JP4527699B2 - Image quality objective evaluation apparatus, method and program - Google Patents

Description

  The present invention relates to a video quality objective evaluation apparatus, method, and program for deriving objective quality of a video from measurement of a characteristic amount of the video signal without performing a subjective evaluation quality test in which a human views the video and evaluates the quality. Is.

The inventor has proposed a concept of an objective video quality evaluation method that comprehensively handles arbitrary deterioration in an arbitrary scene of an image (see Non-Patent Document 1).
Furthermore, the inventor limits the scene to be evaluated as a method of objectively measuring the subjective quality of the video when the degradation that occurs in the video is limited to freeze degradation based on the concept of Non-Patent Document 1. If all the freezes that occurred in the scene to be evaluated are equal in length, the effects of all the freezes that occurred in the scene to be evaluated are integrated into a parameter called the equivalent freeze length. A method for objectively measuring subjective quality has been proposed (see Non-Patent Document 2).

Watanabe, Okamoto, Kurita, “Study on Video Quality Objective Evaluation Method for Freeze Degradation”, IEICE Technical Report, Vol. 105, no. 179, p. 37-42, 2005 Watanabe, Okamoto, Kurita, “Proposal of Objective Video Quality Evaluation Method for Freeze Degradation in Video Communication Services”, IEICE Technical Report, Vol. 106, no. 9, p. 43-46, 2006

  However, with the techniques disclosed in Non-Patent Document 1 and Non-Patent Document 2, the subjective quality of video cannot be objectively measured accurately when a freeze of an arbitrary length occurs or the scene changes. There was a problem. This problem was particularly noticeable when measuring quality for a long time.

  The present invention has been made to solve the above-described problem, and an image quality objective evaluation apparatus, method, and program capable of accurately measuring image quality when a freeze length is not constant or a scene changes. The purpose is to provide.

The present invention relates to a video quality objective evaluation device that objectively measures the subjective quality of a deteriorated video deteriorated due to freeze, the freeze length counting means for detecting the freeze length of the deteriorated video, and the freeze length counting means Equivalent freeze length deriving means for deriving an equivalent freeze length in consideration of the subjective freeze effect on the video quality of all freeze lengths and video scenes detected in the above, and the degradation in consideration of the video scene from the equivalent freeze length Objective evaluation value deriving means for deriving objective quality, which is an estimated value of subjective quality of the video, and the equivalent freeze length deriving means recursively integrates the lengths of two consecutive freezes in the degraded video And repeatedly executing all the freeze lengths detected in the degraded video into the equivalent freeze length, and the lengths of the two freezes. When the two freezes are obtained by referring to the equivalent freeze derivation database prepared in advance from the length of the freeze with the newer occurrence time and the scene feature value when the freeze occurs. It is characterized by deriving an equivalent freeze length in consideration of subjective influence on video quality .
Further, in one configuration example of the video quality objective evaluation apparatus of the present invention, the freeze length counting means performs a process of determining whether or not two consecutive video frames of the deteriorated video are the same one by one in time series. The number of times it is determined to be the same continuously is determined as the length of the freeze.
In addition, one configuration example of the video quality objective evaluation device of the present invention further includes a scene feature quantity deriving unit that derives a scene feature quantity quantitatively representing the scene feature of the degraded video, and the objective evaluation value deriving unit Is to derive the objective quality of the deteriorated video from the equivalent freeze length based on the correspondence between the preset equivalent freeze length and the objective quality and the scene feature amount.
Also, in one configuration example of the video quality objective evaluation apparatus of the present invention, the scene feature quantity deriving means includes a spatial feature quantity and a temporal feature quantity for each frame or between frames of the degraded video as the scene feature quantity. Is used .

