JP2004172753A - Method and apparatus for objectively evaluating video / audio quality - Google Patents

Method and apparatus for objectively evaluating video / audio quality Download PDF

Info

Publication number
JP2004172753A
JP2004172753A JP2002333898A JP2002333898A JP2004172753A JP 2004172753 A JP2004172753 A JP 2004172753A JP 2002333898 A JP2002333898 A JP 2002333898A JP 2002333898 A JP2002333898 A JP 2002333898A JP 2004172753 A JP2004172753 A JP 2004172753A
Authority
JP
Japan
Prior art keywords
quality
evaluation
video
audio
objective
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP2002333898A
Other languages
Japanese (ja)
Inventor
Ginga Kawaguchi
銀河 川口
Atsushi Okamoto
淳 岡本
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nippon Telegraph and Telephone Corp
Original Assignee
Nippon Telegraph and Telephone Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nippon Telegraph and Telephone Corp filed Critical Nippon Telegraph and Telephone Corp
Priority to JP2002333898A priority Critical patent/JP2004172753A/en
Publication of JP2004172753A publication Critical patent/JP2004172753A/en
Pending legal-status Critical Current

Links

Images

Abstract

<P>PROBLEM TO BE SOLVED: To appropriately estimate subjective quality on the basis of the measurement of a physical featured value of a video/audio signal without conducting a subjective evaluation test. <P>SOLUTION: The method and apparatus for objectively evaluating video/audio quality is provided with: a means 110 for obtaining an evaluation value denoting the objective quality of a video/audio signal being an evaluation object received in a streaming form every short time; a means 120 for storing the obtained evaluation value in time series; and a means 130 for obtaining an overall evaluation value (overall quality) reflecting long time subjective evaluation characteristics of human beings on the basis of the stored evaluation value every prescribed time and a predetermined overall quality evaluation value characteristic model. <P>COPYRIGHT: (C)2004,JPO

