Front Inner Page - Volume 2 No.3 June 2015

  • » Back to Index

  • Title

    :

    Robust Scene Text Recognition with Sparse Belief Propagation

    Authors

    :

    S.Priyadharsini1, Mrs. J.Ranjani2

    Keywords

    :

    Scene Text Detection, Maximally Stable External Regions (MSERs), Single-Link Clustering, Distance Metric Learning. 

    Issue Date

    :

    June – 2015

    Abstract

    :

    Content-based image analysis task requires text detection in natural scene images. The character candidates in text detection on minimizing regularized variations termed as Maximally Stable Extremely Regions (MSERs), which are extracted by using a fast and effective pruning algorithm. The single-link clustering algorithm groups character candidates into text candidates. In this grouping, candidates automatically learn the distance weights and clustering threshold. The text classifier used in this grouping performs two operations. They are elimination of non-text probabilities and identification of texts. MSER (Maximally Stable Extremely Regions) efficiently identifies the text and eliminate the non-text area. Forward and Backward algorithm is used to read the different orientation of the text. The retrieved text is converted into audio format by using the classic text to speech engine interface. The audio format which is very useful in various situation in the real world. Then convert the retrieved text into an audio format which is useful in various situations in the real world. Proposed system going to implement audio converted text in the reserved area which is a very useful application for the blind people. 

    Page(s)

    :

    1-5

    ISSN

    :

    2347- 4734

    Source

    :

    Vol. 2, No.3, June 2015

    Download

    :


  • » Back index