The present invention is also an image quality objective evaluation method for objectively measuring the subjective quality of a deteriorated image deteriorated due to freeze, the freeze length counting step for detecting the freeze length of the deteriorated image, and the freeze length. An equivalent freeze length deriving step for deriving an equivalent freeze length in consideration of the subjective influence on video quality of all freeze lengths and video scenes detected in the counting step, and considering the video scene from the equivalent freeze length An objective evaluation value deriving step for deriving an objective quality that is an estimated value of the subjective quality of the degraded video, and the equivalent freeze length deriving step is a process of integrating the lengths of two consecutive freezes in the degraded video Recursively and repeatedly integrating all freeze lengths detected in the degraded video into the equivalent freeze length Yes, when integrating the lengths of the two freezes, the equivalent freeze derivation database prepared in advance is calculated from the length of the freeze with the newest time and the scene feature when this freeze occurs. With reference to the above, the equivalent freeze length is derived in consideration of the subjective influence of the two freezes on the video quality .
The present invention also provides a video quality objective evaluation program for operating a computer as a video quality objective evaluation apparatus for objectively measuring the subjective quality of a deteriorated video deteriorated by freezing, the freeze length counting step, and the equivalent freeze And causing the computer to execute a length derivation step and the objective evaluation value derivation step, and the equivalent freeze length derivation step recursively and repeatedly executes a process of integrating the lengths of two consecutive freezes in the degraded video. And integrating all the freeze lengths detected in the deteriorated video into the equivalent freeze length, and when integrating the lengths of the two freezes, the generated time Based on the scene feature when this freeze occurs, a database for deriving an equivalent freeze prepared in advance is used. Irradiation to the two freezing is characterized in that to derive the equivalent freeze length considering subjective impact on video quality.

  According to the present invention, the length of the freeze of the deteriorated video is detected, and the equivalent freeze length in consideration of the subjective influence on the video quality of all the detected freeze lengths and the video scene is derived, from the equivalent freeze length, By deriving objective quality of degraded video in consideration of video scenes, for example, when any video is transmitted over a network, packet loss or the like occurs, and freeze degradation or frames that have a non-constant length in the video It is possible to accurately measure objective quality when time system degradation such as rate fluctuation occurs. In addition, by replacing the conventional subjective evaluation method with the video quality objective evaluation apparatus of the present invention, it is possible to reduce a great amount of labor and time necessary for performing the subjective evaluation.

  Further, in the present invention, the scene feature amount deriving unit derives a scene feature amount that quantitatively represents the feature of the scene of the deteriorated video, and the objective evaluation value deriving unit calculates the correspondence between the preset equivalent freeze length and the objective quality. Then, by deriving the objective quality of the degraded video from the equivalent freeze length based on the scene feature amount, the scene dependence of the correspondence between the equivalent freeze length and the objective quality can be corrected.

[First Embodiment]
In the prior art, the equivalent freeze length was calculated in a certain evaluation target scene time unit. In the present invention, each time a new freeze occurs, a new freeze length and a new freeze occur. The feature is that the equivalent freeze length is sequentially calculated in consideration of the scene characteristics.

  In the present invention, every time a new freeze occurs, the equivalent freeze length is calculated in consideration of the new freeze length and the scene characteristics, so that it corresponds to the freeze deterioration of any occurrence pattern in any video scene. Equivalent freeze length can be derived. Thus, in the present invention, for example, when evaluating a long-time content, the video quality can be appropriately evaluated without depending on the length of the content or the number of occurrences of deterioration. Also, according to the present invention, subjective quality can be objectively derived with high accuracy even when the influence of freezing changes drastically, such as when the movement in the scene changes significantly during the process. Further, since the change in the frame rate is also a combination of freezes, it is possible to accurately derive the objective video quality when the frame rate changes.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a video quality objective evaluation apparatus according to an embodiment of the present invention.
The video quality objective evaluation apparatus includes a freeze length counting unit 11 that detects a freeze length FL _k of a deteriorated video, and a scene feature value deriving unit 12 that derives a scene feature SV that quantitatively represents the scene characteristics of the deteriorated video. Based on the scene feature value SV, an equivalent freeze length deriving unit 14 and an objective evaluation value deriving unit 15 to be described later derive a coefficient for correcting the scene dependency, a scene correction coefficient deriving unit 13, and a freeze length counting unit 11 Equivalent Freeze Length Deriving Unit 14 for Deriving Equivalent Freeze Length EFL Considering the Freeze Length and Video Subjective Effects of Video Scenes Detected in Video, and Deriving Objective Quality of Degraded Video from Equivalent Freeze Length EFL Objective evaluation value deriving unit 15.

In the video quality objective evaluation apparatus, first, the deteriorated video signal IS is input to the freeze length counting unit 11 and the scene feature amount deriving unit 12.
The freeze length count unit 11 derives a freeze length FL _k (1 ≦ k ≦ n, where n is the number of occurrences of freeze in the degraded video signal IS) in the degraded video signal IS, and outputs it to the equivalent freeze length derivation unit 14. .