Description

【０００１】
【発明の属する技術分野】
本発明は、人間が映像又は音声を観てあるいは聴いてその品質を評価する主観評価試験を行うことなく、映像又は音声信号の物理的特徴量の測定から主観品質を推定する映像・音声品質客観評価方法及び装置に関する。
【０００２】
【従来の技術】
近年、インターネットアクセス回線の高速・広帯域化に伴い、映像・音声のストリーミング配信サービスの普及が本格化している。映像・音声の快適なストリーミング配信サービスを提供するためには、適切な映像・音声品質評価技術に基づく品質設計・管理が重要となる。
【０００３】
映像・音声の品質は、多数の被験者が実際に映像・音声を観てあるいは聴いてその品質を評価する主観評価試験により評価することが基本である（例えば、非特許文献１参照）。しかし、主観評価試験を行うには多くの手間・時間・費用がかかる。また、評価に影響を与える外部要因を極力排除する必要があるため、専用の評価施設が必要なことから、実験を行える場所も限られている。
【０００４】
これを解決する手段として客観品質評価装置がある。客観品質評価装置では、評価対象の映像や音声信号を入力とし、客観評価モデルが出力する歪み量を元に、主観評価実験を行う事なく、主観品質を推定する。なお、代表的なビデオストリーミング向け品質評価モデルツールには、例えばＧｅｎｉｓｔａ社のＶｉｄｅｏ ＱｏＳがある（例えば、非特許文献２、非特許文献３参照）。
【０００５】
【非特許文献１】
川口銀河 他，“映像配信サービスにおけるビットレート変動の影響の主観品質評価”，信学技報ＣＱ２００２−７１，Ｊｕｌ．２００２
【非特許文献２】
ＭＣＭＪａｐａｎ，“ビデオストリーミンク向け品質分析ソフトウエア「ＶｉｄｅｏＱｏＳ」”，平成１３年１０月９日、Ｂｒｏａｄｂａｎｄ Ｗａｔｃｈホームページ、＜ＵＲＬ：ｈｔｔｐ：／／ｗｗｗ．ｗａｔｃｈ．ｉｍｐｒｅｓｓ．ｃｏ．ｊｐ／ｂｒｏａｄｂａｎｄ／ｎｅｗｓ／２００１／１０／０９／ｖｉｄｅｏｑｏｓ．ｈｔｍ＞
【非特許文献３】
Ｇｅｎｉｓｔａ Ｃｏｒｐｒａｔｉｏｎ，“Ｖｉｄｅｏ ＱｏＳ知覚ビデオ品質評価”，平成１３年１０月１０日，株式会社ジェニスタ、＜ＵＲＬ：ｈｔｔｐ：／／ｗｗｗ．ｃｉｄｆ．ｏｒｇ／ｊａｐａｎｅｓｓ／ｉｎｆｏｒｍａｔｉｏｎ／ｄｏｃｓ／ｃｉｄｆ−ｇｅｎ−４９．ｐｄｆ＞
【０００６】
【発明が解決しようとする課題】
従来技術では、映像や音声信号の短時間区間（数１０ｍｓｅｃ）毎に得られる特徴物理量を評価時間長（例えば数秒）に渡って平均した値などを客観評価装置の出力（つまり、総合品質評価値）とする方法が用いられる。この方法は、品質（特徴物理量）の変動が少なく、かつ評価時間長が短時間（例えば数秒）である場合には、主観品質との一致が期待できる。
【０００７】
しかし、評価時間長が長くなり、かつ、その中で品質が時間的に変動する場合には、人間の劣化認知特性は短時間区間の品質の単純な時間平均では表現できず、結果として得られる推定値は実際の主観品質と乖離してしまう恐れがある（非特許文献１）。
【０００８】
一般に、映像・音声配信サービスのコンテンツ（番組等）の時間尺は、通常、数分〜数時間程度であり、この間にネットワークの状態の変化などによって品質が変動することがある点を鑑みれば、従来技術をこのような長時間サービスの評価に直接適用することは品質推定精度の低下を招くことになる。
【０００９】
本発明の目的は、上述の問題を解決した、映像・音声信号の客観的な物理的特徴量の測定から主観品質を適切に推定する映像・音声品質主観評価方法及び装置を提供することにある。
【００１０】
【課題を解決するための手段】
本発明は、評価対象の映像・音声信号の短時間（例えば数秒）毎の客観品質を表す評価値を過去の一定時間分（例えば数分から数十分）、順次、時系列的に保持し、この一定時間分の評価値を元に、予め用意した総合品質評価値特性モデルに基づいて、人間の長時間主観評価特性を反映した総合品質を推定することを特徴とする。
【００１１】
本発明では、人間の劣化知覚特性に対応した特徴物理量時系列元に、予め用意した時間的に変動する品質劣化に対応した人間の認知特性を表わすモデルに基づいて総合品質評価値を推定するため、特徴物理量とモデルの妥当性が保持される限り、劣化時系列から総合品質を推定するという手法において、主観品質を適切に推定できる。
【００１２】
【発明の実施の形態】
以下、本発明の一実施形態について図面により説明する。
図１は本発明の映像・音声品質客観評価装置の一実施形態のブロック図である。図１において、映像・音声品質客観評価装置１００は短時間評価特性計算部１１０、一時記憶部１２０、長時間評価特性計算部１３０、及び、これら各部の動作を制御する制御部１４０から構成される。
【００１３】
短時間評価特性計算部１１０は、評価対象の映像・音声信号１０をストリーミング形式に連続入力し、あらかじめ内蔵した客観評価モデルをもとに、評価対象信号の短時間（例えば数秒）毎の客観品質を表す評価値を計算して順次出力する。即ち、短時間評価特性計算部１１０では、従来の客観評価装置と同様に、評価対象の映像・音声信号における短時間区間（数１０ｍｓｅｃ）毎に得られる特徴物理量を所定の評価時間長（例えば数秒）にわたって平均した値等を、評価対象信号の客観品質を表わす評価値（短時間特性）として順次出力する。この短時間評価特性計算部１１０には、例えばＧｅｎｅｓｔａ社のＶｉｄｅｏＱｏｓをそのまま利用することができる（例えば、非特許文献２，３参照）。この場合、評価対象（加工された）の映像・音声信号１０のほかに、別途、その基準となる映像・音声信号も入力する。
【００１４】
一時記憶部１２０は、短時間評価特性計算部１１０から順次出力される評価値を時系列で記憶する。即ち、一時記憶部１２０は、評価対象である映像・音声信号の物理特徴量の短時間特性を時系列的に保持する。この一時記憶部１２０は、例えばシフトレジスタアレイ構成のバッファメモリなどにより実現し一定時間を過ぎたデータ（評価値）は順次廃棄して行く。一時記憶部１２０に過去のどの時点までの評価値を留めておくかは、長時間評価特性計算部１３０に含まれる総合品質評価特性モデル（主観評価特性モデル）の要求仕様で決まる。
【００１５】
長時間評価特性計算部１３０は、短時間評価特性計算部１１０の処理と平行して、所定時間（例えば数秒）毎に、一時記憶部１２０から過去の一定時間（例えば数分から数十分）分の評価値を取り込み、予め内蔵した総合品質評価特性モデルに基づき、人間の長時間主観評価特性を反映した総合評価値（総合品質）を計算し、客観品質評価結果２０として出力する。この長時間評価特性計算部１３０に用意する総合品質評価特性モデルについては後述する。
【００１６】
制御部１４０は、例えばＣＰＵ構成であり、短時間評価特性計算部１１０、一時記憶部１２０及び長時間評価特性計算部１３０の各動作タイミング等を制御する。
【００１７】
図２に本映像・音声品質客観評価装置１００の全体的処理フローを示す。図２中、時間ｔは短時間評価特性計算部１００の要求仕様の評価時間長（例えば数秒）であり、時間τは長時間特性計算部１３０の要求仕様の評価時間長（例えば数分から数十分）である。
【００１８】
映像・音声品質客観評価装置１００は、装置を立ち上げ、動作を開始することで、評価対象の映像・音声信号１０をストリーミング形式で連続的に入力する（Ｓ１）。先に述べたように、必要ならば別途、基準となる映像・音声信号も同じく入力する。この連続的に入力される評価対象の映像・音声信号１０について、短時間評価特性計算部１１０は、内蔵した客観評価モデルに基づき、所定時間（ｔ）毎に、その時間長（ｔ）の評価値（短時間特性）を計算し、順次出力していく。
【００１９】
まず、動作開始から時間（τ）に該当する最初の一定時間では、短時間評価特性計算部１１０から順次出力される評価値を、単に一時記憶部１２０へ時系列的に記憶するだけの処理を実行する。図２のステップＳ２の前処理がこれを示している。
【００２０】
その後は、所定時間（ｔ）毎に（Ｓ３）、短時間評価特性計算部１１０が、評価対象の映像・音声信号１０について、時間長（ｔ）の評価値を計算し、一時記憶部１２０へ時系列的に記憶する処理（Ｓ５，Ｓ６）と平行して、長時間評価特性計算部１３０では、一時記憶部１２０から一定時間（τ）分の評価値を取り込み、内蔵した総合品質評価特性モデルに基づき、人間の長時間主観評価特性を反映した総合評価値を計算し、それを客観品質評価結果２０として出力する（Ｓ７，Ｓ８，Ｓ９）。これを評価対象の映像・音声信号が入力されなくなるまで繰り返す（Ｓ４）。なお、先にも述べたように、一時記憶部１２０のデータは、一定時間（τ）を過ぎたものから順次廃棄していく。
【００２１】
次に、長時間評価特性計算部１３０の総合品質評価特性モデル（主観評価特性モデル）について説明する。総合品質評価特性モデルは、人間の長時間主観評価特性を反映する必要がある。即ち、映像・音声を連続して視聴している利用者が、過去の品質劣化状態の影響なども含めて、現在、視聴メディアに対して感じている主観品質を推定できる必要がある。映像・音声の視聴時に発生した劣化の影響は、忘却効果（劣化があったことを時間と共に忘れて行く効果）、段差関知効果（変化の平均特性以外に、“品質の変化”が起きたことそのものを妨害要因と感じる効果）などが挙げられる（例えば、非特許文献１参照）。
【００２２】
ここでは、これらを考慮し総合品質評価特性モデルとして、
（１） 同一の劣化時間長・劣化度合いの品質劣化であっても、劣化の発生した時間的位置が品質判断時点から離れるほど総合品質に与える影響が低下するモデル
（２） 同一の総劣化時間長／劣化度合いであっても、品質変動の頻度が高いほど総合品質が低下するモデル
（３） 上記（１）に示すモデルにおいて、ある一定の劣化時間長あるいは劣化度合いを超えた場合にはその影響が支配的となるモデル
を想定する。実際は必要に応じ、このうちの任意の一つまたは任意の二つの組み合せ、あるいは三つ全部を使用して総合品質評価特性モデルを設定する。ここでは、一例として（１）と（２）を組み合せたモデルを仮定する。
【００２３】
時刻ｔでの品質（短時間評価特性計算部１１０により測定される客観評価値）をＱ（ｔ）とする。時刻０における、人間の長時間主観評価特性を考慮した本モデルにおける総合品質（長時間評価特性計算部１３０で計算される総合評価値）Ｑ＿ｇ（ｔ）は以下の式で表現される。
【００２４】
【数１】

