Doktorarbeit: Distanzmaße in der Bildähnlichkeitsanalyse – Neue Verfahren und deren Anwendung im Marketing

Distanzmaße in der Bildähnlichkeitsanalyse – Neue Verfahren und deren Anwendung im Marketing

Schriftenreihe innovative betriebswirtschaftliche Forschung und Praxis, volume 397

Hamburg , 276 pages

ISBN 978-3-8300-7781-7 (print)
ISBN 978-3-339-07781-3 (eBook)

about this book deutsch english

The importance of images in marketing has always been high and is still increasing because of the growing number of digital images uploaded to the Internet. Millions of photos are uploaded to social networks like Facebook or Twitter every day. But this flood of data has not been used for marketing approaches yet. Because there are no systematic, feature-based methods for analysing image databases in context of marketing.

The requirement for classifying huge image data bases, in a way that is useful for marketing, is the implementation of a similarity analysis. The results of such an analysis depend mostly on the used distance measures. Besides of the quality of the clustering results also the computation time and the robustness with regard to the image quality are important.

This work verifies which distance measures are appropriate to classify huge image data bases with regard to robustness, computation time and clustering results with the aim to make them usable for marketing approaches. Therefore a lot of empirical studies had to be implemented and are described in this book, like a Monte Carlo analysis, a confusion test, and many feature-based image similarity analyses. There was one algorithm that was especially robust against typical failures in the digital photography, like noise or overexposure: the `increasing radius algorithm´, which is an approximation of the well-known `earth mover’s distance´ for image similarity analysis.

There are a lot of applications in marketing for which the image similarity analysis could be relevant in the future. This book especially shows how the method of using visual image information could be used for analysing brand positioning and brand confusion. But the obtained findings can also be used in areas like lifestyle segmentation, customer segmentation, searching for campaign images, new product development, or to distinguish buyers from non-buyers. It is also very likely that there will be more images used for questionnaires in the future.

Ihr Werk im Verlag Dr. Kovač

Bibliothek, Bücher, Monitore

Möchten Sie Ihre wissenschaftliche Arbeit publizieren? Erfahren Sie mehr über unsere günstigen Konditionen und unseren Service für Autorinnen und Autoren.

Nach oben ▲