JP4802200B2 - Video quality estimation apparatus, method, and program - Google Patents

Description

  The present invention relates to video communication technology, and more particularly to a system for estimating the quality of video communication such as an IPTV service, a video distribution service, and a video communication service performed via a network.

  With the increase in the speed and bandwidth of Internet access lines, video communication services using audio and video media are expected. Since the Internet is not necessarily a network with guaranteed communication quality, when communication is performed using audio and video media, if the network bandwidth of the network between users is narrow or the line is congested, the audio and video media For example, the quality perceived by the user, that is, the quality of experience (QoE) is deteriorated. Specifically, when quality degradation is added to the video, it is perceived as blurring, blurring, mosaic distortion, jerky feeling, and the like.

In order to provide the above services with high quality, quality design prior to service provision and quality management after service start are important. For this purpose, the quality enjoyed by the user can be expressed appropriately, and it is simple and efficient. Quality evaluation technology is required.
Conventionally, ITU-T (International Telecommunication Union Telecommunication Standardization Sector) recommendation J.I. 144 describes an objective video quality evaluation method. In addition, ITU-T Recommendation G. 1070 describes a quality estimation method for performing quality design of a videophone. These objective evaluation techniques can estimate the subjective quality with an estimation error similar to the statistical ambiguity of the subjective quality under certain conditions.

ITU-T Recommendation J.I. 144 ITU-T Recommendation G. 1070

However, such a conventional technique has a problem that it cannot be applied when estimating the quality of video communication such as IPTV service, video distribution service, and video communication service performed via a network.
First, ITU-T Recommendation J.I. 144 is a quality estimation method using a video media signal (pixel signal), and the amount of calculation related to quality estimation is enormous. Therefore, it is not suitable for extracting quality information from between networks and estimating and managing quality in a large-scale network.

  In addition, ITU-T Recommendation G. Reference numeral 1070 denotes a model for estimating video representative quality from a quality parameter that affects video quality. FIG. 11 shows ITU-T recommendation G.264. It is explanatory drawing which shows the quality estimation precision of the quality estimation model used by 1070. That is, ITU-T recommendation G.I. As shown in FIG. 11, 1070 is estimated based on an average characteristic T obtained from a large number of different videos regarding the amount of motion of the video and the definition of each frame. For this reason, with respect to video A and video B in which the quality of each video, for example, the amount of motion of the video and the resolution in units of frames are different from the average video, errors ΔVa and ΔVb are different from the actual video quality. This causes a large estimation error in the quality of the videos A and B.

  The present invention is intended to solve such problems, and improves the estimation accuracy of video quality even if the video motion amount and the resolution in units of frames are different from the average video. An object of the present invention is to provide a video quality estimation apparatus, method, and program that can be used.

  In order to achieve such an object, a video quality estimation apparatus according to the present invention estimates video quality of video communication in which video signals are compressed and encoded into a plurality of frames including I frames and transmitted in packets. A device that analyzes an input packet and extracts an individual quantization parameter related to a quantization step width of the video and a quality parameter related to video quality of the video communication; and a plurality of video A video representative quality estimation unit for estimating a video representative quality corresponding to a quality parameter of the video communication from a video representative quality estimation model for estimating a video representative quality obtained by averaging the video quality, a video representative quality and the video From the video differential quality estimation model that estimates the video differential quality corresponding to the difference between the video individual quality and the individual quantization parameter of the video communication Includes a video difference quality estimation unit that estimates an image difference quality corresponding to the quality parameters, the video individual quality calculator for calculating the image individual quality of the video communication and a video representative quality and image difference quality.

  At this time, the difference quantization parameter calculation that calculates the difference quantization parameter from the difference between the individual quantization parameter and the representative quantization parameter related to the quantization step width of the image obtained from a plurality of images in the image difference quality estimation unit. Video maximum difference quality estimation that estimates the maximum video differential quality corresponding to the quality parameter from the video maximum differential quality estimation model that estimates the maximum video differential quality that indicates the maximum difference between the video representative quality and the video individual quality And a video difference quality calculation unit that calculates an individual video quality corresponding to the video difference quality and the difference quantization parameter of the video communication from the video difference quality estimation model.

  The video difference quality estimation unit further includes a representative quantization parameter estimation unit that estimates a representative quantization parameter corresponding to the quality parameter from a representative quantization parameter estimation model indicating a relationship between the representative quantization parameter and the quality parameter. May be.

  Further, as the representative quantization parameter estimation model, a characteristic that the representative quantization parameter decreases as the quality parameter including the encoding rate increases may be used.

  Further, as the video maximum difference quality estimation model, the video maximum difference quality increases (or decreases) as the quality parameter including the encoding rate increases, and the video maximum difference quality decreases as the quality parameter further increases ( Or, an increasing characteristic may be used.

  Further, as the video difference quality estimation model, a characteristic that the video difference quality increases as the product of the difference quantization parameter and the video maximum difference quality increases may be used.