The scene feature quantity deriving unit 12 extracts a scene feature quantity SV representing scene dependency from one or both of the degraded video signal IS and the reference video signal RS and outputs the scene feature quantity SV to the scene correction coefficient deriving unit 13. The reference video signal RS may be the video signal before degradation or a value calculated using the video signal before degradation.
The scene correction coefficient deriving unit 13 derives the scene feature amount SCC1 used as a coefficient of the deriving formula for the equivalent freeze length EFL by the equivalent freeze length deriving unit 14 and outputs the scene feature amount SCC1 to the equivalent freeze length deriving unit 14 The scene feature value SCC2 used as the coefficient of the derivation formula of the objective evaluation value EV by the objective evaluation value derivation unit 15 is derived from the scene feature value SV and output to the objective evaluation value derivation unit 15.

The equivalent freeze length derivation unit 14 derives the equivalent freeze length EFL from the input freeze length FL _k (1 ≦ k ≦ n) and the scene feature quantity SSC1, and outputs the equivalent freeze length EFL to the objective evaluation value derivation unit 15.
The objective evaluation value deriving unit 15 derives an objective evaluation value EV from the input equivalent freeze length EFL and scene feature amount SSC2. It should be noted that each unit has a necessary memory for the above operation.

Details of each part will be described below. The freeze length count unit 11 has a function of deriving the number of freezes generated in the deteriorated video signal IS and the length of each freeze. In general, in the unit evaluation section of the deteriorated video signal IS, as shown in FIG. 2, a random freeze occurs n times (n is a natural number). The length of each freeze is FL _k (1 ≦ k ≦ n). The unit evaluation section may be a frame unit or a time unit, and can be determined freely.

The freeze length counting unit 11 executes a process of determining whether or not the two frames are the same based on the difference value between the pixels of two consecutive frames in the degraded video signal IS while shifting the frames one frame at a time. The number of times that the values are determined to be the same continuously is set as a freeze length FL _k . However, when the reference video signal RS is present as an input, it is determined that two consecutive frames of the reference video signal RS are the same, and two consecutive frames corresponding to the RS of the degraded video signal IS are the same. In this case, it is determined that the reference video signal RS is originally a still image, and it is determined that the degraded video signal IS is not frozen.

The freeze length FL _k is determined when it is determined that the two consecutive frames are different when the above-described determination process is executed while being shifted frame by frame. Then, the freeze length counting unit 11 continues the above-described determination processing even after the freeze length FL _k is determined, and starts derivation of the freeze length FL _{k + 1} from the point where the next two consecutive frames are determined to be the same. The freeze length counting unit 11 derives the freeze length FL _k (1 ≦ k ≦ n) as described above for each unit evaluation section of the deteriorated video signal IS. The freeze length counting unit 11 may sequentially output each freeze length FL _k thus obtained to the equivalent freeze length deriving unit 14, or store it in its own memory to some extent and then collectively to the equivalent freeze length deriving unit 14. It may be output.

  Note that the freeze length counting unit 11 determines whether the two consecutive frames are the same by the following equation (1) based on the number of pixels changed between the two consecutive frames and the amount of change of the pixels. This is done using equation (2). When only the degraded video signal IS is input, the freeze length counting unit 11 makes the two consecutive frames identical only when the two consecutive frames of the degraded video signal IS satisfy both the conditions of the expressions (1) and (2). Judge. In addition, when both the degraded video signal IS and the reference video signal RS are input, the freeze length counting unit 11 satisfies both the conditions of the formulas (1) and (2) for the two consecutive frames of the degraded video signal IS. When two consecutive frames of the reference video signal RS corresponding to these frames do not satisfy at least one of the expressions (1) and (2), the two consecutive frames are determined to be the same.

[Number of pixels changed between two consecutive frames of the degraded video signal IS / total number of pixels in the video frame] <α (1)
[Sum of change amount of pixels changed between two consecutive frames] <β (2)
The constants α and β are constants determined in advance by a subjective evaluation test.

  Next, the scene feature quantity deriving unit 12 determines the scene feature quantity SV describing the scene of the video such as the complexity of the video and the amount of movement of the degraded video signal IS for each unit evaluation section, and the degraded video signal IS and the reference video signal RS. Are derived from either or both signals. As the scene feature value SV, it is conceivable to use a spatial feature value or a temporal feature value. For example, “ITU-T Recommendation P.910,“ Subjective video quality ”of International Telecommunication Union (ITU). Assessment methods for multimedia applications ”, Aug. 1996” and “ANSI T1.801.03-1996,“ Digital Transport of One-Way Video signals-Parameters for Objective Performance Assessment ”defined by ANSI (American National Standards Institute) "Spatial Information (SI)" and "Temporal Infromation (TI)" described in "."

