Author : Bharath Vyas B 1
Date of Publication :7th May 2016
Abstract: Text is an important form of information. Any information in the form of text present in a document-image or video, is difficult to be modified, if the text is static in it. Hence, modification or analysis of a text is possible only by separating it from a document-image or a video. This project deals with an efficient method of isolating text present in a video, by using a Laplacian operator. The document image is convolved with Laplacian operator to highlight text regions present in the document image. The pixels related only to text (textual pixels) need to be stressed upon, and isolated from other non-textual pixels. This is achieved by computing a gradient-difference amongst the neighborhood pixels. Clusters of texts are required, in order to differentiate between textual and non-textual pixels. To identify text cluster from non-text cluster, the mean of both the cluster is computed, by employing K-means clustering technique. When this method is employed, it results in the mean of the text cluster possessing a higher value than that of a non-text cluster. In order to prune the text data contained in the identified text blocks, for each candidate text block, the corresponding region in the Sobel edge map of the input image undergoes projection profile analysis to determine the boundary of the text blocks. At the end of the process, false positives regarding the textual clusters are removed by employing geometrical properties-based empirical rules. Experimental results on the standard document image database collected from ICDAR-2003 dataset show that the Laplacian operator based text detection method is able to detect text of different fonts, contrast and backgrounds.
Reference :
-
[1] J. Zang and R. Kasturi, “Extraction of Text Objects in Video Documents: Recent Progress” The Eighth IAPR Workshop on Document Analysis Systems (DAS2008), Nara, Japan, September 2008, pp 5-17.
[2] J. Zhang, D. Goldgof and R. Kasturi, “A New EdgeBased Text Verification Approach for Video”, ICPR, December 2008, pp 1-4.
[3] K. Jung, K.I. Kim and A.K. Jain, “Text information extraction in images and video: a survey”, Pattern Recognition, 37, 2004, pp. 977-997.
[4] A.K. Jain and B. Yu, “Automatic Text Location in Images and Video Frames”, Pattern Recognition, Vol. 31(12), 1998, pp. 2055-2076.
[5] M. Anthimopoulos, B. Gatos and I. Pratikakis, “A Hybrid System for Text Detection in Video Frames”, The Eighth IAPR Workshop on Document Analysis Systems (DAS2008), Nara, Japan, September 2008, pp 286-293.
[6] M. R. Lyu, J. Song and M. Cai, “A Comprehensive Method for Multilingual Video Text Detection, Localization, and Extraction”, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 15, No. 2, February 2005, pp 243-255.
[7] C. Liu, C.Wang and R. Dai, “Text Detection in Images Based on Unsupervised Classification of Edge-based Features”, ICDAR 2005, pp. 610-614.
[8] E. K. Wong and M. Chen, “A new robust algorithm for video text extraction”, Pattern Recognition 36, 2003, pp. 1397-1406.
[9] P. Shivakumara, W. Huang and C. L. Tan, “An Efficient Edge based Technique for Text Detection in Video Frames”, The Eighth IAPR Workshop on Document Analysis Systems (DAS2008), Nara, Japan, September 2008, pp 307-314.
[10] Trung Quy Phan, Palaiahnakote Shivakumara, Chew Lim Tan, “A Laplacian method for video text detection” 2009 10th International Conference on Document Analysis and Recognition, Barcelona, Spain, July 2009, pp 66-70.