JP3967236B2 - Personal digest distribution apparatus, personal digest distribution method and program - Google Patents

Description

【００５８】
イベントａのＲ選手の打席開始からイベントｃのＭ選手による加点の直前のイベントは、Ｒ選手の打席開始から次打者のＭ選手の打席開始の直前までのイベントを含んでいる。したがって、ダイジェストシーン生成部１０１では、イベントａからイベントｃまでのシーンを１５秒に編集してダイジェストシーンが作成される。なお、１つのダイジェストシーンには、打席開始から次打者の打席開始直前までに発生する複数のイベントがあり、ダイジェストシーンとイベントとは、１対複数の関係に対応する。各ダイジェストシーンに含まれるイベントを時系列に並べた集合をイベントリストといい、図中にＲで示す。
【００５９】
映像内容解析部
映像内容解析部１０７は、メタデータＤＢ１０５に蓄積されているイベント情報Ｉを入力し、イベント情報Ｉに基づいて各ダイジェストシーンの重要度を算出する。重要度は、攻撃、守備、状況の３つの要素ごとに算出される。各重要度の算出は、重要度パラメタに基づいてなされる。攻撃、守備、状況の各重要度の算出に用いられる重要度パラメタを、それぞれ攻撃レベル、守備レベル、状況レベルというものとする。
【００６０】
図５は、攻撃レベル、守備レベル、状況レベルの特徴と算出のタイミングを示す図である。なお、守備レベルは、投手に関するイベントを含む守備側のすべてのイベントを考慮した重要度パラメタである。各重要度パラメタは、各ダイジェストシーンにおける攻撃側チーム、守備側チームの有利、あるいは不利、状況の良し悪しを示す。
【００６１】
次に、攻撃レベル、守備レベル、状況レベルを算出する処理を説明する。重要度パラメタは、以下の手順で算出される。
▲１▼イベントごとの守備、攻撃、状況の少なくとも一つに応じた重要度パラメタ算出
▲２▼イベントごとの重要度パラメタをチーム別に算出
【００６２】
▲１▼イベントに基づくパラメタ算出
図６は、攻撃レベル、守備レベル、状況レベルの各重要度パラメタと、各重要度パラメタの値を求めるのに使用されるイベントとの対応を一覧表にして示した図である。各重要度パラメタは、表の右端の欄にあるイベントごとに求められ、１つのイベントは、１種類の重要度パラメタを求めるためにだけ参照され、複数の重要度パラメタを得ることに使用されない。求められた重要度パラメタは、該当するイベントの属性情報としてメタデータＤＢ１０５に保存される。
【００６３】
イベントには、算出される重要度パラメタの値をプラスにするイベント、マイナスにするイベントがある。例えば攻撃レベルは、サヨナラ、逆転、先制といった攻撃側に有利なイベントの発生によって高くなる。また、オーバーラン、守備妨害といった守備側に有利なイベントによって低くなる。守備レベルは、トリプルプレー、三者連続三振、好捕球といった守備側に有利なイベントによって高くなり、エラーのように攻撃側に有利なイベントの発生によって低くなる。
【００６４】
また、状況レベルは、各ダイジェストシーンの先頭のイベント（通常は打席の開始）のイベント情報に付された属性情報に基づいて算出される。属性情報とは、例えば、打席開始時に塁上にいる走者に関する情報や打者の打順に関する情報（クリーンアップ、あるいは下位打線であるか）である。映像内容解析部１０７は、属性情報を参照し、得点や逆転の可能性が高い場合に状況レベルをより高く設定する。
【００６５】
以上述べた処理によれば、ダイジェストシーンに属する各イベントの重要度を判断することができる。また、各ダイジェストシーンが攻撃または守備の観点からどの程度重要なシーンであるかを判断することができる。さらに、状況の観点から、ダイジェストシーンがみどころ（いわゆるいい場面）の度合いを判断することができる。
【００６６】
▲２▼イベントごとの重要度パラメタをチーム別に算出
次に、映像内容解析部１０７は、▲１▼の処理により、イベント単位で算出された守備レベル、攻撃レベル、状況レベルの少なくとも一つに応じた重要度パラメタに基づいて、イベントごとの重要度パラメタをチーム別に算出する。各イベントのチーム別の重要度パラメタは、以下のようにして求められる。
【００６７】
映像内容解析部１０７は、ダイジェストシーンにおいて攻撃側にあるチームに対し、ダイジェストシーン上で発生したイベントの攻撃レベルをそのままの攻撃レベルとする。一方、守備側のチームに対しては、攻撃側にあるチームの攻撃レベルに−１を乗じた値をイベントの攻撃レベルとする。同様に、ダイジェストシーンにおいて守備側にあるチームに対し、ダイジェストシーン上で発生したイベントの守備レベルをそのままの攻撃レベルとする。一方、攻撃側のチームに対しては、守備側にあるチームの守備レベルに−１を乗じた値をイベントの守備レベルとする。以上のように、各イベントの攻撃レベルと守備レベルとが、チーム別に算出される。
【００６８】
チームごとの状況レベルは、ダイジェストシーンの開始イベントの重要度パラメタとして算出される。状況レベルは、チームの攻撃、守備に関わらず、プラス、マイナスが変更されることがない。
【００６９】
以上の処理によれば、各イベントの攻撃、守備、状況に関する重要度を、チーム別に判断することができる。
【００７０】
また、本実施の形態は、上記した実施の形態のようにチームではなく、対象として特定の選手が登録されている場合登録されている選手（登録選手）に関するイベントに基づいて各イベントの攻撃レベル、守備レベル、状況レベルを選手別に算出することができる。
【００７１】
選手別の守備レベル、攻撃レベル、状況レベルは、例えば、登録選手がサヨナラ、逆転のイベントに関わった場合、攻撃レベルが高まるように設定し、登録選手の攻撃レベルに−１を乗じて相手選手の攻撃レベルとすることによって算出することができる。また、登録選手が相手選手をアウトにする、三振にとるといったイベントが発生した場合、守備レベルが高まるように設定し、登録選手の守備レベルに−１を乗じて相手選手の守備レベルとすることによって算出することができる。さらに、状況レベルは、ダイジェストシーンの打席開始時の打者が登録選手であったときのイベントに基づいて算出することができる。