  The scene feature amount deriving unit 12 may use the SI or TI derived for each frame as the scene feature amount SV as it is, or divide one frame into a plurality of blocks and use the SI and TI derived for each block. A value obtained by performing a certain operation on the image, for example, a maximum value, a minimum value, an average value, or the like may be used as the scene feature value SV. In addition, the frame is divided into vertical M × horizontal N pixel blocks, the search window is set to vertical m × horizontal n, a motion vector search is performed for each block, and a certain calculation is performed on the motion amount of each block. A numerical value, for example, the maximum value, the minimum value, the average value of the motion vectors of all the blocks in the frame, or the rate of motion that is greater than or equal to a certain value can be used as the scene feature value SV. Further, the scene feature quantity SV can be calculated from one type or a plurality of feature quantities derived from a specific frame or a plurality of frames. In the present embodiment, these feature quantities are output to the scene correction coefficient deriving unit 13 as scene feature quantities SV.

The scene correction coefficient derivation unit 13 calculates the scene correction coefficients SCC1 and SCC2 based on the scene feature quantity SV of the unit evaluation section unit input from the scene feature quantity derivation unit 12, and the scene correction coefficients SCC1 and SCC2 are equivalent to each other. The result is passed to the freeze length deriving unit 14 and the objective evaluation value deriving unit 15.
Specific scene correction coefficient SCC1 is less _{a, b, p i, q} i, r i to derive for each unit evaluation section, the scene correction coefficient SCC2 the following a deriving for each unit evaluation section, b.

a = α ₁ [SI] + β ₁ [TI] + γ ₁ [MV] (3)
b = α ₂ [SI] + β ₂ [TI] + γ ₂ [MV] (4)
p _i = α _3i [SI] + β _3i [TI] + γ _3i [MV] (2 ≦ i ≦ ε) (5)
q _i = α _4i [SI] + β _4i [TI] + γ _4i [MV] (2 ≦ i ≦ ε) (6)
r _i = α _5i [SI] + β _5i [TI] + γ _5i [MV] (2 ≦ i ≦ ε) (7)

However, α ₁ , β ₁ , γ ₁ , α ₂ , β ₂ , and γ ₂ are a plurality of scene images having different characteristics so that the scene dependency can be accurately taken into account in the calculation formula used in the objective evaluation value deriving unit 15. This coefficient is derived in advance from the subjective evaluation characteristics when deterioration is added to. Similarly, α _3i , β _3i , γ _3i , α _4i , β _4i , γ _4i , α _5i , β _5i , and γ _5i can be accurately considered in the scene dependence by a calculation formula used in the equivalent freeze length deriving unit 14. As described above, the coefficient is derived in advance from the subjective evaluation characteristics when deterioration is added to a plurality of scene videos having different characteristics. ε is a value determined in advance by a subjective evaluation test, and can take all integer values of 2 or more. As will be described later, ε is a threshold value for selecting a method for deriving the equivalent freeze length.

  [MV] is a ratio of blocks having a motion vector of a certain size or more in the unit evaluation section. [SI] and [TI] are average values of SI and TI obtained for each frame in the unit evaluation section.

Note that the scene correction coefficient deriving unit 13, when receiving a correction coefficient request accompanying the specification of the frame start number and unit evaluation time information from the objective evaluation value deriving unit 15, performs scene correction for the section specified in accordance with the correction coefficient request. The coefficients SCC1 (a, b, p _i , q _i , r _i ) and SCC 2 (a, b) may be derived and passed to the equivalent freeze length deriving unit 14 and the objective evaluation value deriving unit 15.

Next, the equivalent freeze length deriving unit 14 includes the freeze length FL _k (1 ≦ k ≦ n) input from the freeze length counting unit 11 and the SCC1 (a, b, p _i ) input from the scene correction coefficient deriving unit 13. , Q _i , r _i ), an equivalent freeze length EFL is derived in consideration of the subjective influence of all freezes on video quality. The method for deriving the equivalent freeze length EFL is characterized by repeating the integration of two freeze lengths into one equivalent freeze length, as shown in FIG.