  The video quality estimation method according to the present invention is a video quality estimation method for estimating video quality of video communication in which a video signal is compressed and encoded into a plurality of frames including an I frame and transmitted in packets. A packet analysis step for analyzing a packet and extracting an individual quantization parameter relating to the quantization step width of the video and a quality parameter relating to the video quality of the video communication, and averaging the video quality of a plurality of videos. From the video representative quality estimation model for estimating the video representative quality, the video representative quality estimation step for estimating the video representative quality corresponding to the quality parameter of the video communication and the difference between the video representative quality and the individual video quality of the video From the video difference quality estimation model that estimates the corresponding video difference quality, individual quantization parameters and quality parameters of the video communication Includes a video differential quality estimation step of estimating an image difference quality corresponding, the video individual quality calculation step of calculating a video individual quality of the video communication and a video representative quality and image difference quality.

  In addition, the program according to the present invention analyzes a packet input to a computer of a video quality estimation device that estimates video quality of video communication in which a video signal is compressed and encoded into a plurality of frames including an I frame and transmitted in packets. A packet analysis step for extracting an individual quantization parameter relating to the quantization step width of the video and a quality parameter relating to the video quality of the video communication, and a video representative obtained by averaging video quality of a plurality of videos A video representative quality estimation step for estimating a video representative quality corresponding to a quality parameter of the video communication from a video representative quality estimation model for estimating the quality, and a video corresponding to a difference between the video representative quality and the video individual quality of the video From the video difference quality estimation model that estimates the difference quality, the individual quantization parameter and quality parameter of the video communication. A video differential quality estimation step of estimating an image difference quality corresponding to bets, to execute the video individual quality calculation step of calculating a video individual quality of the video communication and a video representative quality and image difference quality.

According to the present invention, it is possible to estimate the video quality in consideration of the individual quantization parameter of the target video communication. For this reason, video quality with less error is estimated based on parameters obtained from IP packets, even for video communications where the amount of motion of the video and the resolution in units of frames are different from the average video. Can do.
Therefore, it is possible to easily determine whether or not video quality of a certain level or more is maintained for a user who uses a video communication service such as an IPTV service, a video distribution service, and a video communication service performed via a network. As a result, it is possible to design and control the quality parameters used in the service and the actual quality of the service being provided.

Next, embodiments of the present invention will be described with reference to the drawings.
First, a video quality estimation apparatus according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a configuration of a video quality estimation apparatus according to an embodiment of the present invention.
The video quality estimation apparatus 1 is composed of an information processing apparatus such as a computer that performs arithmetic processing on input information. After compressing and encoding a video signal encoded into a plurality of frames, the communication terminal 2A to the communication terminal 2B terminal. Estimating the subjective video quality perceived by the viewer from the video that is decoded and reproduced by the communication terminal 2B using the video communication packet as an input for video communication transmitted in packets via the IP network 3 such as the Internet This is a device that calculates a value as individual video quality.

  The video quality estimation apparatus 1 includes, as main functional units, an operation input unit 11, a screen display unit 12, a communication interface unit (hereinafter referred to as a communication I / F unit) 13, a storage unit 14, a characteristic coefficient storage unit 15, An arithmetic processing unit 16 is provided. The arithmetic processing unit 16 includes, as main processing units, a packet analysis unit 16A, a video representative quality estimation unit 16B, a video maximum difference quality estimation unit 16C, a representative quantization parameter estimation unit 16D, a differential quantization parameter calculation unit 16E, A video difference quality calculation unit 16F and a video individual quality calculation unit 16G are provided. Among these, the video difference quality estimation unit 17 is configured by the video maximum difference quality estimation unit 16C, the representative quantization parameter estimation unit 16D, the difference quantization parameter calculation unit 16E, and the video difference quality calculation unit 16F.

In compression coding schemes such as MPEG-2 (Moving Picture Experts Group-2), I-frame information allocates code amounts for B frames and P frames in the inter-frame prediction scheme in the case of video with a large amount of motion. In the case of an image to be tried and a video with high definition, there is a feature that the code amount of the B frame or the P frame is reduced and the code amount is increased for the I frame of the intra-frame prediction method.
In the present embodiment, paying attention to the above feature that the individual quantization parameter changes according to the amount of motion of the video and the definition of each frame, the video quality is considered in consideration of the individual quantization parameter of the target video communication. Estimated.

  Specifically, the packet analysis unit 16A analyzes the input packet, extracts individual quantization parameters related to the quantization step width of the video, and quality parameters related to the video quality of the video communication. The quality estimation unit 16B estimates the video representative quality corresponding to the quality parameter of the video communication from the video representative quality estimation model for estimating the video representative quality obtained by averaging the video quality of a plurality of videos, and the video difference Corresponding to the individual quantization parameter and the quality parameter of the video communication from the video differential quality estimation model for estimating the video differential quality corresponding to the difference between the video representative quality and the video individual quality of the video by the quality estimation unit 17. The video differential quality is estimated, and the video individual quality calculation unit 16G uses the video representative quality and the video differential quality to determine the video individual video communication quality. To calculate the quality.