Figure 2004172753
【００２５】
ここで、右辺の第１項は、視聴直後のものはよく覚えているので影響が大きいが、時と共に忘れていくことを表わしてする。但し、時間τが過ぎたものは寄与が少ないので完全に忘れたものとする。第２項は、ビットレートの変化（品質の変化）が発生すると“変化”の絶対値が劣化として感じられ、それは視聴直後ほど強いことを表わしている。
【００２６】
このモデルの特徴は以下の通りである。
（１） 常時一定の品質Ｑ＿ｃであるものは時間変化を考慮してもＱ＿ｃとなる
（２） 時間τ以内に品質が劣化したことがある場合には現在再生している短時間品質がＱ＿ｃであっても、時間的影響を組み込むとＱ＿ｃよりも低くなる。
（３） 発生した品質低下がどんなに短時間であっても、第二項により、変化が発生したことの影響が出る（単なる平均などではない主観特性の効果）。
【００２７】
このモデルの場合、ある時刻ｔにおける品質は過去τ時間にのみ影響されることになるので、必然的に図１中の一時記憶部１２０に必要な量はτの時間分となる（それ以前の品質は本モデルにおいては影響しないため不要である）。
【００２８】
図３に本モデルの適用例を示す。また、図４に本モデルの重み（評価特性重み）ｗ（ｔ）を示す。
【００２９】
図３において、上が短時間品質（短時間評価特性計算部１１０で測定される客観評価値）Ｑ（ｔ）の時間変化の例を表わし、下がこれに対して、本モデルによる時間変動の影響を含む総合品質（長時間評価特性計算部１３０で得られる総合評価値）Ｑ−ｇ（ｔ）の時間変化の例を表わしたものである。短時間品質Ｑ（ｔ）の時間変化は、
時刻 −ｒ〜 −０．７τ４， −０．３τ〜 ０ ：品質 １００
時刻 −０．７τ４，〜−０．３τ ：品質 ５０
である。現在の表示／再生の短時間品質は１００となっているが、過去の時刻１−０．７τから−０．３τでの品質５０への劣化はユーザが覚えているという効果を本モデルで評価すると、式（１）を積分して現在の時間変化を考慮した総合品質は
１００＊（１−（ｅｘｐ（０．３）（ｅｘｐ（０．４）−１））−Ｂ＊（２＊５０）
となる。第一項で１００より減少分が存在し、また、さらに第二項が段差を含めており、過去の劣化の影響が反映されていることが分かる。
【００３０】
なお、この例においては、説明の都合上、過去の品質による影響として極めて単純なモデルを用い、品質評価についても１次元の総合品質のみを用いているが、より精緻なモデルや、あるいは総合品質だけでなく、映像の特性情報、音声品質などより細かい情報などを用いるといった方法により、映像・音声品質のより正確な、人間の長時間主観特性を考慮した客観評価が可能である。
【００３１】
また、ここでは、人間の長時間主観評価特性を（１）式のような式でモデル化し、長時間評価特性計算部１３０では、該式により総合品質を求める適用例を示したが、長時間評価特性計算部１３０に主観品質の特性をモデル化して反映させる手段としては、あらゆるパターンについての特性を表などとして格納しておく手段（予め主観評価実験などにより値を決定しておく）、代表的なパターンについての値を元に近似する値を返す手段などが考えられる。
【００３２】
なお、図１で示した装置における各部の一部もしくは全部の処理機能をコンピュータのプログラムで構成し、そのプログラムをコンピュータを用いて実行して本発明を実現することができること、あるいは、図２で示した処理手順をコンピュータのプログラムで構成し、そのプログラムをコンピュータに実行させることができることは言うまでもない。また、コンピュータでその処理機能を実現するためのプログラム、あるいは、コンピュータにその処理手順を実行させるためのプログラムを、そのコンピュータが読み取り可能な記録媒体、例えば、ＦＤやＭＯ、ＲＯＭ、メモリカード、ＣＤ、ＤＶＤ、リムーバブルディスクなどに記録して、保存したり、提供したりすることができるとともに、インターネット等のネットワークを通してそのプログラムを配布したりすることが可能である。
【００３３】
【発明の効果】
本発明によれば、人間の劣化知覚特性に対応した特徴物理量時系列を元に、予め用意した時間的に変動する品質劣化に対応した人間の認知特性を表わすモデルに基づいて総合品質評価値を推定するため、主観品質を適切に推定できる。
【図面の簡単な説明】
【図１】本発明の一実施形態の装置構成のブロック図である。
【図２】本発明の一実施形態の処理フロー図である。
【図３】本発明による適用例を示す図である。
【図４】本発明による適用例の重みを示す図である。
【符号の説明】
１０ 映像・音声信号
２０ 客観品質評価結果
１００ 映像・音声品質客観評価装置
１１０ 短時間評価特性計算部
１２０ 一時記憶部
１３０ 長時間評価特性計算部
１４０ 制御部[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention provides an objective video / audio quality objective for estimating subjective quality from measurement of physical features of a video or audio signal without performing a subjective evaluation test for evaluating the quality of the video or audio by watching or listening to the video or audio. It relates to an evaluation method and an apparatus.
[0002]
[Prior art]
2. Description of the Related Art In recent years, with the increase in the speed and bandwidth of Internet access lines, streaming of video and audio distribution services has become widespread. In order to provide a comfortable streaming service of video and audio, quality design and management based on appropriate video and audio quality evaluation technology is important.
[0003]
The quality of video / audio is basically evaluated by a subjective evaluation test in which a large number of subjects actually watch or listen to video / audio and evaluate the quality (for example, see Non-Patent Document 1). However, performing a subjective evaluation test requires a lot of trouble, time and cost. In addition, since it is necessary to eliminate external factors that influence the evaluation as much as possible, a dedicated evaluation facility is required, and thus the places where experiments can be performed are limited.
[0004]
As a means for solving this, there is an objective quality evaluation device. The objective quality evaluation device receives a video or audio signal to be evaluated as an input, and estimates the subjective quality without performing a subjective evaluation experiment based on the distortion amount output by the objective evaluation model. A typical quality evaluation model tool for video streaming is, for example, Video QoS by Genista (see, for example, Non-Patent Documents 2 and 3).
[0005]
[Non-patent document 1]
Ginga Kawaguchi et al., "Subjective Quality Evaluation of the Effect of Bit Rate Fluctuation on Video Distribution Services", IEICE Technical Report CQ2002-71, Jul. 2002
[Non-patent document 2]
MCM Japan, “VideoQoS, Quality Analysis Software for Video Streaming,” October 9, 2001, Broadband Watch website, <URL: http://www.watch.impress.co.jp/broadband/news/2001. /10/09/videoqos.htm>
[Non-Patent Document 3]
Genista Corporation, "Video QoS Perceived Video Quality Evaluation", October 10, 2001, Genista, Inc., <URL: http: // www. cidf. org / japanese / information / docs / cidf-gen-49. pdf>
[0006]
[Problems to be solved by the invention]
In the prior art, a value obtained by averaging characteristic physical quantities obtained for each short time section (several tens of msec) of a video or audio signal over an evaluation time length (for example, several seconds) is output from an objective evaluation device (that is, an overall quality evaluation value). ) Is used. This method can be expected to be consistent with the subjective quality when the fluctuation of the quality (feature physical quantity) is small and the evaluation time length is short (for example, several seconds).
[0007]
However, when the evaluation time length is long and the quality fluctuates over time, the human degradation perception characteristic cannot be represented by a simple time average of the quality of the short-term section, and is obtained as a result. The estimated value may deviate from the actual subjective quality (Non-Patent Document 1).
[0008]
Generally, the time length of the contents (programs and the like) of the video / audio distribution service is usually several minutes to several hours, and in view of the fact that the quality may fluctuate due to a change in the state of the network during this time, Applying the prior art directly to the evaluation of such a long-time service leads to a reduction in quality estimation accuracy.
[0009]
An object of the present invention is to provide a video and audio quality subjective evaluation method and apparatus that solve the above-described problems and appropriately estimate subjective quality from measurement of objective physical features of video and audio signals. .
[0010]
[Means for Solving the Problems]
According to the present invention, an evaluation value representing an objective quality of a video / audio signal to be evaluated for each short time (for example, several seconds) is held for a predetermined time in the past (for example, several minutes to several tens of minutes), and sequentially and sequentially. On the basis of the evaluation value for a certain period of time, a total quality reflecting a long-term subjective evaluation characteristic of a human is estimated based on a comprehensive quality evaluation value characteristic model prepared in advance.
[0011]