A method for integrating equivalent freeze lengths will be specifically described with reference to FIG. First, the equivalent freeze length deriving unit 14 substitutes the length FL ₁ of freeze ₁ and the length FL _{2 of} freeze 2 into a predetermined integration function f in order to integrate freeze 1 and freeze 2. The freeze length EFL ₂ is derived and stored. Here, the integration function f is a function that converts the subjective influence of the two freezes on the video quality into an equivalent effect (freeze length) when the freeze occurs once. Subsequently, the equivalent freeze length deriving unit 14 substitutes the length FL ₃ of the freeze ₃ and the equivalent freeze length EFL ₂ obtained by integrating the freeze 1 and the freeze 2 into the integrated function f to derive the equivalent freeze length EFL _3. Remember.

Equivalent freeze length deriving section 14 recursively repeated until integrating length FL _n of the last freeze n generated the aforementioned processing in the unit evaluation interval of the deteriorated video signal IS, the equivalent freeze length EFL _n To derive. This EFL _n is an equivalent freeze length EFL obtained by integrating the lengths of all freezes generated during the unit evaluation section of the deteriorated video signal IS. The equivalent freeze length deriving unit 14 outputs the equivalent freeze length EFL to the objective evaluation value deriving unit 15.

Incidentally, FL ₂ when the equivalent freeze length deriving section 14, which integrates two freeze length times caused among the newer freeze length FL _{new new} (e.g., a freeze length FL ₁ and freeze length FL ₂ to integrate, the equivalent When FL ₃ ) when integrating the freeze length EFL ₂ and the freeze length FL ₃ is larger than the ε frame, a function that simply takes the sum of two freeze lengths is used as the integration function f.

The equivalent freeze length deriving unit 14 uses an equivalent freeze derivation database prepared in advance when the freeze time FL _new, which has a newer time out of the two freeze lengths to be integrated, is equal to or less than ε frames. The integrated function f is determined, and an equivalent freeze length EFL is derived. An example of this equivalent freeze derivation database is shown in FIG. The equivalent freeze derivation database corresponds to the freeze length FL _new from 2 frames of the minimum freeze length to ε frame and the coefficients p _i , q _i , r _i (2 ≦ i ≦ ε) of the following equation (8) It is what you remember.
EFL = p _i (FL _sum ) + q _i log (FL _sum ) + r _i (8)

Expression (8) represents the correspondence between the total FL _{sum of the} freeze lengths generated in the video and the equivalent freeze length EFL when the freeze length per time is fixed to i (2 ≦ i ≦ ε) frames. It is. The coefficients p _i , q _i , and r _i are determined in advance by a subjective evaluation test. However, 2 ≦ i ≦ epsilon coefficient p _i in subjective tests in the entire range of, q _i, there is no need to predetermine the r _i, 2 ≦ _i ≦ coefficient subjective evaluation tests at two points or more of the epsilon p _i , Q _i , r _i can be obtained, the coefficients of the remaining points can be derived by interpolation.
Further, since the coefficients p _i , q _i , and r _i vary depending on the scene change as shown in the explanation of the operation of the scene correction coefficient deriving unit 13, the coefficients of the scene correction coefficient SCC1 output from the scene correction coefficient deriving unit 13 are changed. According to the value, the equivalent freeze derivation database is updated as appropriate. As long as the update cycle is a natural number in frame units or time units, any value can be used as the update cycle.

FIG. 5 illustrates the characteristic of the equation (8). The vertical axis in FIG. 5 was obtained in a subjective evaluation test in which the freeze length per time was fixed to an i (2 ≦ i ≦ ε) frame, and freezes of this fixed length were generated multiple times. This is an equivalent freeze length EFL obtained by converting the subjective evaluation value SEV as an influence of one freeze length. The equivalent freeze length EFL can be obtained by substituting the subjective evaluation value SEV into the following equation (9).
EFL = power (10, (SEV−b) / a) (9)
Where power (s, t) = s ^t

Expression (9) shows the relationship between the length (equivalent freeze length EFL) when the freeze obtained from the subjective evaluation test occurs once and the subjective evaluation value SEV. The relationship of this equation (9) is shown in FIG. If it is possible to obtain the equivalent freeze length EFL from the subjective evaluation value SEV using Equation (9), by performing the sum of the freezing length FL _sum equivalent freeze length EFL regression analysis using the equation (8), coefficient p It is possible to derive _i , q _i and r _i .