【００７２】
シーン選択部
シーン選択部１１１は、嗜好情報登録部１０９に登録された対象を含む対象およびイベント、さらに配信希望の度合いに基づいてダイジェストシーンを選択する構成である。このため、シーン選択部１１１は、チーム別の各ダイジェストシーンの重要度を、ユーザごとに算出する。各ダイジェストシーンのチーム別、ユーザ別の重要度は、以下のようにして算出される。
【００７３】
本実施の形態のシーン選択部１１１は、映像内容解析部１０７でチーム別に算出されたイベントごとの重要度パラメタに加え、嗜好情報登録部１０９に登録されたイベントをも用いて各ダイジェストシーンのチーム別、ユーザ別の重要度を算出する。すなわち、見たいイベント（ｅｖｅｎｔ）を登録したユーザにとってのダイジェストシーンＳｉのチームＴｊの攻撃、守備、状況に関する重要度パラメタ（Ｓｉ．Ｔｊ．Ｐｋ）は、以下の式によって算出される。
Ｓｉ．Ｔｊ．Ｐｋ＝Ｆｕｎｃ（Ｅｋ，Ｔｊ，Ｐｋ，Ｆａｖ＿ｅｖｅｎｔ）
Ｓｉ：ダイジェストシーン
Ｔｊ：チーム識別子（ホームチームまたはアウェイチーム）
Ｐｋ：攻撃、守備、状況レベルを示す識別子
Ｅｋ：ダイジェストシーンＳｉに属するイベントの集合
Ｆａｖ＿ｅｖｅｎｔ：見たいイベント
Ｆｕｎｃ：ダイジェストシーンＳｉに属するイベントの集合、チームの識別子（チームＩＤ）、算出すべき重要度パラメタの要素（攻撃、守備、状況）、ユーザの見たいイベントの型名を引数として、ダイジェストシーンＳｉの攻撃、守備、状況のいずれかに対応する重要度パラメタを、ユーザごと、チームごとに算出する関数
【００７４】
次に、シーン選択部１１１は、ユーザごと、チームごとに算出されたダイジェストシーンＳｉの（攻撃、守備、状況のいずれかに対応する）重要度パラメタ、嗜好情報登録部１０９に登録されたチームおよび利用者が登録されたチームを好きであるか嫌いであるかの別、配信希望の度合い、に基づいて、各ダイジェストシーンの重要度をユーザごとに算出する。算出された重要度は、パーソナルダイジェストを構成するダイジェストシーンの選択に直接用いられるため、最終重要度ともいう。最終重要度を算出する式は、以下の通りである。
シーンＳｉの最終重要度＝
（Ｆａｖ＿ｏｆｆｅｎｓｅ＊ｆｌａｇ＊Ｓｉ．Ｔｊ．Ｐｋ（攻撃レベル））
＋（Ｆａｖ＿ｄｅｆｅｎｓｅ＊ｆｌａｇ＊Ｓｉ．Ｔｊ．Ｐｋ（守備レベル））
＋（Ｆａｖ＿ｃｈａｎｃｅ＊ｆｌａｇ＊Ｓｉ．Ｔｊ．Ｐｋ（状況レベル））
ｆｌａｇ 対象やイベントである嗜好項目を好きか嫌いかの別を示すフラグ、好き：１、嫌い：−１、デフォルト：１
Ｓｉ 映像シーン
Ｔｊ 対象識別子
Ｆａｖ＿ｏｆｆｅｎｓｅ 攻撃的要素の配信希望の度合い
Ｆａｖ＿ｄｅｆｅｎｓｅ 防御的要素の配信希望の度合い
Ｆａｖ＿ｃｈａｎｃｅ 状況的要素の配信希望の度合い
【００７５】
シーン選択部１１１は、上記した式にしたがって各ダイジェストシーンの重要度を算出する。そして、算出された重要度で各ダイジェストシーンをランキングし、ランキングの上位のダイジェストシーンを使って配信を要求したユーザ用のパーソナルダイジェストを作成する。
【００７６】
なお、前記したように、対象を特定の選手に設定し、上記した式のＳｉ．Ｔｊ．重要度パラメタ（Ｐｋ）の項にダイジェストシーンＳｉのユーザごと、選手ごと、攻撃、守備、状況といった要素ごとの重要度パラメタを代入すれば、本実施の形態は、特定の選手の最終重要度を算出することができる。
【００７７】
以上述べた構成は、以下のように動作する。すなわち、ダイジェストシーン生成部１０１は、スポーツ映像Ｓを入力し、平均１５秒程度の映像であるダイジェストシーンを作成する。作成されたダイジェストシーンは、動画ＤＢ１０３に蓄積される。また、作成された各ダイジェストシーン上で発生した複数のイベントを示す情報は、メタデータとしてメタデータＤＢ１０５に蓄積される。
【００７８】
映像内容解析部１０７は、メタデータＤＢ１０５に蓄積されたイベントの情報を読み出し、各イベントの重要度パラメタＰ１を求める。各イベントの重要度パラメタＰ１は、予め設定しておくものであってもよい。また、予め設定された演算式を使って算出するものであってもよい。さらに、映像内容解析部１０７は、各イベントの重要度パラメタＰ１を使って、各イベントの重要度パラメタＰ２をチーム別に算出する。重要度パラメタＰ１、重要度パラメタＰ２は、いずれもメタデータＤＢ１０５に蓄積される。
【００７９】
ところで、嗜好情報登録部１０９には、ユーザの嗜好に関する情報が予め携帯端末のＵＩＦからユーザ自身によって設定、登録されている。ユーザがパーソナルダイジェストの配信を希望したタイミングでパーソナルダイジェストを作成する場合、シーン選択部１１１は、配信要求を受けてユーザＤＢ１１３から配信を要求したユーザの嗜好に関する嗜好項目（配信を希望する対象、見たいイベント）や嗜好の属性（対象が好きか嫌いかの別、配信を希望する度合いなど）の情報ｆを読み出す。
【００８０】
そして、シーン選択部１１１は、情報ｆと重要度パラメタＰ２とに基づいて、ダイジェストシーンの重要度パラメタを、ユーザ別、チーム別に算出する。続いて、算出されたユーザ別、チーム別のダイジェストシーンの重要度パラメタに基づいて、最終重要度パラメタを算出する。さらに、重要度パラメタに基づいてダイジェストシーンをランキングし、パーソナルダイジェストを作成して配信を要求したユーザに配信する。
【００８１】
図７、図８は、以上述べたパーソナルダイジェスト配信装置において行われるパーソナルダイジェストの配信方法を説明するためのフローチャートである。図７はパーソナルダイジェスト作成以前になされる処理を説明するためのフローチャートであり、図８はユーザの嗜好に基づいて、パーソナルダイジェストを作成する方法を説明するためのフローチャートである。
【００８２】