In the present invention, to estimate a total quality evaluation value based on a model representing a human cognitive characteristic corresponding to a temporally varying quality deterioration prepared in advance from a characteristic physical quantity time series element corresponding to a human deterioration perception characteristic. As long as the validity of the feature physical quantity and the model is maintained, the subjective quality can be appropriately estimated in the method of estimating the overall quality from the deterioration time series.
[0012]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram of one embodiment of the objective video / audio quality evaluation apparatus of the present invention. In FIG. 1, the video / audio quality objective evaluation device 100 includes a short-time evaluation characteristic calculation unit 110, a temporary storage unit 120, a long-time evaluation characteristic calculation unit 130, and a control unit 140 that controls the operations of these units. .
[0013]
The short-time evaluation characteristic calculation unit 110 continuously inputs the video / audio signal 10 to be evaluated in a streaming format and, based on an objective evaluation model built in advance, objective quality of the evaluation target signal every short time (for example, several seconds). Are calculated and sequentially output. That is, the short-time evaluation characteristic calculation unit 110 calculates the characteristic physical quantity obtained for each short-time section (several tens of msec) in the video / audio signal to be evaluated in a predetermined evaluation time length (for example, several seconds), similarly to the conventional objective evaluation device. ) Are sequentially output as an evaluation value (short-time characteristic) representing the objective quality of the signal to be evaluated. For this short-time evaluation characteristic calculation unit 110, for example, VideoQos from Genesta can be used as it is (for example, see Non-Patent Documents 2 and 3). In this case, in addition to the video / audio signal 10 to be evaluated (processed), a video / audio signal serving as the reference is separately input.
[0014]
Temporary storage section 120 stores the evaluation values sequentially output from short-time evaluation characteristic calculation section 110 in time series. That is, the temporary storage unit 120 stores the short-term characteristics of the physical feature amounts of the video / audio signals to be evaluated in a time-series manner. The temporary storage unit 120 is realized by, for example, a buffer memory having a shift register array configuration, and data (evaluation values) after a certain time is sequentially discarded. The point in the past at which the evaluation value is stored in the temporary storage unit 120 is determined by the required specifications of the comprehensive quality evaluation characteristic model (subjective evaluation characteristic model) included in the long-time evaluation characteristic calculation unit 130.
[0015]
The long-term evaluation characteristic calculation unit 130 stores a predetermined period of time (for example, several minutes to several tens of minutes) from the temporary storage unit 120 every predetermined time (for example, several seconds) in parallel with the processing of the short-time evaluation characteristic calculation unit 110. , And based on a built-in comprehensive quality evaluation characteristic model, a comprehensive evaluation value (total quality) that reflects human long-term subjective evaluation characteristics is calculated and output as an objective quality evaluation result 20. The comprehensive quality evaluation characteristic model prepared in the long-time evaluation characteristic calculation unit 130 will be described later.
[0016]
The control unit 140 has, for example, a CPU configuration, and controls each operation timing of the short-time evaluation characteristic calculation unit 110, the temporary storage unit 120, and the long-time evaluation characteristic calculation unit 130.
[0017]
FIG. 2 shows an overall processing flow of the objective video / audio quality evaluation apparatus 100. In FIG. 2, the time t is the evaluation time length of the required specification of the short-time evaluation characteristic calculation unit 100 (for example, several seconds), and the time τ is the evaluation time length of the required specification of the long-time characteristic calculation unit 130 (for example, several minutes to several tens of seconds). Minute).
[0018]
The video / audio quality objective evaluation apparatus 100 starts up the apparatus and starts operation to continuously input the video / audio signal 10 to be evaluated in a streaming format (S1). As described above, a video / audio signal serving as a reference is separately input if necessary. For the video / audio signal 10 to be evaluated which is continuously input, the short-time evaluation characteristic calculation unit 110 evaluates the time length (t) at predetermined time intervals (t) based on a built-in objective evaluation model. Values (short-time characteristics) are calculated and sequentially output.
[0019]
First, in the first fixed time corresponding to the time (τ) from the start of the operation, a process of simply storing the evaluation values sequentially output from the short-time evaluation characteristic calculation unit 110 in the temporary storage unit 120 in time series is performed. Execute. This is indicated by the pre-processing of step S2 in FIG.
[0020]
Thereafter, for each predetermined time (t) (S3), the short-time evaluation characteristic calculation unit 110 calculates the evaluation value of the time length (t) for the video / audio signal 10 to be evaluated, and sends the evaluation value to the temporary storage unit 120. In parallel with the processing of storing in chronological order (S5, S6), the long-term evaluation characteristic calculation unit 130 fetches the evaluation value for a predetermined time (τ) from the temporary storage unit 120 and stores the integrated quality evaluation characteristic model , A total evaluation value reflecting the human long-term subjective evaluation characteristics is calculated and output as an objective quality evaluation result 20 (S7, S8, S9). This is repeated until the video / audio signal to be evaluated is no longer input (S4). As described above, the data in the temporary storage unit 120 is discarded sequentially after a certain time (τ).
[0021]
Next, a comprehensive quality evaluation characteristic model (subjective evaluation characteristic model) of the long-time evaluation characteristic calculation unit 130 will be described. The comprehensive quality evaluation characteristic model needs to reflect human long-term subjective evaluation characteristics. That is, it is necessary that the user who is continuously viewing the video and audio can estimate the subjective quality currently felt by the viewing media, including the influence of the past quality deterioration state. The effects of the deterioration that occurred when viewing video and audio were the forgetting effect (the effect of forgetting that there was deterioration over time) and the step-related effect (the "quality change" in addition to the average change characteristics). (For example, an effect of sensing itself as a disturbing factor) (for example, see Non-Patent Document 1).