When integrating the already accumulated equivalent freeze length EFL _old (or the first freeze length FL ₁ ) and the freeze length FL _new , the freeze time FL _new that has the newest time out of these freeze lengths is an ε frame. A method of integrating the equivalent freeze length EFL _old (or the freeze length FL ₁ ) and the freeze length FL _new using the equivalent freeze derivation database will be described with reference to FIG. The coefficients p _i , q _i , r _i in the case of i = FL _new (2 ≦ FL _new ≦ ε) are obtained from the equivalent freeze derivation database, and the freeze length per time is FL _new frame from the equation (8). FIG. 7 shows the relationship between the freeze length total FL _sum and the equivalent freeze length EFL when this occurs multiple times.

First, the equivalent freeze length deriving section 14, coefficient p _i corresponding to freeze length FL _{new new,} q _i, obtains the r _i from the database for the equivalent freeze derived, obtained coefficients p _i, q _i, r _i and equation ( 8) is used to calculate the equivalent freeze length EFL _old (or the first freeze length FL ₁ ) that has been accumulated so far when the freeze length per occurrence is multiple times in the FL _new frame. _Is converted into a total _{sum FLsum} (step S1 in FIG. 7).

Next, the equivalent freeze length deriving unit 14 sets a value obtained by adding the freeze length FL _new to the freeze length total FL _sum obtained in step S1 as a new freeze length total FL _sum (step S2 in FIG. 7). Finally, the equivalent freeze length deriving section 14, the coefficient obtained from the database for the equivalent freeze deriving p _i, q _i, the sum FL _sum freeze length determined by r _i and step S2 are substituted into Equation (8), An equivalent freeze length EFL is derived (step S3 in FIG. 7). In FIG. 7, the equivalent freeze length obtained in step S3 is described as EFL _new . Thus, the two freeze lengths EFL _old (or FL ₁ ) and FL _new can be integrated.

Next, the objective evaluation value deriving unit 15 is based on the equivalent freeze length EFL input from the equivalent freeze length deriving unit 14 and the scene correction coefficient SCC2 (a, b) input from the scene correction coefficient deriving unit 13. The objective evaluation value EV is derived using the following formula (10) representing the relationship between the a priori obtained equivalent freeze length EFL and the objective evaluation value EV.
EV = alog (EFL) + b (10)

  The processing content of the objective evaluation value deriving unit 15 is shown in FIG. The function of Expression (10) is a function for converting the equivalent freeze length EFL (the length when the freeze occurs once) into the objective evaluation value EV. In addition, although Formula (10) is represented in the format using a logarithm, it is not necessary to be a logarithm in particular, and may be obtained in advance from a subjective evaluation test as a subjective evaluation characteristic for freeze.

  As described above, according to the present embodiment, when an arbitrary video is transmitted through a network, packet loss or the like occurs, and time-based degradation such as freeze degradation or frame rate fluctuation occurs in the video. By substituting the conventional subjective evaluation method with the video quality objective evaluation device of this embodiment, it is possible to estimate the subjective evaluation value at the time, and to reduce a great amount of labor and time necessary for subjective evaluation Can do.

  Note that the video quality objective evaluation device of the present embodiment can be realized by a computer having a CPU, a storage device, and an external interface, and a program for controlling these hardware resources. In such a computer, the video quality objective evaluation program for realizing the video quality objective evaluation method of the present invention is provided in a state of being recorded on a recording medium such as a flexible disk, a CD-ROM, a DVD-ROM, or a memory card. Is done. The CPU writes the program read from the recording medium into the storage device, and executes the above-described processing according to the program.

  The present invention can be applied to a technique for measuring the objective quality of video from the measurement of physical feature quantities of the video signal.

It is a block diagram which shows the structure of the video quality objective evaluation apparatus which concerns on embodiment of this invention. It is a figure which shows the example which freezed n times in the degradation video signal. It is a figure which shows the derivation method of the equivalent freeze length in the equivalent freeze length derivation | leading-out part of the video quality objective evaluation apparatus of FIG. It is a figure which shows the example of the database for equivalent freeze derivation in embodiment of this invention. It is a figure which shows the relationship between the sum total of freeze length, and an equivalent freeze length at the time of fixing the freeze length per time in embodiment of this invention. It is a figure which shows the relationship between the length when a freeze generate | occur | produces once in embodiment of this invention, and a subjective evaluation value. It is a figure which shows the method of integrating | accumulating the equivalent freeze length accumulate | stored in embodiment of this invention, and the newly generated freeze length, and deriving a new equivalent freeze length. It is a figure which shows the relationship between an equivalent freeze length and objective evaluation value in embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 11 ... Freeze length count part, 12 ... Scene feature-value derivation part, 13 ... Scene correction coefficient derivation part, 14 ... Equivalent freeze length derivation part, 15 ... Objective evaluation value derivation part

