2010 | OriginalPaper | Buchkapitel
A Vertical Search Engine Based on Visual and Textual Features
verfasst von : Kun Wu, Hai Jin, Ran Zheng, Qin Zhang
Erschienen in: Entertainment for Education. Digital Techniques and Systems
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
It is very difficult for general image search to get higher retrieval accuracy and relevancy with thousands of irrelevant image results. Vertical search engine can collect web information from multiple and different resources in specific domain, and provide more professional and individualized image retrieval services for various users in their domain. A new approach is to combine textual and visual features to search content-related images from Internet. The new topic identification and hybrid segmentation are proposed to effectively improve retrieval accuracy and coverage. A new SVM-based classification with RBF kernel integrates visual and textual features to implement better classification to enhance image retrieval accuracy. The proposed methods are tested with Pet Image Search Engine (PISE) database. Numerous experiments demonstrate the superior performance of developed PISE system, which is attractive in practical applications.