[0022]
Here, considering these, as a comprehensive quality evaluation characteristic model,
(1) A model in which even if quality degradation has the same degradation time length and degradation degree, the influence on the overall quality decreases as the temporal position at which the degradation occurs moves away from the point of quality judgment. (2) The same total degradation time (3) In the model shown in the above (1), when the frequency of the quality change is higher, the quality of the quality changes more frequently. Assume a model where the effect is dominant. Actually, if necessary, an overall quality evaluation characteristic model is set using any one or a combination of any two of them, or all three. Here, as an example, a model combining (1) and (2) is assumed.
[0023]
The quality at time t (the objective evaluation value measured by the short-time evaluation characteristic calculation unit 110) is defined as Q (t). At time 0, the total quality (comprehensive evaluation value calculated by the long-time evaluation characteristic calculation unit) Q_g (t) in this model in consideration of the human long-term subjective evaluation characteristics is expressed by the following equation.
[0024]
(Equation 1)
Figure 2004172753
[0025]
Here, the first term on the right-hand side represents that the user immediately remembers the one immediately after viewing and thus has a great influence, but forgets that over time. However, since the time τ has passed little, it is assumed that it has been completely forgotten. The second term indicates that when a change in the bit rate (change in quality) occurs, the absolute value of the “change” is perceived as degradation, which is stronger immediately after viewing.
[0026]
The features of this model are as follows.
(1) If the quality is always constant Q_c, it will be Q_c even if time change is considered. (2) If the quality has deteriorated within time τ, the currently reproduced short-time quality is Q_c. Even if there is a time effect, it becomes lower than Q_c.
(3) Regardless of how short the quality deterioration occurs, the second term has an effect of the occurrence of the change (the effect of subjective characteristics other than mere average).
[0027]
In the case of this model, since the quality at a certain time t is affected only by the past τ time, the amount necessary for the temporary storage unit 120 in FIG. Quality is not required in this model because it has no effect.)
[0028]
FIG. 3 shows an application example of the present model. FIG. 4 shows the weight (evaluation characteristic weight) w (t) of this model.
[0029]
In FIG. 3, the upper part shows an example of the temporal change of the short-time quality (the objective evaluation value measured by the short-time evaluation characteristic calculator 110) Q (t), and the lower part shows the temporal change of the present model. It shows an example of a temporal change of the overall quality including the influence (the overall evaluation value obtained by the long-time evaluation characteristic calculation unit 130) Qg (t). The temporal change of the short-term quality Q (t) is
Time -r ~ -0.7τ4, -0.3τ ~ 0: Quality 100
Time −0.7τ4, to −0.3τ: quality 50
It is. The present short-time quality of display / playback is 100, but the effect that the user remembers the deterioration from the past time of 1-0.7τ to -50 at −0.3τ is evaluated by this model. Then, the total quality considering the current time change by integrating the equation (1) is 100 * (1- (exp (0.3) (exp (0.4) -1))-B * (2 * 50 )
It becomes. In the first term, there is a decrease from 100, and further, the second term includes a step, which indicates that the influence of the past deterioration is reflected.
[0030]
In this example, for the sake of explanation, a very simple model is used as the influence of past quality, and only one-dimensional overall quality is used for quality evaluation. Not only that, but also by using more detailed information such as video characteristic information and audio quality, it is possible to perform more accurate video and audio quality objective evaluation in consideration of human long-term subjective characteristics.
[0031]
Also, here, an application example in which the human long-term subjective evaluation characteristics are modeled by an expression such as Expression (1), and the long-time evaluation characteristic calculation unit 130 obtains the overall quality by the expression is shown. Means for modeling and reflecting the characteristics of the subjective quality in the evaluation characteristic calculation unit 130 include means for storing characteristics of all patterns as a table (a value is determined in advance by a subjective evaluation experiment or the like), representative. Means for returning an approximate value based on the value of a typical pattern can be considered.
[0032]
It should be noted that some or all of the processing functions of each unit in the apparatus shown in FIG. 1 are configured by a computer program, and that the program can be executed using a computer to realize the present invention. Needless to say, the processing procedure shown can be configured by a computer program and the computer can execute the program. In addition, a program for realizing the processing function of the computer or a program for causing the computer to execute the processing procedure may be stored in a computer-readable recording medium such as an FD, an MO, a ROM, a memory card, and a CD. , A DVD, a removable disk, or the like, and can be stored or provided, and the program can be distributed through a network such as the Internet.
[0033]
【The invention's effect】
According to the present invention, based on a characteristic physical quantity time series corresponding to human deterioration perception characteristics, a comprehensive quality evaluation value is calculated based on a model prepared in advance and representing a human cognitive characteristic corresponding to temporally varying quality deterioration. Because of the estimation, the subjective quality can be appropriately estimated.
[Brief description of the drawings]
FIG. 1 is a block diagram of an apparatus configuration according to an embodiment of the present invention.
FIG. 2 is a processing flowchart of an embodiment of the present invention.
FIG. 3 is a diagram showing an application example according to the present invention.
FIG. 4 is a diagram showing weights of an application example according to the present invention.
[Explanation of symbols]
Reference Signs List 10 Video / audio signal 20 Objective quality evaluation result 100 Video / audio quality objective evaluation device 110 Short-time evaluation characteristic calculation unit 120 Temporary storage unit 130 Long-time evaluation characteristic calculation unit 140 Control unit

