A Survey On Video Scene Text Deblurring Using Text-Specific Multiscale Dictionaries

Abstract

This paper is concerned with the study of scene text detec tion and recognition from blurry natural video scene i.e. blurred images extracted from video which is imp ortant for image related purposes. In digital imaging,it�s a challenging task to restore a clear image fro m a single motion-blurred image due to camera shake. E.g. includes image retrieval,comparative studies of different images,as well as the tool that aid visually impaired individuals to access the pictorial i nformation. Particularly when we use handheld cameras to capture natural scene images,a general probl em,i.e.,blur,frequently happens. There are several techniques available for text detection and recogni tion. For representation of the text and non-text fields,a string of text-specific multi-scale dictionaries (TMD) and a natural scene dictionary is used separately. The TMD-based text field reconstruction make s easier to deal with the different scales of strings in a blurry image effectively. A lot of work has been done for detecting text in images and a lot has to be done. In this survey,we extend an existing end- to-end solution for text detection and recognition in natural images to video https://www.ijiert.org/paper-details?paper_id=140426