Video Search Systems based on Overlay Text

International Journal of Artificial Intelligence and Knowledge Discovery

B B Meshram¹  Avinash N. Bhute¹ 

[1] Veermata Jijabai Technological Institute (VJTI), Matunga, Mumbai-19

关键词: Accuracy;    Bagging;    Genetic Algorithm (GA);    Naive Bayes (NB);    Sentiment Mining;    Support Vector Machine;   

DOI  :  

学科分类：建筑学

来源: RG Education Society

【 摘 要 】

Text Information present in images and video which is useful for automatic annotation, indexing, and structuring of images. Regions of text are identified within the frames of the videos. Traditionally, Optical Character Recognition (OCRs) are used to extract the text within frame or videos, but OCRs has many constraints in term of language, font, size, orientation, style, alignment, etc.  In this paper, we have proposed the approach that enables search based on text information presented in videos. The process is divided into two steps. Firstly, we build the metadata labels by detecting and extracting the text/overlay from videos. Secondly, these metadata labels are used for searching the videos from database. The Search is based on image-matching i.e the overlays that are extracted from images or videos are stored in database and matched for similarity according to query.

【 授权许可】

Unknown