[Configuration of video quality estimation device]
Next, the configuration of the video quality estimation apparatus according to the embodiment of the present invention will be described in detail with reference to FIG. 1 and FIG. FIG. 2 is an explanatory diagram showing an operation flow of the video quality estimation apparatus according to the embodiment of the present invention.

The operation input unit 11 includes an operation input device such as a keyboard and a mouse, and has a function of detecting an operator operation and outputting the operation to the arithmetic processing unit 16.
The screen display unit 12 includes a screen display device such as an LCD or a PDP, and has a function of displaying an operation menu and video individual quality as an estimation result on the screen in response to an instruction from the arithmetic processing unit 16.
The communication I / F unit 13 includes a dedicated communication circuit and has a function of performing data communication with the IP network 3 and an external device (not shown). The video communication packet subject to quality estimation may be acquired by the communication I / F unit 13 via the IP network 3, or may be acquired by the communication I / F unit 13 from the communication terminal 2B on the receiving side. Good.

  The storage unit 14 includes a storage device such as a hard disk and a memory, and has a function of storing various processing information used for processing operations in the arithmetic processing unit 16 and a program 14P. The program 14P is read in advance from an external device or a recording medium (not shown) via a data input / output function such as the communication I / F unit 13 provided in the video quality estimation device 1, and stored in the storage unit 14. ing.

Main processing information stored in the storage unit 14 includes an individual quantization parameter 14A, a coding rate 14B, and a characteristic coefficient specifying parameter 14C.
Of these, the individual quantization parameter 14 </ b> A and the coding rate 14 </ b> B are both extracted by the packet analysis unit 16 </ b> A from the target video communication packet input from the communication I / F unit 13 and stored in the storage unit 14. . The characteristic coefficient specifying parameter 14 </ b> C is stored in the storage unit 14 in advance via the operation input unit 11 and the communication I / F unit 13.

The individual quantization parameter 14 </ b> A is a parameter related to a quantization step width related to a video of target video communication.
The encoding rate 14B is the amount of information transmitted per unit time in the target video communication.
The characteristic coefficient specifying parameter 14C includes the compression encoding method used in the target video communication, the display resolution when the video is displayed on the screen, the key frame related to encoding, the monitor size related to the media playback function in the terminal, the monitor It consists of parameters such as resolution and room illuminance during media playback on the terminal.

  The characteristic coefficient storage unit 15 includes a storage device such as a hard disk or a memory, and includes a plurality of individual storage units that store various characteristic coefficients used for processing operations in the arithmetic processing unit 16. As main individual storage units of the characteristic coefficient storage unit 15, a video representative quality characteristic coefficient storage unit 15A, a video maximum difference quality characteristic coefficient storage unit 15B, a representative quantization parameter characteristic coefficient storage unit 15C, and a video differential quality characteristic coefficient storage unit There is 15D.

  The video representative quality characteristic coefficient storage unit 15A specifies a video representative quality characteristic coefficient for specifying a video representative quality estimation model for estimating the video representative quality 22 from the coding rate 14B in correspondence with the value of the characteristic coefficient specifying parameter 14C. A function of storing in advance, and a function of reading a video representative quality characteristic coefficient corresponding to the inputted characteristic coefficient specifying parameter 14C and outputting it to the video representative quality estimation unit 16B.

  The video maximum difference quality characteristic coefficient storage unit 15B sets the video maximum difference quality characteristic coefficient for specifying the video maximum difference quality estimation model for estimating the video maximum difference quality 23 from the encoding rate 14B, and the value of the characteristic coefficient specifying parameter 14C. And a function of reading out the video maximum difference quality characteristic coefficient corresponding to the inputted characteristic coefficient specifying parameter 14C and outputting it to the video maximum difference quality estimation unit 16C.

  The representative quantization parameter characteristic coefficient storage unit 15C sets the representative quantization parameter characteristic coefficient for specifying the representative quantization parameter estimation model for estimating the representative quantization parameter 24 from the coding rate 14B, and the value of the characteristic coefficient specification parameter 14C. And a function of reading a representative quantization parameter characteristic coefficient corresponding to the inputted characteristic coefficient specifying parameter 14C and outputting it to the representative quantization parameter estimation unit 16D.

  The video difference quality characteristic coefficient storage unit 15D sets a video difference quality characteristic coefficient for specifying a video difference quality estimation model for estimating the video difference quality 26 from the video maximum difference quality 23 and the difference quantization parameter 25, and a characteristic coefficient specifying parameter. A function of storing in advance corresponding to the value of 14C, and a function of reading the video quality characteristic coefficient corresponding to the video maximum difference quality 23 and the difference quantization parameter 25 and outputting the same to the video difference quality calculation unit 16F. is doing.

  FIG. 3 is an explanatory diagram showing the relationship between the characteristic coefficient specifying parameter and various characteristic coefficients. In the example of FIG. 3, CODEC (encoding system), video format, and service type are used as the characteristic coefficient specifying parameter 14C. For CODEC, see H.C. 264, H. 263, H261, MPEG1, MPEG2, MPG4, etc., the video format is HD, SD, VGA, QVGA, CIF, QCIF, etc., and the service type is IPTV, VoD, TV phone, TV Take a value such as a meeting.

