Text size
  • Small
  • Medium
  • Large
Contrast
  • Standard
  • Blue text on blue
  • High contrast (Yellow text on black)
  • Blue text on beige

    Which Entry is More Similar? A Non-linear Visualisation of Query Results in Image Retrieval and Image Recognition Problem

    Electronic Visualisation and the Arts (EVA 2017)

    London, UK, 11 - 13 July 2017

    AUTHORS

    Joanna Gancarczyk & Anna Olszewska

    ABSTRACT

    http://dx.doi.org/10.14236/ewic/EVA2017.14

    Content based image retrieval (CBIR) has been a subject of exploration in digital humanities since 1990’s (Gudivada 1995). Various descriptors were implemented to represent shape, texture and colour content of the image as sequences of numerical values (Zhang & Lu 2004, Veltkamp, Latecki 2006, Zha & Yang 2010). At the same time similarity measures and learning algorithms were designed to enable efficient image classification and retrieval (LeCun 1998). The issue, however, remains in a simple question: which descriptor and which similarity measure best reflects the human perception of similarity of visual objects? And is this the same one, that best responds to the ground truth in a retrieval query?

    In this paper, we move for a while away from the very technical issues of shape descriptors definition and verification and we focus on the question how visualisation of the computed data affects the final result of a visual query. We are replacing a traditional, linear presentation of n most similar outputs to a set of graph-like and scatterplot based visualisation modes. The research study is performed on a particular example of visual search in large databases of historical watermarks, trademarks and monograms.

    We believe, that the approach to search across a digital print room repository involving intuitive user interaction is a step toward fully making use of its potential. We state, that a modern interface, that allows the end user an intuitive navigation through options and partial results is a milestone on the way to fill a technological gap between users familiar with image processing issues and with computer science background and those for whom obtaining an answer for a particular research question is worth more, than understanding how the result was actually computed. Finally, we proof the concept with a proposition of a shape descriptor followed by a set of flexible interfaces designed to display and navigate through the results of a visual query.

    PAPER FORMATS

    PDF Icon PDF Version of this Paper 1,006(kb)