図７に示すように、パーソナルダイジェスト配信装置は、先ず、スポーツ映像が入力されたか否かを判断する(ステップＳ７０１）。スポーツ映像を入力していない場合(ステップＳ７０１：Ｎｏ）、スポーツ映像が入力されるまで待機する。また、スポーツ映像を入力した場合（ステップＳ７０１：Ｙｅｓ）、ダイジェストシーン生成部１０１はダイジェストシーンと共にメタデータを生成する（ステップＳ７０２、７０３）。ダイジェストシーン、メタデータの生成の処理は、スポーツ映像のすべてを処理するまで継続される。生成されたダイジェストシーンは動画ＤＢ１０３に、メタデータはメタデータＤＢ１０５にそれぞれ蓄積される。
【００８３】
ダイジェストシーン生成部１０１は、ダイジェストシーン、メタデータの生成が終了したか否かを判断する（ステップＳ７０４）。処理が終了していないと判断された場合（ステップＳ７０４：Ｎｏ）、終了するまでダイジェストシーン、メタデータの生成を継続する。
【００８４】
また、ダイジェストシーン生成部１０１がダイジェストシーン、メタデータの生成が終了したと判断した場合（ステップＳ７０４：Ｙｅｓ）、映像内容解析部１０７は、ダイジェストシーンに含まれるイベント（メタデータで表される）の重要度パラメタを算出する（ステップＳ７０５）。そして、さらにイベント単位で算出された重要度パラメタから、チーム別の重要度パラメタをイベント単位で算出する（ステップＳ７０６）。イベントごと（イベント単位）の重要度パラメタＰ1、チーム別のイベント単位の重要度パラメタＰ2は、メタデータＤＢ１０５にメタデータと共に蓄積される。
【００８５】
次に、シーン選択部１１１は、図８に示すように、ユーザのパーソナルダイジェストの配信の希望があるか否かを判断する（ステップＳ８０１）。判断の結果、配信の希望がなされていない場合（ステップＳ８０１：Ｎｏ）、配信希望がなされるまで待機する。
【００８６】
また、シーン選択部１１１は、配信を希望したユーザの嗜好情報を嗜好情報登録部１０９から取得する（ステップＳ８０２）。そして、メタデータＤＢ１０５に蓄積された重要度パラメタＰ1、重要度パラメタＰ2を読み出し、チーム別に各ダイジェストシーンの重要度を配信の希望があったユーザごとに算出する（ステップＳ８０３）。続いて、算出されたチーム別のダイジェストシーンの重要度に基づいて、各ダイジェストシーンの最終的な重要度をユーザごとに算出する（ステップＳ８０４)。さらに、ダイジェストシーンを算出した重要度が高い順にランキングし、ランクの高いものから所定数選択してパーソナルダイジェストを作成する（ステップＳ８０５）。シーン選択部１１１は、作成されたパーソナルダイジェストを配信を希望したユーザに配信し、処理を終了する（ステップＳ８０６）。
【００８７】
なお、本実施形態のパーソナルダイジェスト配信方法をコンピュータに実行させるためのプログラムは、インストール可能な形式又は実行可能な形式のファイルでＣＤ−ＲＯＭ、フロッピーディスク（ＦＤ）、ＤＶＤ等のコンピュータで読み取り可能な記録媒体に記録されて提供される。また、パーソナルダイジェスト配信方法をコンピュータに実行させるためのプログラムを、インターネット等のネットワークに接続されたコンピュータ上に格納し、ネットワーク経由でダウンロードさせることにより提供するように構成しても良い。
【００８８】
【発明の効果】
本発明は、映像に関して利用者が配信を希望する嗜好項目と、嗜好項目に関し、利用者が配信を希望する度合いを映像の攻撃、防御、状況の少なくとも一つの要素ごとに登録し、攻撃、防御、状況の少なくとも一つの要素に応じて映像に含まれる各シーンの重要度パラメタを算出し、利用者ごとの各シーンの重要度に基づいて算出された重要度に基づいて映像のうちのシーンをパーソナルダイジェスト選択するため、利用者の配信を希望する嗜好項目、さらには配信希望の度合いを反映した利用者ごとの各シーンの重要度を比較的簡易な演算で算出することができ、より利用者の希望にそったパーソナルダイジェストを作成し、配信することができるという効果を奏する。
【図面の簡単な説明】
【図１】本発明の実施の形態のパーソナルダイジェスト配信装置を説明するためのブロック図である。
【図２】本発明の一実施の形態の嗜好項目の配信希望の度合いを登録するためにユーザがＵＩＦ上で指定する属性を示す図である。
【図３】本発明の一実施の形態のイベントを説明するための図である。
【図４】本発明の一実施の形態のダイジェストシーンとイベントとの関係を説明するための図である。
【図５】本発明の一実施の形態の攻撃レベル、守備レベル、状況レベルの特徴と算出のタイミングを示す図である。
【図６】本発明の一実施の形態の攻撃レベル、守備レベル、状況レベルの各重要度パラメタと、各重要度パラメタの値を求めるのに使用されるイベントとの対応を一覧表にして示した図である。
【図７】本発明の一実施の形態のパーソナルダイジェストの配信方法を説明するためのフローチャートである。
【図８】本発明の一実施の形態のパーソナルダイジェストの配信方法を説明するための他のフローチャートである。
【符号の説明】
１００ パーソナルダイジェスト作成部
１０１ ダイジェストシーン生成部
１０７ 映像内容解析部
１０９ 嗜好情報登録部
１１１ シーン選択部
１０３ 動画ＤＢ
１０５ メタデータＤＢ
Ｐ1 重要度パラメタ
Ｐ2 重要度パラメタ
Ｓ スポーツ映像[0001]
[Prior art]
In recent years, portable information terminals (portable terminals) such as PDAs (Personal Digital Assistants) and mobile phones have become widespread. In particular, third-generation mobile terminals have started services for receiving moving images as well as text and still images at a maximum of 384 Kbps. Furthermore, it is expected that a speech-to-speech function will be implemented in the future.