In the characteristic coefficient storage unit 15, for each combination of the characteristic coefficient specifying parameters 14C, the video representative quality characteristic coefficients (v1 to v3), the video maximum difference quality characteristic coefficients (v4 to v9), and the representative quantization parameter characteristic coefficients (t1 to t1). t3) and the video differential quality characteristic coefficient (a, b) are associated in advance, and the video representative quality characteristic coefficient storage unit 15A, the video maximum differential quality characteristic coefficient storage unit 15B, and the representative quantization parameter characteristic coefficient storage unit 15C, respectively. , And the video difference quality characteristic coefficient storage unit 15D.

The arithmetic processing unit 16 includes a microprocessor such as a CPU and its peripheral circuits, and reads and executes the program 14P in the storage unit 14, thereby causing the hardware and the program 14P to cooperate with each other for video quality estimation processing. It has a function to realize various processing units to be used.
As main processing units realized by the arithmetic processing unit 16, a packet analysis unit 16A, a video representative quality estimation unit 16B, a video maximum difference quality estimation unit 16C, a representative quantization parameter estimation unit 16D, a differential quantization parameter calculation unit 16E, A video difference quality calculation unit 16F and a video individual quality calculation unit 16G are provided. Among these, the video difference quality estimation unit 17 is configured by the video maximum difference quality estimation unit 16C, the representative quantization parameter estimation unit 16D, the difference quantization parameter calculation unit 16E, and the video difference quality calculation unit 16F.

  The packet analysis unit 16A has a function of analyzing an input packet relating to target video communication, extracting the individual quantization parameter 14A and the encoding rate 14B, and storing them in the storage unit 14.

  The video representative quality estimation unit 16B has a function of deriving a video representative quality estimation model from the video representative quality characteristic coefficient read from the video representative quality characteristic coefficient storage unit 15A according to the characteristic coefficient specifying parameter 14C, and the video representative quality And a function of estimating the video representative quality 22 corresponding to the encoding rate 14B of the storage unit 14 from the estimation model.

  The video maximum difference quality estimation unit 16C derives a video maximum difference quality estimation model from the video maximum difference quality characteristic coefficient read from the video maximum difference quality characteristic coefficient storage unit 15B in accordance with the characteristic coefficient specifying parameter 14C. It has a function of estimating the video maximum difference quality 23 corresponding to the coding rate 14B of the storage unit 14 from this video maximum difference quality estimation model.

  The representative quantization parameter estimation unit 16D derives a representative quantization parameter estimation model from the representative quantization parameter characteristic coefficient read from the representative quantization parameter characteristic coefficient storage unit 15C according to the characteristic coefficient specifying parameter 14C, The representative quantization parameter estimation model has a function of estimating the representative quantization parameter 24 corresponding to the coding rate 14B of the storage unit 14.

  The differential quantization parameter calculation unit 16E has a function of calculating the differential quantization parameter 25 from the representative quantization parameter 24 estimated by the representative quantization parameter estimation unit 16D and the individual quantization parameter 14A of the storage unit 14. ing.

  The video difference quality calculation unit 16F has a function of deriving a video difference quality estimation model from the video difference quality characteristic coefficient read from the video difference quality characteristic coefficient storage unit 15D according to the characteristic coefficient specifying parameter 14C, and the video difference quality. A function of estimating the video difference quality 26 corresponding to the video maximum difference quality 23 estimated by the video maximum difference quality estimation unit 16C and the difference quantization parameter 25 calculated by the difference quantization parameter calculation unit 16E from the estimation model; have.

  The video individual quality calculating unit 16G has a function of calculating the video individual quality 27 from the video representative quality 22 estimated by the video representative quality estimating unit 16B and the video differential quality 26 estimated by the video differential quality calculating unit 16F. ing.

[Operation of video quality estimation device]
Next, with reference to FIGS. 2 to 4, the operation of the video quality estimation apparatus according to the embodiment of the present invention will be described. FIG. 4 is a flowchart showing video individual quality estimation processing of the video quality estimation apparatus according to the embodiment of the present invention.

First, definitions of various data used in the following description will be described.
The video individual quality 27 (Vq _in ) indicates individual video quality for each video communication, and the video quality estimation device 1 estimates the video individual quality 27 based on the inputted packet relating to the target video communication. To do. On the other hand, the video representative quality 22 (Vq _rep ) is a video quality defined by the average of arbitrary video subsets in an infinite video set. Actually, the individual video quality 27 and the representative video quality 22 are composed of subjective video quality that the viewer feels from the video decoded and reproduced by the communication terminal 2B on the receiving side.

