012-08-08 北京捷成世纪科技股份有限公司 基于场景切换的码率控制视频压缩方法和装置
CN103826121A (zh) * 2013-12-20 2014-05-28 电子科技大学 低延迟视频编码基于场景切换检测的码率控制方法
CN103945281A (zh) * 2014-04-29 2014-07-23 中国联合网络通信集团有限公司 视频传输处理方法、装置和系统
WO2016027410A1 (ja) * 2014-08-21 2016-02-25 パナソニックＩｐマネジメント株式会社 検知装置および検知システム

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3315766B2 (ja) * 1992-09-07 2002-08-19 富士通株式会社 画像データ符号化方法、その方法を用いた画像データ符号化装置、画像データ復元方法、その方法を用いた画像データ復元装置、シーン変化検出方法、その方法を用いたシーン変化検出装置、シーン変化記録装置、及び画像データのシーン変化記録・再生装置
JPH09322174A (ja) 1996-05-30 1997-12-12 Hitachi Ltd 動画データの再生方法
JP2002010254A (ja) 2000-06-20 2002-01-11 Sony Corp 特徴点検出方法および記録再生装置
US7525579B2 (en) * 2004-12-27 2009-04-28 Konica Minolta Holdings, Inc. Image sensing apparatus and image processing method for use therein
US8208536B2 (en) * 2005-04-28 2012-06-26 Apple Inc. Method and apparatus for encoding using single pass rate controller
JP2010219929A (ja) 2009-03-17 2010-09-30 Oki Networks Co Ltd 動画像解析装置、プログラム及び方法、並びに、動画像処理装置及び動画像配信装置
US8588296B2 (en) * 2009-07-02 2013-11-19 Dialogic Corporation Bitrate control algorithm for video transcoding systems
CN102740108B (zh) 2011-04-11 2015-07-08 华为技术有限公司 一种视频数据质量评估方法和装置
US10063866B2 (en) * 2015-01-07 2018-08-28 Texas Instruments Incorporated Multi-pass video encoding

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1758759A (zh) * 2005-11-18 2006-04-12 清华大学 一种视频场景切换检测方法
CN101072342A (zh) * 2006-07-01 2007-11-14 腾讯科技（深圳）有限公司 一种场景切换的检测方法及其检测系统
CN102630013A (zh) * 2012-04-01 2012-08-08 北京捷成世纪科技股份有限公司 基于场景切换的码率控制视频压缩方法和装置
CN103826121A (zh) * 2013-12-20 2014-05-28 电子科技大学 低延迟视频编码基于场景切换检测的码率控制方法
CN103945281A (zh) * 2014-04-29 2014-07-23 中国联合网络通信集团有限公司 视频传输处理方法、装置和系统
WO2016027410A1 (ja) * 2014-08-21 2016-02-25 パナソニックＩｐマネジメント株式会社 検知装置および検知システム

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ITU-T P: "P.1201.2 - Parametric Non-Intrusive Assessment of Audiovisual Media Streaming Quality-Higher Resolution Application Area", SERIES P: TERMINALS AND SUBJECTIVE AND OBJECTIVE ASSESSMENT METHODS MODELS AND TOOLS FOR QUALITY ASSESSMENT OF STREAMED MEDIA, 31 October 2012 (2012-10-31), XP044142718 *
See also references of EP3499460A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111629261A (zh) * 2019-02-28 2020-09-04 阿里巴巴集团控股有限公司 信息处理方法、装置、电子设备及计算机可读存储介质
CN111629261B (zh) * 2019-02-28 2022-04-22 阿里巴巴集团控股有限公司 信息处理方法、装置、电子设备及计算机可读存储介质

Also Published As

Publication number Publication date
CN107770538A (zh) 2018-03-06
KR20190039265A (ko) 2019-04-10
EP3499460A4 (en) 2019-06-19
US20190260999A1 (en) 2019-08-22
CN107770538B (zh) 2020-09-11
US10917643B2 (en) 2021-02-09
EP3499460A1 (en) 2019-06-19
JP2019528643A (ja) 2019-10-10

Similar Documents

Publication Publication Date Title
WO2018036481A1 (zh) 一种检测场景切换帧的方法、装置和系统
US11423942B2 (en) Reference and non-reference video quality evaluation
KR101414435B1 (ko) 비디오 스트림 품질 평가 방법 및 장치
JP5938786B2 (ja) ビデオデータ品質評価方法およびビデオデータ品質評価デバイス
US10109316B2 (en) Method and apparatus for playing back recorded video
EP2876881B1 (en) Method and system for determining a quality value of a video stream
CN109155840B (zh) 运动图像分割装置及监视方法
US11310489B2 (en) Method, apparatus, and system for implementing video quality assessment
JP2014519735A (ja) ビットストリームレベルで動画品質を推定する方法及び装置
Yamada et al. Accurate video-quality estimation without video decoding
KR102350570B1 (ko) 영상프레임의 손실을 측정하기 위한 iptv 셋탑박스 및 그 동작방법
CN110602507A (zh) 丢帧处理方法、设备及系统
Venkatesh Babu et al. Evaluation and monitoring of video quality for UMA enabled video streaming systems
Kwon et al. A novel video quality impairment monitoring scheme over an ipty service with packet loss
KR101199470B1 (ko) 주관적 화질 열화 측정 장치
KR20150046036A (ko) 비디오 비트스트림에서 점진적 전환 픽쳐를 검출하기 위한 방법 및 장치

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: 17842903

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2019510927

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 20197007484

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2017842903

Country of ref document: EP

Effective date: 20190313