[0002]
By the way, there is a service for transmitting information related to sports to a mobile terminal. This service informs the user of information about the progress of sports matches such as baseball and soccer, and breaking news of the game results. The user of the service can know the situation and result of the game even when the user is unable to view the television or the like due to reasons such as moving.
[0003]
The service edits the highlights of the game, creates a digest of the game, and creates information that is sent to the mobile terminal. The digest information is composed of a character string and a scene corresponding to the character string, for example, “Team G is reversed due to a timely hit by a player T of team G3-1, team C3-1 and team G”.
[0004]
[Patent Document 1]
JP-A No. 13-275058 (Claims 1 and 6)
[0005]
[Problems to be solved by the invention]
By the way, a user who wants to distribute a personal digest of sports generally has a preference such as a favorite team or a favorite player. Since many portable terminals are owned by individuals, it is conceivable to reflect the preference of each user in the personal digest.
[0006]
Some users who have a preference for sports games want to distribute a scene that their favorite team has an advantage, but do not want to distribute an unfavorable scene such as getting a score. On the other hand, there are some who do not want to distribute a scene that is advantageous to a specific team or player, but want to distribute an unfavorable scene. A person who wants to distribute a scene that is disadvantageous for a specific team or player is called an anti-fan. The conventional personal digest distribution system cannot generate a digest by judging an advantageous scene or an unfavorable scene corresponding to the user's preference.
[0007]
Also, in sports games, although there was no scoring, there are scoring opportunities and defeated scenes. Some users wish to distribute scenes where their favorite players and teams get a chance. However, since the conventional personal digest determines the importance of the scene based on the result of the score or victory, the distribution of the chance scene could not be met.
[0008]
Furthermore, in sports matches, one of the two teams generally attacks and the other defends against the attack. In other words, the game has at least two elements, attack and defense. For example, when distributing a baseball broadcast personal digest, a user who is a fan of a team with excellent pitcher power may desire to see more scenes related to team defense. In addition, a fan of a team that is superior in striking power may desire to see more scenes related to the attack. However, the conventional personal digest device cannot set the degree of distribution desired for each video element, and there is room for improvement to fully reflect the user's preference.
[0009]
  The present invention has been made in view of the above points, and provides a personal digest distribution device, a personal digest distribution method, and a personal digest distribution method for distributing a personal digest more reflecting the user's preference.BipThe purpose is to provide a program. It also meets the demands of users who want to deliver scenes that are particularly disadvantaged, called anti-fans, and can handle not only the results of the game but also the chance scenes to be delivered as digest scenes. Digest distribution device, personal digest distribution method andBipThe purpose is to provide a program.
[0010]
[Means for Solving the Problems]
  In order to solve the above-described problem, a personal digest distribution apparatus according to claim 1 of the present invention is based on a video composed of a plurality of continuous scenes.Personal for the user who requested deliveryA personal digest distribution device that creates and distributes a digest, and the user wishes to distribute the videoPreferences items that are targets and eventsRegistered in the target registration means and the target registration means for registeringPreference itemsThe degree to which the user wishes to distributeAt least one of attack, defense, situationDesired degree registration means to register for each element,Video analysis means for calculating the importance parameter of each scene included in the video according to at least one element of attack, defense, and situation, and the importance of each scene for each user is calculated by the following formula: The calculated importanceBased on the scene of the videoPersonal digestSelected asSelectRuScene selectionMeans,
    Importance = (Fav_offense * flag * Attack element importance parameter)
          + (Fav_defense * flag * defensive element importance parameter)
          + (Fav_chance * flag * situation element importance parameter)
    However, Fav_offense: Degree of desirability of distributing aggressive elements
            Fav_defense: Degree of desirability of distributing defensive elements
            Fav_chance: Degree of distribution of situational elements
            flag: Flag indicating whether the user likes or dislikes the preference item
It is characterized by providing.
[0011]
  According to the first aspect of the present invention, the user desires to distribute the video.Preference itemsWhen,Preference itemsThe degree to which the user wishes to distributeAt least one element of attack, defense, situationYou can register for each. AndCalculate the importance parameter of each scene included in the video according to at least one element of attack, defense, and situation, and calculate the importance based on the importance of each scene for each userBased on the scene in the videoPersonal digest selectioncan do. For this reason, I would like users to distributePreference itemsAnd also reflect the degree of deliveryThe importance of each scene for each user can be calculated with relatively simple calculations.It is possible to create and distribute a personal digest that better meets the user's wishes.
[0012]
  Personal digest delivery according to claim 2MethodIsPersonal digest distribution for creating and distributing personal digests for users who have requested distribution from videos composed of multiple continuous scenes, and including a preference information registration unit, video content analysis unit, and scene selection unit A personal digest distribution method in the apparatus, wherein the preference information registration unit accepts registration of a preference item that is an object or event that a user desires to distribute regarding the video, and the preference information regarding the registered preference item. The registration unit accepts registration of the degree to which the user wants to distribute for at least one element of attack, defense, and situation of the video, and at least one element of attack, defense, and situation by the video content analysis unit And calculating the importance parameter of each scene included in the video according to Selecting said each user by the scene selector importance of each scene, is calculated by the following equation, the scene of the image based on the calculated significance as the personal digest,
    Importance = (Fav_offense * flag * Attack element importance parameter)
          + (Fav_defense * flag * defensive element importance parameter)
          + (Fav_chance * flag * situation element importance parameter)
    However, Fav_offense: Degree of desirability of distributing aggressive elements
            Fav_defense: Degree of desirability of distributing defensive elements
            Fav_chance: Degree of distribution of situational elements
            flag: Flag indicating whether the user likes or dislikes the preference item
includingIt is characterized by that.
[0013]
  According to the invention described in claim 2,The preference item that the user desires to distribute regarding the video and the degree of the user's desire to distribute the preference item can be registered for each of at least one of the attack, defense, and status of the video. Then, the importance parameter of each scene included in the video is calculated according to at least one element of attack, defense, and situation, and the video based on the importance calculated based on the importance of each scene for each user. Of these scenes, a personal digest can be selected. For this reason, it is possible to calculate the importance level of each scene reflecting the preference items desired to be distributed by the user and the degree of distribution desired by a relatively simple calculation. A personal digest can be created and distributed.