The video difference quality 26 (ΔIc _± ) is a value defined by the difference between the video representative quality 22 and the video individual quality 27. However, the video difference quality 26 takes positive and negative values.
The video maximum difference quality 23 (ΔIc _{Max ±} ) is a value defined by the difference quality that maximizes the difference between the video representative quality 22 and the video individual quality 27, for example, the video representative quality 22 and any video A. If the difference from the video individual quality 27 is the maximum, this value is defined as the video maximum difference quality. However, the video maximum difference quality 23 takes a positive or negative value.

The individual quantization parameter 14A (QP _in ) is defined by an average value of quantization parameters of individual video blocks extracted from the IP packet of the target video in a certain quality estimation section (for example, 10 seconds, 100 seconds, etc.). Is done. Actually, the quantization parameter of each video block is extracted from the IP packet and is calculated by averaging within the estimation interval.

The representative quantization parameter 24 (QP _rep ) is defined as a representative value obtained by averaging an average value of quantization parameters calculated from IP packets of each video in the video subset in the quality estimation section. Is done.
The difference quantization parameter 25 (ΔQP) is defined by a difference value between the individual quantization parameter 14A and the representative quantization parameter 24.

Next, the video individual quality estimation process of the video quality estimation apparatus 1 will be described in detail with reference to FIGS.
The arithmetic processing unit 16 of the video quality estimation apparatus 1 starts the video individual quality estimation process of FIGS. 2 and 4 in response to the operator's processing start instruction detected by the operation input unit 11.

First, the arithmetic processing unit 16 analyzes the IP packet used for video communication received from the IP network 3 by the communication I / F unit 13 by the packet analysis unit 16A. This IP packet includes an IP header, a TS (Transport stream) packet, an ES (Elementary Stream), and the like.
The packet analysis unit 16A extracts the individual quantization parameter 14A and the coding rate 14B from these input IP packets, and stores them in the storage unit 14 (step 100).

Next, the arithmetic processing unit 16 uses the video representative quality characteristic coefficient (v) from the video representative quality characteristic coefficient storage unit 15A according to the characteristic coefficient specifying parameter 14C previously stored in the storage unit 14 by the video representative quality estimation unit 16B. _{1 to} v ₃ ) are read out, and a video representative quality estimation model is derived based on these video representative quality characteristic coefficients (step 101). The video representative quality estimation unit 16B estimates the video representative quality 22 corresponding to the encoding rate 14B of the storage unit 14 from the video representative quality estimation model (step 102).

FIG. 5 is a graph showing the relationship between video representative quality and encoding rate. Usually, the video representative quality 22 has a characteristic of gradually approaching a certain maximum (minimum) quality as the encoding rate increases (decreases). The video representative quality estimation model can be approximated using this characteristic. When the video representative quality characteristic value (a value obtained by subtracting 1 from the video representative quality) is Ic _rep and the encoding rate is Br, the video representative quality Vq _rep is expressed by the following functional equation (1). The image representative quality characteristic coefficients v _{1 to} v ₃ are constants.

Next, the arithmetic processing unit 16 uses the video maximum difference quality characteristic coefficient (v ₄ ~) from the video maximum difference quality characteristic coefficient storage unit 15B according to the characteristic coefficient specifying parameter 14C of the storage unit 14 by the video maximum difference quality estimation unit 16C. v ₉₎ reads, derives the video maximum differential quality estimation model these picture maximum differential quality characteristic coefficient (step 103). The video maximum difference quality estimation unit 16C estimates the video maximum difference quality 23 corresponding to the encoding rate 14B of the storage unit 14 from the video maximum difference quality estimation model (step 104).

FIG. 6 is a graph showing the relationship between the video maximum difference quality and the logarithmic value of the encoding rate. When the difference quantization parameter 25 is larger than 0, the video maximum difference quality 23 has a characteristic that the video maximum difference quality increases as the coding rate increases and the video maximum difference quality decreases as the coding parameter further increases. Yes. In addition, when the difference quantization parameter 25 is smaller than 0, the video maximum difference quality 23 decreases as the encoding rate increases, and the video maximum difference quality increases as the encoding parameter further increases. is doing. The video maximum difference quality estimation model can be approximated using this characteristic. When the encoding rate is Br, the differential quantization parameter ΔIc _{Max ±} is expressed by the following function equation (2). However, the functional expressions ΔIc _{Max +} and ΔIc _Max− are switched according to the sign of the differential quantization parameter 25 (ΔQP). The video maximum difference quality characteristic coefficient (v _{4 to} v ₉ ) is a constant.

Next, the arithmetic processing unit 16 causes the representative quantization parameter estimation unit 16D to change the representative quantization parameter characteristic coefficient (q ₁ to q ₁₎ from the representative quantization parameter characteristic coefficient storage unit 15C according to the characteristic coefficient specifying parameter 14C of the storage unit 14. q ₃ ) is read out, and a representative quantization parameter estimation model is derived from these representative quantization parameter characteristic coefficients (step 105). The representative quantization parameter estimation unit 16D estimates the representative quantization parameter 24 corresponding to the coding rate 14B of the storage unit 14 from the representative quantization parameter estimation model (step 106).

