Recent techniques of Extraction and Recognition Textual Information from Natural Scene: Review

Abstract

With the growing reputation of the digital camera and embedded visual devices, the text extraction commencing natural scene pictures has become the key issue that is considered to alter the real time applications. Text data contain the natural scene pictures for the analysis of the videos, indexing and recovery process.   Normally, the textual data in the scene pictures present the high level semantic data regarding the picture and background which are helpful for proper understanding of the pictures. However, an effective text extraction method has been a challenging approach at present years because  of different font-style, text magnitude, alignment and complicated background. To overcome this issue, text extraction method for complicated video scene pictures has been proposed in this paper.  Text extraction from video analysis is done through various phases; pre-processing, feature extraction and combing of the key-frames. A comprehensive survey has been done on the extraction of the natural scene pictures. In addition, various generalized texts, images has been descried with figures.  Moreover, different method of the text extraction has been elaborated along with advantages and disadvantages.  The methods includes region, texture, color, and hybrid based and its sub types has been described.