[0014]
  Claim 3programIsA program for causing a computer to execute a process of creating and distributing a personal digest for a user who has requested distribution from a video composed of a plurality of continuous scenes. For the target registration function for registering preference items that are events, and the preference items registered in the target registration function, the degree to which the user desires to be distributed is registered for at least one element of the attack, defense, and status of the video A desired degree registration function, a video analysis function that calculates an importance parameter of each scene included in the video according to at least one element of attack, defense, and situation, and an importance level of each scene for each user. , Calculated by the following formula, and selecting a scene of the video as the personal digest based on the calculated importance And the scene selection function that,
    Importance = (Fav_offense * flag * Attack element importance parameter)
          + (Fav_defense * flag * defensive element importance parameter)
          + (Fav_chance * flag * situation element importance parameter)
    However, Fav_offense: Degree of desirability of distributing aggressive elements
            Fav_defense: Degree of desirability of distributing defensive elements
            Fav_chance: Degree of distribution of situational elements
            flag: Flag indicating whether the user likes or dislikes the preference item
Is executed by the computer.
[0015]
  According to the invention described in claim 3,The preference item that the user desires to distribute regarding the video and the degree of the user's desire to distribute the preference item can be registered for each of at least one of the attack, defense, and status of the video. Then, the importance parameter of each scene included in the video is calculated according to at least one element of attack, defense, and situation, and the video based on the importance calculated based on the importance of each scene for each user. Of these scenes, a personal digest can be selected. For this reason, it is possible to calculate the importance level of each scene reflecting the preference items desired to be distributed by the user and the degree of distribution desired by a relatively simple calculation. A personal digest can be created and distributed.
[0040]
DETAILED DESCRIPTION OF THE INVENTION
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, an embodiment of a personal digest delivery apparatus, a personal digest delivery method, and a program for causing a computer to execute the method according to the present invention will be described in detail with reference to the drawings.
[0041]
FIG. 1 is a block diagram for explaining a personal digest distribution apparatus according to an embodiment of the present invention. The personal digest distribution apparatus according to the present embodiment is a personal digest distribution apparatus that creates and distributes a digest from a video composed of a plurality of continuous scenes. The personal digest distribution device according to the present embodiment is a digest that generates metadata and a digest scene that indicate events (events) that occur on a game relay video (sports video) of sports (baseball in the present embodiment). A scene generation unit 101, a metadata database (metadata DB) 105 for storing generated metadata, and a moving image DB 103 for storing digest scenes are provided.
[0042]
In addition, the personal digest distribution apparatus according to the present embodiment includes a target registration unit that registers a target that the user desires to distribute with respect to the sports video S, and a degree that the user desires to distribute with respect to the registered target for each video element. A preference information registration unit 109 which is a desired degree registration unit to be registered, and a personal digest creation unit 100 which selects a digest scene in the sports video S based on the registered target and the degree of distribution desired and creates a personal digest. I have.
[0043]
The preference information registration unit 109 registers the degree of distribution request according to at least one of defense, attack, and situation. In this embodiment, since the sports video S is used as a baseball broadcast video, the defense is particularly referred to as defense.
[0044]
The personal digest creation unit 100 according to the present embodiment selects the digest scene based on the importance, the video content analysis unit 107 that determines the importance of each digest scene of the sports video S according to the defense, the attack, and the situation. And a scene selection unit 111 for performing the operation.
[0045]
Furthermore, the personal digest distribution apparatus according to the present embodiment includes a user DB 113 that stores information registered in the preference information registration unit 109. Hereinafter, among the above-described configurations, the preference information registration unit 109, the digest scene generation unit 101, the video content analysis unit 107, and the scene selection unit 111 will be described in order.
[0046]
Preference information registration department
The preference information registration unit 109 is an object registration unit for registering an object (a favorite team in the present embodiment) that the user desires to distribute regarding the sports video S, and an event (home run) that the user desires to distribute regarding the registered object. It functions as a desired degree registration means for registering the degree of desired delivery of the reversal, etc.). In addition, the preference information registration unit 109 according to the present embodiment can also register an event (an event that the user wants to see) that the user wants to distribute for the sports video S.
[0047]
The user, for example, on the UIF (User Inter Face), the target of preference (the team he is interested in) and the event (the scene he wants to see) (the target team and the event are collectively described as a preference item) Are registered in the preference information registration unit 109.
Interested team (Fav_team)
Events you want to see (Fav_event)
For teams that are interested, the user registers teams that want to get more information. Also, for the event that you want to see, you register the event that you want to see, such as HR and reverse scene.
[0048]
Furthermore, in the personal digest distribution apparatus according to the present embodiment, the preference information registration unit 109 registers the degree of distribution preference of the user's preference items. FIG. 2 shows attributes specified by the user on the UIF in order to register the degree of distribution preference of the preference item.
[0049]
As illustrated in FIG. 2, the preference information registration unit 109 registers whether the user likes or dislikes the target as Fav_flag. Fav_flag is set to 1 when the user likes the target of the preference item, and is set to -1 when the user does not like it. The default of Fav_flag is 1. In addition, the preference information registration unit 109 registers the degree of distribution request according to at least one of a defensive element, an aggressive element, and a situation element.
[0050]
Among the attributes shown in FIG. 2, the degree of distribution request according to the attack level that is an aggressive element is registered as Fav_offense. In Fav_offense, the user registers how much of an aggressively important scene is to be viewed within a range of 1 to α. The default for Fav_offense is 1. Also, the degree of distribution request according to the defensive level, which is a defensive element, is registered as Fav_defense. In Fav_defense, the user registers himself / herself in the range of 1 to β to see how many scenes are important for defensive purposes. The default for Fav_defense is 1. Further, the degree of distribution request according to the situation element is registered as Fav_chance. In Fav_chance, the user registers himself in the range of 1 to γ how much the scene that was a chance is to be viewed.
[0051]
For example, a user who likes Team T very much, strongly desires to deliver an advantageous scene on Team T's attack and defense side, and strongly desires to deliver an opportunity scene for Team T is as follows: By registering each attribute in, it is possible to receive the distribution of the digest scene as desired.