FIG. 7 is a graph showing the relationship between the representative quantization parameter and the coding rate. Normally, the representative quantization parameter 24 has a characteristic that increases as the encoding rate increases. The representative quantization parameter estimation model can be approximated by, for example, an exponential function using this characteristic. When the coding rate is Br, the representative quantization parameter QP _rep is expressed by the following functional equation (3). The representative quantization parameter characteristic coefficients (q _{1 to} q ₃ ) are constants.

Next, the arithmetic processing unit 16 calculates a difference from the difference between the representative quantization parameter 24 estimated by the representative quantization parameter estimation unit 16D and the individual quantization parameter 14A of the storage unit 14 by the differential quantization parameter calculation unit 16E. The quantization parameter 25 is calculated (step 107). The representative quantization parameter is QP _rep , the individual quantization parameter is QP _in , the frame number is f, the number of frames in the quality estimation section is F, the block number is m, and the number of blocks in the frame is M. In this case, the difference quantization parameter ΔQP is calculated based on the following equation (4).

  Next, the arithmetic processing unit 16 reads out the video difference quality characteristic coefficient (a, b) from the video difference quality characteristic coefficient storage unit 15D according to the characteristic coefficient specifying parameter 14C of the storage unit 14 by the video difference quality calculation unit 16F. The video difference quality estimation model is derived from these video difference quality characteristic coefficients (step 108). From this video difference quality estimation model, the video difference quality calculation unit 16F uses the video maximum difference quality 23 estimated by the video maximum difference quality estimation unit 16C and the difference quantization parameter 25 calculated by the difference quantization parameter calculation unit 16E. The video difference quality 26 corresponding to is estimated (step 109).

FIG. 8 is a graph showing the relationship between the video difference quality and ΔTS _I · ΔIc _{Max ±} . Normally, the video difference quality 26 has a characteristic that the product of the video maximum difference quality 23 and the difference quantization parameter 25 is highly correlated. The video difference quality estimation model can be approximated by, for example, a linear function using this characteristic. When the difference quantization parameter is ΔQP, the video difference quality ΔIc _± is expressed by the following function equation (5). The video difference quality characteristic coefficient (a, b) is a constant. However, ΔIc ₊ and ΔIc ₋ are switched according to ΔIc _{Max +} and ΔIc _Max− .

Next, the arithmetic processing unit 16 calculates the sum of the video representative quality 22 estimated by the video representative quality estimation unit 16B and the video differential quality 26 estimated by the video difference quality calculation unit 16F by the video individual quality calculation unit 16G. The individual video quality 27 is calculated (step 110). When the video representative quality is Vq _rep , the video differential quality is ΔIc _± , and the video representative quality characteristic value (a value obtained by subtracting 1 from the video representative quality) is Ic _rep , the video individual quality Vq _in is expressed by the following formula (6 ).

  The individual video quality calculation unit 16G stores the calculated individual video quality 27 in the storage unit 14, displays the screen on the screen display unit 12, or transmits the communication I / F unit 13 to an external device (not shown). Thereafter, the series of video individual quality estimation processing is terminated.

  FIG. 9 is a graph showing an estimation result obtained by a conventional video quality estimation method. FIG. 10 is a graph showing an estimation result by the video quality estimation method according to the present embodiment. In any of these graphs, the subjective evaluation value Vqs (vertical axis) obtained by actual opinion evaluation and the video for the video communication of the video in which the amount of motion of the video and the definition in units of frames are different from the average video. The estimated evaluation value Vqe (horizontal axis), which is the individual video quality 27 estimated by the quality estimation method, is shown.

In these graphs, the closer the data related to each video communication is to a straight line with an inclination of 45 degrees passing through the origin, the smaller the difference between the subjective evaluation value and the estimated evaluation value, that is, the estimation error, and the higher the accuracy.
In FIG. 9, there is a region where subjective evaluation values vary greatly with respect to almost the same estimated evaluation value, whereas in FIG. 10, each data is distributed along a straight line with a 45-degree inclination passing through the origin. Therefore, the video quality estimation method according to the present embodiment reduces the video quality estimation error even for video communications in which the amount of motion of the video and the definition in units of frames differ from the average video. You can see that

[Effects of the present embodiment]
Thus, in the present embodiment, the video representative quality estimation unit 16B uses the video representative quality estimation model that estimates the video representative quality obtained by averaging the video quality of a plurality of videos, and the quality parameter of the video communication. The video differential quality estimation model 17 estimates the video differential quality corresponding to the difference between the video representative quality and the video individual quality of the video by the video differential quality estimation unit 17. Since the video differential quality corresponding to the individual quantization parameter and the quality parameter is estimated, the video individual quality calculation unit 16G calculates the video individual quality of the video communication from the video representative quality and the video differential quality. The video quality can be estimated in consideration of the individual quantization parameter of the target video communication.

