3D Image Flow Visualization System for Inter-Media Trend Analysis

We propose a novel 3D visualization system for exploring temporal changes in trends using image flows in multiple medium. Our use of media has changed dynamically in the last decade; mass and social media affect each other. It is important to compare how multiple medium are affected by real-world events and how each medium affects other media. Media provides many representative images, such as scenes of accidents and disaster, the design of products and commercial pictures, to explain the reality of events without text. It is important for analyzing media and society to trace and explore such images flowing on the media. Our system visualizes flows of images extracted from one or multiple medium in a 3D space. We arrange histograms of images related to multiple topics from different types of media in the 3D space by stacking them on timelines to explore changes in trends in each topic, and compare differences in exposure among topics and medium. We implement application systems using proposing visualization system on a huge blog archive and a news video archive, and report the usefulness of our system by using various exploration examples.

■Case Study 1: Visual Trends in Social Media 


Example of image flow visualization using histograms of images that visualizes changes in trends in political and social issues for given query related to ``Prime Minister Hatoyama''. We can read stories about “Prime Minister Hatoyama” by exploring the movements of topics.

■Case Study 2: Inter-Media Event Exploration on TV and Blogs 


This example is visualization system for intermedia events that displays images from TV on the Image Flow View and bloggers’ activities on Event View to analyze society. Image Flow View that represents histograms of images from TV corresponding to four key people related to “nuclear power plant”, and line charts that represent frequencies of their appearance on blogs to compare different types of media. EventView shows detail about events on May 9, 2011 related to “Prime Minister Kan” (shown at root of tree structure) extracted from blog archive.

■Case Study 3： Image Flows Visualization for Inter-Media Comparison 


We extract blog image clusters from our six-year blog archive and search for similar TV shots in each cluster from a broadcast news video archive by using image similarities. We then visualize such image flows on a timeline in 3D space to visually and interactively explore time sequential changes in influences among media resources and differences and/or similarities between them such as topics that become popular on only blogs or that become popular on blogs earlier than on TV.


Extracting image cluster related to construction of nuclear power plant.




多メディア間の話題探索のための時系列画像3次元可視化システム 


Web上の画像や放送映像を含む多メディア画像用いた時系列画像3次元可視化システムを提案する． 近年，Webメディアは，従来型のマスメディアの影響を受けると同時に，マスメディアへ影響を与えるようになってきており，社会事象を分析する際には複数メディア間の話題の広がりを分析することが重 要となってきている． また，これらのメディアでは，テキストの代用として積極的に映像・画像を用いることで，文章だけでは伝えきれない，その時々の話題および興味を視覚的に伝えており，社会分析において，これらの映像・画像情報を追跡した分析は不可欠となる．提案するシステムでは，Webメディアおよび放送映像から抽出された様々な話題に関する時系列画像群を画像ヒストグラムとして3次元空間に可視化することで，流行の推移，トピック間の違い，メディア間の関係等を視覚的に探索可能にする新たな可視化システムを実現した．実際に，本システムを，大規模なブログアーカイブとニュース映像アーカイブに適応したアプリケーションをいくつか紹介し，様々な探索事例を示すことで提案システムの有用性を示す．



■References


  • 伊藤正彦、豊田正史、喜連川優, 多メディア間の話題探索のための時系列画像3次元可視化システム，情報処理学会論文誌：データベース，Vol. 8, No. 1, (TOD65), pp. 27-44，Mar. 2015
  • Masahiko Itoh, Masashi Toyoda, Cai-Zhi Zhu, Shin’ichi Satoh, and Masaru Kitsuregawa, Image Flows Visualization for Inter-Media Comparison, In Proceedings of IEEE Pacific Visualization (PVis 2014), pp. 129-136, Feb.2014
  • Masahiko Itoh, Masashi Toyoda, Cai-Zhi Zhu, Shin’ichi Satoh, and Masaru Kitsuregawa,Visualization Framework for Inter-Media Comparison using Image Flows, In Proceedings of IEEE Conference on Visual Analytics Science and Technology (VAST 2013), Oct. 2013, (poster)
  • Masahiko Itoh, Masashi Toyoda, and Masaru Kitsuregawa, Visualizing Time-Varying Topics via Images and Texts for Inter-Media Analysis, In Proceedings of the 17th International Conference on Information Visualization (IV2013), pp. 568-576, Jul. 2013
  • Masahiko Itoh, Masashi Toyoda, Tetsuya KAMIJO, and Masaru Kitsuregawa, Visualizing Flows of Images in Social Media, In Proceedings of IEEE Conference on Visual Analytics Science and Technology (VAST 2012), pp. 129-130, Oct. 2012, (poster)
  • 上條哲也，伊藤正彦，豊田正史, CGM画像の時系列トレンド可視化システム, 第73回情報処理学会全国大会論文集, Mar. 2011


登録: 投稿 (Atom)