Claims (4)

人間が感じる映像・音声品質（これを主観品質と呼ぶ）を評価対象の映像・音声信号の物理的特徴量から推定する映像・音声品質客観評価方法であって、
評価対象の映像・音声信号の短時間毎の客観品質を表わす評価値を求めるステップと、
前記求めた評価値を時系列的に記憶するステップと、
前記記憶された一定時間分の評価値をもとに、予め定めた総合品質評価特性モデルに基づき、人間の長時間主観評価特性を反映した総合評価値（総合品質）を求めるステップと、
を有することを特徴とする映像・音声品質客観評価方法。An objective video / audio quality evaluation method for estimating video / audio quality perceived by humans (this is referred to as subjective quality) from physical features of video / audio signals to be evaluated.
Obtaining an evaluation value representing the objective quality of the video / audio signal to be evaluated for each short time;
Storing the obtained evaluation values in chronological order;
Obtaining a comprehensive evaluation value (total quality) reflecting human long-term subjective evaluation characteristics based on a predetermined comprehensive quality evaluation characteristic model based on the stored evaluation values for a predetermined time;
A video / audio quality objective evaluation method characterized by having: 請求項１記載の映像・音声品質客観評価方法において、
前記総合品質評価特性モデルとして、
（１） 同一の劣化時間長・劣化度合いの品質劣化であっても、劣化の発生した時間的位置が品質判断時点から離れるほど総合品質に与える影響が低下するモデル、
（２） 同一の総劣化時間長／劣化度合いであっても、品質変動の頻度が高いほど総合品質が低下するモデル、
（３） 上記（１）に示すモデルにおいて、ある一定の劣化時間長あるいは劣化度合いを超えた場合にはその影響が支配的となるモデル、
の一つまたは複数を用いることを特徴とする映像・音声品質客観評価方法。The video / audio quality objective evaluation method according to claim 1,
As the comprehensive quality evaluation characteristic model,
(1) Even if the quality degradation has the same degradation time length and degradation degree, the effect on the overall quality decreases as the temporal position at which the degradation occurs moves away from the quality judgment point.
(2) Even if the total deterioration time length / deterioration degree is the same, the higher the frequency of quality change, the lower the overall quality,
(3) In the model shown in the above (1), when a certain deterioration time length or a certain degree of deterioration is exceeded, the influence is dominant;
A video / audio quality objective evaluation method characterized by using one or more of the following. 人間が感じる映像・音声品質（主観品質）を評価対象の映像・音声信号の物理的特徴量から推定する映像・音声品質客観評価装置であって、
連続的に入力される評価対象の映像・音声信号の短時間毎の客観品質を表わす評価値を求める手段と、
前記求めた評価値を時系列的に記憶する手段と、
前記記憶された一定時間分の評価値をもとに、予め定めた総合品質評価特性モデルに基づき、人間の長時間主観評価特性を反映した総合評価値（総合品質）を求めて出力する手段と、
を有することを特徴とする映像・音声品質客観評価装置。An objective video / audio quality evaluation device for estimating video / audio quality (subjective quality) felt by humans from physical features of video / audio signals to be evaluated,
Means for obtaining an evaluation value representing the objective quality of the video / audio signal to be evaluated continuously input for each short time;
Means for storing the obtained evaluation value in chronological order,
Means for obtaining and outputting an overall evaluation value (total quality) reflecting a long-term subjective evaluation characteristic of a human being based on a predetermined overall quality evaluation characteristic model based on the stored evaluation values for a certain period of time; ,
An objective video / audio quality evaluation apparatus characterized by having: 請求項３記載の映像・音声品質客観評価装置において、
前記総合品質評価特性モデルとして、
（１） 同一の劣化時間長・劣化度合いの品質劣化であっても、劣化の発生した時間的位置が品質判断時点から離れるほど総合品質に与える影響が低下するモデル、
（２） 同一の総劣化時間長／劣化度合いであっても、品質変動の頻度が高いほど総合品質が低下するモデル、
（３） 上記（１）に示すモデルにおいて、ある一定の劣化時間長あるいは劣化度合いを超えた場合にはその影響が支配的となるモデル、
の一つまたは複数を用いることを特徴とする映像・音声品質客観評価装置。The video / audio quality objective evaluation device according to claim 3,
As the comprehensive quality evaluation characteristic model,
(1) Even if the quality degradation has the same degradation time length and degradation degree, the effect on the overall quality decreases as the temporal position at which the degradation occurs moves away from the quality judgment point.
(2) Even if the total deterioration time length / deterioration degree is the same, the higher the frequency of quality change, the lower the overall quality,
(3) In the model shown in the above (1), when a certain deterioration time length or a certain degree of deterioration is exceeded, the influence is dominant;
An objective video / audio quality evaluation apparatus characterized by using one or more of the following.
JP2002333898A 2002-11-18 2002-11-18 Method and apparatus for objectively evaluating video / audio quality Pending JP2004172753A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2002333898A JP2004172753A (en) 2002-11-18 2002-11-18 Method and apparatus for objectively evaluating video / audio quality