In particular, in a compression encoding method such as MPEG-2, in the case of a video with a large amount of motion, the feature is to reduce the code amount of the I frame and increase the code amount for the B frame and P frame of the inter-frame prediction method, In the case of video with high definition, there is a feature that the code amount of the B frame or P frame is reduced and the code amount is increased for the I frame of the intra-frame prediction method.
In the present embodiment, paying attention to the above feature that the individual quantization parameter changes according to the amount of motion of the video and the definition of each frame, the video quality is considered in consideration of the individual quantization parameter of the target video communication. Estimated.

For this reason, video quality with less error is estimated based on the parameters that can be extracted from the IP packet, even for video communications in which the amount of motion of the video and the definition in units of frames are different from the average video. be able to.
Therefore, it is possible to easily determine whether or not video quality of a certain level or more is maintained for a user who uses a video communication service such as an IPTV service, a video distribution service, and a video communication service performed via a network. As a result, it is possible to design and control the quality parameters used in the service and the actual quality of the service being provided.

  In the present embodiment, the difference quantization parameter calculation unit 16E that calculates the difference quantization parameter from the difference between the individual quantization parameter and the representative quantization parameter indicating the average value of the quantization parameters obtained from a plurality of videos. And a video maximum difference quality estimation unit that estimates a video maximum difference quality corresponding to a quality parameter from a video maximum difference quality estimation model that estimates a video maximum difference quality indicating a maximum difference between a video representative quality and a video individual quality. 16C, and the video difference quality calculation unit 16F calculates the individual video quality corresponding to the video difference quality and the difference quantization parameter of the video communication from the video difference quality estimation model. As a result, the video maximum difference quality can be estimated with high accuracy in consideration of the quality parameter of the target video communication, and as a result, it is possible to accurately estimate the video differential quality considering the individual quantization parameter. .

  In the present embodiment, the representative quantization parameter estimation unit 16D estimates the representative quantization parameter corresponding to the quality parameter from the representative quantization parameter estimation model indicating the relationship between the representative quantization parameter and the quality parameter. Yes. As a result, the representative quantization parameter can be estimated with high accuracy in consideration of the quality parameter of the target video communication, and as a result, it is possible to accurately estimate the video differential quality considering the individual quantization parameter. .

[Extended embodiment]
In the above description, when the video difference quality 26 is estimated from the video difference quality estimation model, the estimated and calculated video maximum difference quality 23 and the representative are calculated based on the encoding rate of the target video communication and the individual quantization parameter. Although the case where the video difference quality 26 is estimated using the intermediate parameter including the quantization parameter 24 and the difference quantization parameter 25 has been described as an example, the present invention is not limited to this. For example, the video difference quality 26 may be estimated through another intermediate parameter based on the encoding rate of the target video communication and the individual quantization parameter. Alternatively, a model for directly estimating the video difference quality 26 may be created based on the encoding rate of the target video communication and the individual quantization parameter.

  For the video representative quality estimation model, video maximum difference quality estimation model, representative quantization parameter estimation model, and video differential quality estimation model, the measurement data obtained by measuring individual characteristics from a plurality of video communications in advance. For example, each model may be created by performing a convergence operation by the least square method. In addition, although these models have been described by way of example using functional expressions, they may be realized using a black box model in which only input / output characteristics are specified, such as a neural network or a case base. . Further, when the target video is limited to a specific video, or when a certain amount of estimation error is allowed, for example, a fixed value is set as the video representative quality 22, the video maximum difference quality 23, or the representative quantization parameter 24. It may be used.

  In the above description, the case where the encoding rate 14B is used as the quality parameter of the target video communication has been described as an example. However, the present invention is not limited to this. For example, other quality parameters such as a frame rate may be used. Good. At this time, as described above, a model using the quality parameter may be created in advance. Further, the number of quality parameters is not one, but a plurality of quality parameters may be used at the same time, or estimation may be performed using an arbitrary quality parameter for each intermediate parameter.

It is a block diagram which shows the structure of the video quality estimation apparatus concerning one embodiment of this invention. It is explanatory drawing which shows the operation | movement flow of the video quality estimation apparatus concerning one embodiment of this invention. It is explanatory drawing which shows the relationship between a characteristic coefficient specific parameter and various characteristic coefficients. It is a flowchart which shows the video separate quality estimation process of the video quality estimation apparatus concerning one embodiment of this invention. It is a graph which shows the relationship between video representative quality and a coding rate. It is a graph which shows the relationship between video maximum difference quality and the logarithm value of an encoding rate. It is a graph which shows the relationship between a representative quantization parameter and a coding rate. Is a graph showing the relationship of the video difference quality and ΔTS _I · ΔIc _{Max ±.} It is a graph which shows the estimation result by the conventional video quality estimation method. It is a graph which shows the estimation result by the video quality estimation method concerning this Embodiment. ITU-T Recommendation G. It is explanatory drawing which shows the quality estimation precision of the quality estimation model used by 1070.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Video quality estimation apparatus, 11 ... Operation input part, 12 ... Screen display part, 13 ... Communication I / F part, 14 ... Memory | storage part, 14A ... Individual quantization parameter, 14B ... Coding rate, 14C ... Characteristic coefficient specification Parameter, 14P ... Program, 15 ... Characteristic coefficient storage unit, 15A ... Video representative quality characteristic coefficient storage unit, 15B ... Video maximum difference quality characteristic coefficient storage unit, 15C ... Representative quantization parameter characteristic coefficient storage unit, 15D ... Video differential quality Characteristic coefficient storage unit, 16 ... arithmetic processing unit, 16A ... packet analysis unit, 16B ... video representative quality estimation unit, 16C ... video maximum difference quality estimation unit, 16D ... representative quantization parameter estimation unit, 16E ... differential quantization parameter calculation 16F ... Video difference quality calculation unit 16G ... Video individual quality calculation unit 17 ... Video difference quality estimation unit 2A, 2B ... Communication terminal 3 ... IP network Over click.