(Fav_name, Fav_flag, Fav_offense, Fav_defense, Fav_chance)
= ("Team T", 1, α, β, γ)
[0052]
In addition, for example, a user who is very disliked by Team G and strongly desires delivery of a disadvantageous scene on Team G's attack side and defense side, and who is not interested in the opportunity scene for Team G is as follows: By registering each attribute in, it is possible to receive the distribution of the digest scene as desired.
(Fav_name, Fav_flag, Fav_offense, Fav_defense, Fav_chance)
= ("Team G", -1,1,1,1)
[0053]
In the present embodiment described above, an example is shown in which the target is a team, and a team T fan or an anti-team G fan distributes a personal digest desired to be distributed. However, the present invention does not limit the target to a team, but can also register a player to meet the demand of a specific player's fan or a specific player's anti-fan.
[0054]
Digest scene generator
The digest scene generation unit 101 is configured as an information terminal device that edits a moving image of a video shot in a baseball field. When an operator presses a registration button on a screen (not shown) of the information terminal, metadata corresponding to one play video scene is generated. The generated video scene is called a digest scene, and the metadata is called event information I in the present embodiment. A digest scene is a scene that is a candidate for a scene for generating a personal digest, and an event indicates an event that has occurred on each digest scene. When events are grouped for each digest scene, this group is called an event list.
[0055]
FIG. 3 is a diagram for explaining an event. Each row of the table shown in FIG. 3 corresponds to one of the basic events that have occurred on the digest scene, and indicates event information that is event information. The event information is represented by an event type name [attribute 1, attribute 2,... Attribute n], and the attribute i is set depending on the event type (start at bat, advance, out, etc.).
[0056]
Event information is created for each digest scene. A digest scene is created for each video (one play scene) that is a digest scene candidate. In the present embodiment, the digest scene is created based on a video including a batter's batter start event to an event just before the next batter's batter start.
[0057]
FIG. 4 is a diagram for explaining the relationship between the digest scene and the event. In the figure, a plurality of events a, b, c... Exist in one play indicated by P. Examples of the events a, b, and c include the following contents.

[0058]
The event immediately before the addition by the M player of the event c from the start of the bat of the R player of the event a includes the event from the start of the batting of the R player to immediately before the start of the bat of the M player of the next batter. Therefore, the digest scene generation unit 101 edits the scene from event a to event c in 15 seconds to create a digest scene. Note that one digest scene has a plurality of events that occur from the start of the bat to the start of the next batter's bat, and the digest scene and the event correspond to a one-to-many relationship. A set in which events included in each digest scene are arranged in time series is called an event list, and is indicated by R in the figure.
[0059]
Video content analysis department
The video content analysis unit 107 inputs the event information I stored in the metadata DB 105 and calculates the importance of each digest scene based on the event information I. The importance is calculated for each of the three elements of attack, defense, and situation. Each importance is calculated based on the importance parameter. The importance parameters used to calculate the importance of attack, defense, and situation are referred to as an attack level, a defense level, and a situation level, respectively.
[0060]
FIG. 5 is a diagram showing characteristics of attack level, defense level, and situation level, and calculation timing. The defensive level is an importance parameter that considers all events on the defensive side including events related to pitchers. Each importance parameter indicates the advantage or disadvantage of the attacking team and the defensive team in each digest scene, and whether the situation is good or bad.
[0061]
Next, processing for calculating an attack level, a defense level, and a situation level will be described. The importance parameter is calculated by the following procedure.
(1) Importance parameter calculation according to at least one of defense, attack, and situation for each event
(2) Calculate the importance parameter for each event by team
[0062]
(1) Parameter calculation based on events
FIG. 6 is a diagram showing a list of correspondences between the importance level parameters of the attack level, the defense level, and the situation level, and the events used to obtain the value of each importance level parameter. Each importance parameter is obtained for each event in the rightmost column of the table, and one event is referred to only for obtaining one kind of importance parameter, and is not used for obtaining a plurality of importance parameters. The obtained importance parameter is stored in the metadata DB 105 as attribute information of the corresponding event.
[0063]
The event includes an event that makes the calculated importance parameter value positive and an event that makes it negative. For example, the attack level is raised by the occurrence of an event advantageous to the attacker such as goodbye, reversal, and preemptive. It is also lowered by events that are advantageous to the defensive side, such as overrun and defensive defensive. The defense level is increased by an event advantageous to the defender such as triple play, triple strikeout, and good catching ball, and is lowered by occurrence of an event advantageous to the attack side such as an error.
[0064]
The situation level is calculated based on the attribute information attached to the event information of the first event (usually the start of a bat) in each digest scene. The attribute information is, for example, information on the runner who is on the heel at the start of the batting and information on the batting order of the batter (whether it is cleanup or lower batting). The video content analysis unit 107 refers to the attribute information and sets the situation level higher when the score or the possibility of reversal is high.
[0065]
According to the processing described above, the importance of each event belonging to the digest scene can be determined. In addition, it is possible to determine how important each digest scene is from the viewpoint of attack or defense. Furthermore, from the viewpoint of the situation, it is possible to determine the degree of highlight of the digest scene (so-called good scene).
[0066]
(2) Calculate the importance parameter for each event by team
Next, the video content analysis unit 107 performs the importance level for each event based on the importance level parameter corresponding to at least one of the defense level, the attack level, and the situation level calculated in units of events by the processing of (1). Calculate parameters by team. The importance parameter for each event team is determined as follows.
[0067]
The video content analysis unit 107 sets the attack level of the event occurring on the digest scene as it is to the attack side of the team on the attack side in the digest scene. On the other hand, for the defensive team, the attack level of the event is a value obtained by multiplying the attack level of the team on the attacking side by -1. Similarly, the defense level of the event occurring on the digest scene is set to the attack level as it is for the team on the defense side in the digest scene. On the other hand, for the attacking team, a value obtained by multiplying the defensive level of the team on the defensive side by -1 is set as the defensive level of the event. As described above, the attack level and the defense level of each event are calculated for each team.
[0068]
The situation level for each team is calculated as the importance parameter of the start event of the digest scene. The situation level will not change, regardless of the team's attack and defense.
[0069]
According to the above processing, it is possible to determine the importance of each event for attack, defense, and situation for each team.
[0070]
Further, in the present embodiment, the attack level of each event is based on an event relating to a registered player (registered player) when a specific player is registered as a target instead of a team as in the above-described embodiment. The defense level and the situation level can be calculated for each player.
