Robust text detection and recognition in blurred images

Abstract

In this paper, we propose a novel method for detecting and recognizing the text from the blurred images. Text detection in natural scenery images is an important issue in the processing stage. All the previously proposed methods use different algorithms to detect text in images; however, they suffer from poor performance while performing detection in blurred images. The proposed algorithm is capable of removing blur with an iterative deconvolution method and a linear invariant filter. The proposed method can achieve detection and recognition of the text with a time complexity of 4.53 s. Experiments show our method achieves a better text detection than the other existing methods.