Claims (8)

A video quality estimation device for estimating video quality of video communication in which a video signal is compressed and encoded into a plurality of frames including an I frame and transmitted in packets,
A packet analysis unit that analyzes the input packet and extracts an individual quantization parameter related to the quantization step width of the video and a quality parameter related to the video quality of the video communication;
A video representative quality estimation unit for estimating a video representative quality corresponding to a quality parameter of the video communication from a video representative quality estimation model for estimating a video representative quality obtained by averaging video quality of a plurality of videos;
Estimate the video differential quality corresponding to the individual quantization parameter and the quality parameter of the video communication from the video differential quality estimation model that estimates the video differential quality corresponding to the difference between the video representative quality and the video individual quality of the video A video difference quality estimation unit,
A video quality estimation apparatus comprising: a video individual quality calculation unit that calculates a video individual quality of the video communication from the video representative quality and the video differential quality. The video quality estimation apparatus according to claim 1,
The video difference quality estimation unit
A differential quantization parameter calculation unit that calculates a differential quantization parameter from a difference between the individual quantization parameter and a representative quantization parameter related to a quantization step width of the video obtained from a plurality of videos;
Video maximum difference quality estimation for estimating a video maximum difference quality corresponding to the quality parameter from a video maximum difference quality estimation model for estimating a video maximum difference quality indicating a maximum value of a difference between the video representative quality and the video individual quality And
The video quality estimation apparatus comprising: the video differential quality calculation unit that calculates the video individual quality corresponding to the video differential quality and the differential quantization parameter of the video communication from the video differential quality estimation model. The video quality estimation apparatus according to claim 2, wherein
The video difference quality estimation unit further includes a representative quantization parameter estimation unit that estimates a representative quantization parameter corresponding to the quality parameter from a representative quantization parameter estimation model indicating a relationship between the representative quantization parameter and the quality parameter. A video quality estimation apparatus characterized by the above. The video quality estimation apparatus according to claim 3,
The video quality estimation apparatus, wherein the representative quantization parameter estimation model has a characteristic that the representative quantization parameter decreases as the quality parameter including the coding rate increases. The video quality estimation apparatus according to claim 2, wherein
In the video maximum difference quality estimation model, the video maximum difference quality increases (or decreases) as the quality parameter including the encoding rate increases, and the video maximum difference quality decreases (or decreases) as the quality parameter further increases. A video quality estimation apparatus characterized by comprising characteristics that increase). The video quality estimation apparatus according to claim 2, wherein
The video difference quality estimation model is characterized in that the video difference quality increases as the product of the difference quantization parameter and the video maximum difference quality increases. A video quality estimation method for estimating video quality of video communication in which video signals are compressed and encoded into a plurality of frames including I frames and transmitted in packets,
A packet analysis step of analyzing the input packet and extracting an individual quantization parameter related to a quantization step width of the video and a quality parameter related to the video quality of the video communication;
A video representative quality estimation step for estimating a video representative quality corresponding to a quality parameter of the video communication from a video representative quality estimation model for estimating a video representative quality obtained by averaging video quality of a plurality of videos;
Estimate the video differential quality corresponding to the individual quantization parameter and the quality parameter of the video communication from the video differential quality estimation model that estimates the video differential quality corresponding to the difference between the video representative quality and the video individual quality of the video Video difference quality estimation step to perform,
A video quality estimation method comprising: a video individual quality calculating step of calculating a video individual quality of the video communication from the video representative quality and the video differential quality. A computer of a video quality estimation apparatus that estimates video quality of video communication that compresses and encodes a video signal into a plurality of frames including an I frame and transmits the packet in a packet,
A packet analysis step of analyzing the input packet and extracting an individual quantization parameter related to a quantization step width of the video and a quality parameter related to the video quality of the video communication;
A video representative quality estimation step for estimating a video representative quality corresponding to a quality parameter of the video communication from a video representative quality estimation model for estimating a video representative quality obtained by averaging video quality of a plurality of videos;
Estimate the video differential quality corresponding to the individual quantization parameter and the quality parameter of the video communication from the video differential quality estimation model that estimates the video differential quality corresponding to the difference between the video representative quality and the video individual quality of the video Video difference quality estimation step to perform,
A program for executing a video individual quality calculating step of calculating a video individual quality of the video communication from the video representative quality and the video differential quality.

Images