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2002333898A JP2004172753A (en) 2002-11-18 2002-11-18 Method and apparatus for objectively evaluating video / audio quality

Publications (1)

Publication Number Publication Date
JP2004172753A true JP2004172753A (en) 2004-06-17

Family

ID=32698480

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2002333898A Pending JP2004172753A (en) 2002-11-18 2002-11-18 Method and apparatus for objectively evaluating video / audio quality

Country Status (1)

Country Link
JP (1) JP2004172753A (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006043500A1 (en) * 2004-10-18 2006-04-27 Nippon Telegraph And Telephone Corporation Video quality objective evaluation device, evaluation method, and program
JP2007194893A (en) * 2006-01-19 2007-08-02 Nippon Telegr & Teleph Corp <Ntt> Image quality evaluation apparatus and method
EP1864511A1 (en) * 2005-03-29 2007-12-12 France Telecom S.A. Method and device for evaluating degradation of quality caused by an invariance of a stimulus, as perceived by a recipient of said stimulus
JP2009273127A (en) * 2008-04-30 2009-11-19 Thomson Licensing Method for assessing quality of distorted version of frame sequence
JP2011250049A (en) * 2010-05-26 2011-12-08 Nippon Hoso Kyokai <Nhk> Live atmosphere estimation device and program for the same
WO2017104416A1 (en) * 2015-12-16 2017-06-22 日本電信電話株式会社 Audio/visual quality estimation device, method for estimating audio/visual quality, and program
JP2019526190A (en) * 2016-06-29 2019-09-12 テレフオンアクチーボラゲット エルエム エリクソン（パブル） Quality estimation for adaptive multimedia streaming
WO2020036004A1 (en) * 2018-08-16 2020-02-20 日本電信電話株式会社 Evaluating method, evaluating device, and program