Claims (6)

A video quality objective evaluation device that objectively measures the subjective quality of degraded video that has deteriorated due to freeze,
Freeze length counting means for detecting the length of the freeze of the deteriorated image;
Equivalent freeze length deriving means for deriving equivalent freeze length in consideration of the subjective influence of all freeze lengths detected by the freeze length counting means and the video scene on the video quality;
An objective evaluation value deriving unit for deriving an objective quality that is an estimated value of the subjective quality of the degraded video in consideration of a video scene from the equivalent freeze length ;
The equivalent freeze length deriving means recursively and repeatedly executes a process of integrating the lengths of two consecutive freezes in the deteriorated image, and all the freeze lengths detected in the deteriorated image are converted into the equivalent freeze length. When integrating the lengths of the two freezes, it is prepared in advance from the length of the freeze with the newest time and the scene feature when the freeze occurred. An image quality objective evaluation apparatus , wherein an equivalent freeze length is derived in consideration of a subjective influence of the two freezes on image quality with reference to an equivalent freeze derivation database . The video quality objective evaluation apparatus according to claim 1,
The freeze length counting means executes a process of determining whether two consecutive video frames of the deteriorated video are the same while shifting the frames one by one in time series, and determines the number of times that the same is determined as the same An image quality objective evaluation device characterized by determining the length of the freeze. The video quality objective evaluation apparatus according to claim 1,
Furthermore, a scene feature quantity deriving unit for deriving a scene feature quantity quantitatively representing the scene feature of the degraded video is provided,
The objective evaluation value deriving means derives the objective quality of the deteriorated video from the equivalent freeze length based on a correspondence relationship between a preset equivalent freeze length and objective quality and the scene feature amount. Video quality objective evaluation device. In the video quality objective evaluation apparatus according to claim 3,
The video feature objective evaluation apparatus, wherein the scene feature quantity deriving means uses a spatial feature quantity and a temporal feature quantity for each frame or between frames of the degraded video as the scene feature quantity. A video quality objective evaluation method that objectively measures the subjective quality of degraded video degraded by freeze,
A freeze length counting step for detecting a freeze length of the deteriorated image;
An equivalent freeze length deriving step for deriving an equivalent freeze length in consideration of the subjective influence of all the freeze lengths detected in this freeze length counting step and the video scene on the video quality;
An objective evaluation value deriving step for deriving an objective quality that is an estimated value of a subjective quality of the degraded video in consideration of a video scene from the equivalent freeze length;
The equivalent freeze length deriving step recursively and repeatedly executes a process of integrating the lengths of two consecutive freezes in the degraded video, and all the freeze lengths detected in the degraded video are converted to the equivalent freeze length. This is a step of integrating the lengths of the two freezes, and is prepared in advance from the length of the freeze with the newest time and the scene feature when the freeze occurred. A video quality objective evaluation method, wherein an equivalent freeze length is derived in consideration of a subjective influence of the two freezes on video quality with reference to an equivalent freeze derivation database . A video quality objective evaluation program that operates a computer as a video quality objective evaluation device that objectively measures the subjective quality of degraded video degraded by freeze,
A freeze length counting step for detecting a freeze length of the deteriorated image;
An equivalent freeze length deriving step for deriving an equivalent freeze length in consideration of the subjective influence of all the freeze lengths detected in this freeze length counting step and the video scene on the video quality;
From the equivalent freeze length, let the computer execute an objective evaluation value derivation step for deriving objective quality that is an estimated value of subjective quality of the degraded video in consideration of the video scene,
The equivalent freeze length deriving step recursively and repeatedly executes a process of integrating the lengths of two consecutive freezes in the degraded video, and all the freeze lengths detected in the degraded video are converted to the equivalent freeze length. This is a step of integrating the lengths of the two freezes, and is prepared in advance from the length of the freeze with the newest time and the scene feature when the freeze occurred. An image quality objective evaluation program, wherein an equivalent freeze length is derived in consideration of a subjective influence of the two freezes on image quality with reference to an equivalent freeze derivation database.

Images