[0071]
The defense level, attack level, and situation level for each player are set so that, for example, if the registered player is involved in a goodbye or reversal event, the attack level is increased, and the attack level of the registered player is multiplied by -1, and the opponent player It can be calculated by setting the attack level. In addition, when an event occurs such that the registered player takes the opponent out, or strikes out, the defense level is set to increase, and the defense level of the opponent player is set by multiplying the defense level of the registered player by -1. Can be calculated. Further, the situation level can be calculated based on an event when the batter at the start of the batting of the digest scene is a registered player.
[0072]
Scene selector
The scene selection unit 111 is configured to select a digest scene based on a target and an event including the target registered in the preference information registration unit 109, and the degree of distribution desired. For this reason, the scene selection unit 111 calculates the importance of each digest scene for each team for each user. The importance of each digest scene for each team and each user is calculated as follows.
[0073]
  The scene selection unit 111 according to the present embodiment uses the event registered in the preference information registration unit 109 in addition to the importance parameter for each event calculated by the video content analysis unit 107 for each team. The importance for each user is calculated. That is, the attack of the team Tj of the digest scene Si for the user who registered the event to be viewed, Defensive, situationThe importance parameter (Si.Tj.Pk) is calculated by the following equation.
  Si. Tj. Pk = Func (Ek, Tj, Pk, Fav_event)
  Si: digest scene
  Tj: Team identifier (home team or away team)
  Pk: Attack, Defensive, situationAn identifier indicating the level
  Ek: A set of events belonging to the digest scene Si
  Fav_event: Event you want to see
  Func: A set of events belonging to the digest scene Si, a team identifier (team ID), elements of importance parameters to be calculated (attack, defense, situation), and a type name of the event that the user wants to seeargumentAs a function to calculate the importance parameter corresponding to any of the attack, defense, and situation of the digest scene Si for each user and each team
[0074]
  Next, the scene selection unit 111 determines the importance parameter (corresponding to any one of attack, defense, and situation) of the digest scene Si calculated for each user and each team, the team registered in the preference information registration unit 109, and The degree of importance of each digest scene is calculated for each user based on whether the user likes or dislikes the registered team and the degree of desire for distribution. Since the calculated importance is directly used for selecting a digest scene constituting the personal digest, it is also referred to as a final importance. The formula for calculating the final importance is as follows.
  Final importance of scene Si =
      (Fav_offense * flag * Si.Tj.Pk (attack level))
    + (Fav_defense* Flag * Si. Tj.Pk (Defense level))
    + (Fav_chance* Flag * Si. Tj.Pk (status level))
  flag targetPreference items that are eventsFlag that indicates whether you like or dislike, like: 1, dislike: -1, default: 1
  Si video scene
  Tj Target identifier
  Fav_offense Degree of desirability of aggressive element delivery
  Fav_defense Degree of desirability of defensive elements
  Fav_chance Degree of delivery of situational elements
[0075]
The scene selection unit 111 calculates the importance of each digest scene according to the above formula. Then, the digest scenes are ranked according to the calculated importance, and a personal digest for the user who has requested distribution is created using the digest scenes at the top of the ranking.
[0076]
As described above, the target is set to a specific player, and Si. Tj. If the importance parameter for each element such as each user of the digest scene Si, each player, attack, defense, and situation is substituted into the item of importance parameter (Pk), this embodiment determines the final importance of a specific player. Can be calculated.
[0077]
The configuration described above operates as follows. That is, the digest scene generation unit 101 inputs the sports video S and creates a digest scene that is an average video of about 15 seconds. The created digest scene is accumulated in the moving image DB 103. In addition, information indicating a plurality of events that occur on each created digest scene is accumulated in the metadata DB 105 as metadata.
[0078]
The video content analysis unit 107 reads the event information stored in the metadata DB 105 and obtains the importance parameter P1 of each event. The importance parameter P1 of each event may be set in advance. Further, it may be calculated using a preset arithmetic expression. Further, the video content analysis unit 107 calculates the importance parameter P2 of each event for each team using the importance parameter P1 of each event. Both the importance parameter P1 and the importance parameter P2 are stored in the metadata DB 105.
[0079]
By the way, in the preference information registration unit 109, information related to the user's preference is set and registered in advance by the user himself from the UIF of the mobile terminal. When creating a personal digest at the timing when the user wishes to deliver a personal digest, the scene selection unit 111 receives a delivery request and requests a preference item related to the preference of the user who requested delivery from the user DB 113. Desired event) and preference attributes (whether they like or dislike the object, the degree of desire for distribution, etc.) are read out.
[0080]
Then, the scene selection unit 111 calculates the importance parameter of the digest scene for each user and each team based on the information f and the importance parameter P2. Subsequently, the final importance parameter is calculated based on the calculated importance parameter of the digest scene for each user and each team. Furthermore, the digest scene is ranked based on the importance parameter, and a personal digest is created and distributed to the user who has requested distribution.
[0081]
FIG. 7 and FIG. 8 are flowcharts for explaining a personal digest distribution method performed in the personal digest distribution apparatus described above. FIG. 7 is a flowchart for explaining a process performed before creating a personal digest, and FIG. 8 is a flowchart for explaining a method for creating a personal digest based on user preferences.
[0082]
As shown in FIG. 7, the personal digest distribution apparatus first determines whether or not a sports video has been input (step S701). If no sports video is input (step S701: No), the system waits until a sports video is input. When a sports video is input (step S701: Yes), the digest scene generation unit 101 generates metadata together with the digest scene (steps S702 and 703). The digest scene and metadata generation process is continued until all of the sports video is processed. The generated digest scene is stored in the moving image DB 103, and the metadata is stored in the metadata DB 105.
[0083]
The digest scene generation unit 101 determines whether the generation of the digest scene and metadata has been completed (step S704). If it is determined that the process has not ended (step S704: No), the generation of the digest scene and metadata is continued until the process ends.
[0084]
If the digest scene generation unit 101 determines that the generation of the digest scene and metadata has ended (step S704: Yes), the video content analysis unit 107 includes an event (represented by metadata) included in the digest scene. Is calculated (step S705). Then, an importance parameter for each team is calculated for each event from the importance parameters calculated for each event (step S706). The importance parameter P1 for each event (event unit) and the importance parameter P2 for each event unit for each team are stored in the metadata DB 105 together with the metadata.
