Tamil text extraction using dual tree complex wavelet transform and a weighted hybrid thresholding approach

Abstract

In recent times, Extraction of text and caption from images and videos is important and in great demand for video retrieval, annotation,indexing and content analysis. A challenging problem for character recognition applications is text extraction from images, due to language characteristic, complex background and unknown text color. The application areas are vast. There are numerous text extraction techniques existing for extracting text from the images. In spite of availability of some really good text extraction algorithms, there is still a need to develop more efficient ones. The present work describes a novel and an efficient text extraction scheme for text extraction using Dual Tree Complex Wavelet Transform and a Weighted hybrid Thresholding Approach for text binarization. The proposed text extraction scheme is designed in such a way as to preserve the finer details in the text. The theory and algorithm for extracting the text from images are established in detail. The methodology developed along with all the algorithms are presented and documented. It is demonstrated that the proposed method achieved reasonable accuracy of the text extraction for moderately difficult examples. The present work considers the Tamil Text extraction from images. Different level of Dual Tree Complex Wavelet Transform is used for Tamil Text extraction. The text from the input images is extracted. The text can be either scene text or superimposed text. Superimposed text is the text which embedded in the image after the image is captured. The Dual Tree wavelet Transform is used to extract the text from the images and a weighted hybrid thresholding method is used for text binarization. In text extraction, the text is extracted along with the background. The background is removed using text binarization. Text extraction involves two steps, filtering and morphological dilation.