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006043500A1 (en) * 2004-10-18 2006-04-27 Nippon Telegraph And Telephone Corporation Video quality objective evaluation device, evaluation method, and program
US8130274B2 (en) 2004-10-18 2012-03-06 Nippon Telegraph And Telephone Corporation Video quality objective assessment device, assessment method, and program
JP4733049B2 (en) * 2004-10-18 2011-07-27 日本電信電話株式会社 Image quality objective evaluation apparatus, evaluation method and program
JP2008538666A (en) * 2005-03-29 2008-10-30 フランス テレコム Method and apparatus for assessing quality degradation perceived by a recipient of a stimulus caused by the invariance of the stimulus
EP1864511A1 (en) * 2005-03-29 2007-12-12 France Telecom S.A. Method and device for evaluating degradation of quality caused by an invariance of a stimulus, as perceived by a recipient of said stimulus
JP4490374B2 (en) * 2006-01-19 2010-06-23 日本電信電話株式会社 Video quality evaluation apparatus and method
JP2007194893A (en) * 2006-01-19 2007-08-02 Nippon Telegr & Teleph Corp <Ntt> Image quality evaluation apparatus and method
JP2009273127A (en) * 2008-04-30 2009-11-19 Thomson Licensing Method for assessing quality of distorted version of frame sequence
JP2011250049A (en) * 2010-05-26 2011-12-08 Nippon Hoso Kyokai <Nhk> Live atmosphere estimation device and program for the same
WO2017104416A1 (en) * 2015-12-16 2017-06-22 日本電信電話株式会社 Audio/visual quality estimation device, method for estimating audio/visual quality, and program
JPWO2017104416A1 (en) * 2015-12-16 2018-08-30 日本電信電話株式会社 Audio visual quality estimation apparatus, audio visual quality estimation method, and program
US10869072B2 (en) 2015-12-16 2020-12-15 Nippon Telegraph And Telephone Corporation Audio-visual quality estimation device, method for estimating audio-visual quality, and program
JP2019526190A (en) * 2016-06-29 2019-09-12 テレフオンアクチーボラゲット エルエム エリクソン（パブル） Quality estimation for adaptive multimedia streaming
US11463742B2 (en) 2016-06-29 2022-10-04 Telefonaktiebolaget Lm Ericsson (Publ) Quality estimation of adaptive multimedia streaming
WO2020036004A1 (en) * 2018-08-16 2020-02-20 日本電信電話株式会社 Evaluating method, evaluating device, and program

Similar Documents

Publication Publication Date Title
Streijl et al. Mean opinion score (MOS) revisited: methods and applications, limitations and alternatives
EP1924101B1 (en) Video communication quality estimation device, method, and program
Chen et al. Quadrant of euphoria: a crowdsourcing platform for QoE assessment
EP3659344B1 (en) Calibration system for audience response capture and analysis of media content
Chen et al. Modeling the time—Varying subjective quality of HTTP video streams with rate adaptations
Hines et al. ViSQOLAudio: An objective audio quality metric for low bitrate codecs
WO2017126288A1 (en) Viewing material evaluation method, viewing material evaluation system, and program
CN101309400B (en) Stream media service pause information acquisition method and device
WO2006110472A2 (en) Website evaluation tool
Ghadiyaram et al. A time-varying subjective quality model for mobile streaming videos with stalling events
JP6952061B2 (en) Quality estimation of adaptive multimedia streaming
Weiss et al. Temporal development of quality of experience
JP2004172753A (en) Method and apparatus for objectively evaluating video / audio quality
Ickin et al. VLQoE: Video QoE instrumentation on the smartphone
US9571360B2 (en) Method and score management node for supporting service evaluation
JP5957419B2 (en) QoE estimation apparatus, QoE estimation method and program
US20150269952A1 (en) Method, an apparatus and a computer program for creating an audio composition signal
Zamora et al. Smartphone tuning for accurate ambient noise assessment
JP6845187B2 (en) Quality estimation device, quality estimation method and program
Salvador et al. QoE assessment of HTTP adaptive video streaming
Gallardo et al. Perceptual Ratings of Voice Likability Collected Through In-Lab Listening Tests vs. Mobile-Based Crowdsourcing.
Bockstael et al. Verifying the attenuation of earplugs in situ: Method validation using artificial head and numerical simulations
EP3391592B1 (en) Quality of experience monitoring system and method
Völk Headphone Selection for Binaural Synthesis with Blocked Auditory Canal Recording
JP5779616B2 (en) QoE evaluation apparatus, QoE evaluation method and program

Legal Events

Date Code Title Description
A621 Written request for application examination

Effective date: 20050118

Free format text: JAPANESE INTERMEDIATE CODE: A621

A977 Report on retrieval

Effective date: 20061227

Free format text: JAPANESE INTERMEDIATE CODE: A971007

A131 Notification of reasons for refusal

Effective date: 20070117

Free format text: JAPANESE INTERMEDIATE CODE: A131

A02 Decision of refusal

Free format text: JAPANESE INTERMEDIATE CODE: A02

Effective date: 20070523