[0085]
Next, as shown in FIG. 8, the scene selection unit 111 determines whether there is a user's desire to distribute a personal digest (step S801). As a result of the determination, if the distribution request is not made (step S801: No), the process waits until the distribution request is made.
[0086]
Further, the scene selection unit 111 acquires the preference information of the user who desires distribution from the preference information registration unit 109 (step S802). Then, the importance level parameter P1 and the importance level parameter P2 stored in the metadata DB 105 are read, and the importance level of each digest scene is calculated for each user who wishes to distribute by team (step S803). Subsequently, the final importance of each digest scene is calculated for each user based on the calculated importance of the digest scene for each team (step S804). Further, the digest scenes are ranked in descending order of importance, and a predetermined number is selected from the highest ranked ones to create a personal digest (step S805). The scene selection unit 111 distributes the created personal digest to the user who desires distribution, and ends the process (step S806).
[0087]
Note that a program for causing a computer to execute the personal digest distribution method of the present embodiment is an installable or executable file that can be read by a computer such as a CD-ROM, floppy disk (FD), or DVD. Recorded on a recording medium and provided. In addition, a program for causing a computer to execute the personal digest distribution method may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network.
[0088]
【The invention's effect】
  BookThe inventionAbout videoUser wants to deliverUsers want to distribute preference items and preference itemsDegree of videoAt least one element of attack, defense, situationRegister everyAnd calculates the importance parameter of each scene included in the video according to at least one element of attack, defense, and situation, and the importance calculated based on the importance of each scene for each userBased on the scene in the videoPersonal digest selectionUser'sIt is possible to calculate the importance of each scene reflecting the preference items to be distributed and the degree of distribution desired with a relatively simple calculation, and create a personal digest that better matches the user's wishes. And deliveryThere is an effect that can be done.
[Brief description of the drawings]
FIG. 1 is a block diagram for explaining a personal digest distribution apparatus according to an embodiment of the present invention.
FIG. 2 is a diagram showing attributes specified by the user on the UIF in order to register the degree of desired distribution of preference items according to the embodiment of the present invention.
FIG. 3 is a diagram for explaining an event according to the embodiment of this invention;
FIG. 4 is a diagram for explaining a relationship between a digest scene and an event according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating characteristics of attack level, defense level, and situation level according to an embodiment of the present invention, and calculation timing;
FIG. 6 shows a list of correspondences between the importance level parameters of the attack level, the defense level, and the situation level according to an embodiment of the present invention and the events used to calculate the value of each importance level parameter. It is a figure.
FIG. 7 is a flowchart for explaining a personal digest distribution method according to an embodiment of the present invention;
FIG. 8 is another flowchart for explaining the personal digest distribution method according to the embodiment of the present invention;
[Explanation of symbols]
100 Personal digest creation part
101 digest scene generator
107 Video content analysis unit
109 Preference information registration part
111 Scene selection section
103 Movie DB
105 Metadata DB
P1 Importance parameter
P2 importance parameter
S sports video

Claims (3)

A personal digest distribution device for creating and distributing a personal digest for a user who has requested distribution from a video composed of a plurality of continuous scenes,
A target registration means for registering a preference item that is a target or event that the user wishes to distribute with respect to the video;
Regarding the preference item registered in the target registration means, a desired degree registration means for registering the degree of user's desire for distribution for each of at least one of the attack, defense, and situation of the video,
Video analysis means for calculating an importance parameter of each scene included in the video according to at least one element of attack, defense, and situation;
The importance of each scene for each user, and the following is calculated by the equation, it selects the scene of the image as the personal digest based on the calculated significance scene selection means,
Importance = (Fav_offense * flag * Attack element importance parameter)
+ (Fav_defense * flag * defensive element importance parameter)
+ (Fav_chance * flag * situation element importance parameter)
However, Fav_offense: Degree of desirability of distributing aggressive elements
Fav_defense: Degree of desirability of distributing defensive elements
Fav_chance: Degree of distribution of situational elements
flag: A personal digest distribution device comprising a flag indicating whether a user likes or dislikes a preference item . Personal digest distribution that creates and distributes personal digests for users who have requested distribution from video composed of multiple consecutive scenes, and includes a preference information registration unit, video content analysis unit, and scene selection unit A personal digest distribution method in an apparatus,
Receiving a registration of a preference item that is an object or event that a user desires to distribute with respect to the video by the preference information registration unit;
With respect to the registered preference items, a step of accepting registration of the degree that the user desires distribution by the preference information registration unit for at least one element of attack, defense, and situation of the video;
Calculating an importance parameter of each scene included in the video according to at least one element of attack, defense, and situation by the video content analysis unit;
Calculating the importance of each scene for each user by the scene selection unit, and selecting a scene of the video as the personal digest based on the calculated importance;
Importance = (Fav_offense * flag * Attack element importance parameter)
+ (Fav_defense * flag * defensive element importance parameter)
+ (Fav_chance * flag * situation element importance parameter)
However, Fav_offense: Degree of desirability of distributing aggressive elements
Fav_defense: Degree of desirability of distributing defensive elements
Fav_chance: Degree of distribution of situational elements
flag: Flag indicating whether the user likes or dislikes the preference item
A personal digest distribution method comprising: A program for causing a computer to execute a process of creating and distributing a personal digest for a user who has requested distribution from video composed of a plurality of continuous scenes,
A target registration function for registering a preference item that is a target or event that a user wishes to distribute with respect to the video;
Regarding the preference item registered in the target registration function, a desired degree registration function for registering the degree to which a user desires distribution for at least one element of attack, defense, and situation of the video,
A video analysis function that calculates an importance parameter of each scene included in the video according to at least one element of attack, defense, and situation;
The importance of each scene for each user is calculated by the following formula, and a scene selection function for selecting a scene of the video as the personal digest based on the calculated importance,
Importance = (Fav_offense * flag * Attack element importance parameter)
+ (Fav_defense * flag * defensive element importance parameter)
+ (Fav_chance * flag * situation element importance parameter)
However, Fav_offense: Degree of desirability of distributing aggressive elements
Fav_defense: Degree of desirability of distributing defensive elements
Fav_chance: Degree of distribution of situational elements
flag: Flag indicating whether the user likes or dislikes the preference item
That causes the computer to execute the program.

Images