Video text information plays an important role in semantic-based video analysis, indexing and retrieval. It is observed that the detection of texts in video remains as a challenging task due to its complex varying conditions. In this paper, we present a study on local features based text detection in document images and more focus is provided for text detection based on Laplacian method. The document image is convolved with Laplacian operator to filter the document image. Then the maximum gradient difference value is computed for each pixel to generate threshold. Based on the computed threshold, a binarized frame is obtained which highlights the text block. The candidate text block regions are further verified and refined that is, the corresponding region in the Sobel edge map of the input image undergoes projection profile analysis to determine the boundary of the text blocks. Finally, empirical rules are employed to eliminate false positives based on geometrical properties. In addition, a comparative study of the Laplacian method with a novel text detection and localization method based on Corner response and Multi scale edge based method for video text detection is made. The techniques are evaluated on documents taken from ICDAR 2003 robust reading and text locating database. Experimental results show that the Laplacian method is able to detect texts of different fonts, contrast and backgrounds. To give an objective comparison of the Laplacian approach, we have used detection rate and false positive rate as decision parameters and metrics.

1. TrungQuyPhan, PalaiahnakoteShivakumara and Chew Lim Tan," A Laplacian Method for Video Text Detection", IEEE DOI vol. 10, 2009, pp. 153.
2. Li Sun, Guizhong Liu, XuemingQian, DanpingGuo," A Novel Text Detection And Localization Method Based On Corner Response", Pattern Recognition, vol. 37, 2004, pp. 595–608.
3. Ching-Tung " Embedded-Text Detection and Its Application to Anti-Spam Filtering", IEEE Trans. on Pattern Analysis and MachineIntelligence, vol. 10, 2008, pp. 910-918.
4. XuemingQian, Guizhong Liu, Huan Wang, Rui Su," Text detection, localization, and tracking in compressed video", Signal Processing: Image Communication, vol. 22, 2007, pp. 752 – 768.
5. 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, February 2005, pp. 243-255.
6. C. Liu, C. Wang and R. Dai, "Text Detection in Images Based on Unsupervised Classification of Edge-based Features", ICDAR, 2005, pp. 610-614.
7. E. K. Wong and M. Chen, "A new robust algorithm for video text extraction", Pattern Recognition , vol. 36, 2003, pp. 1397-1406.
8. 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.
9. Qixiang Ye*, Jianbin Jiao, Jun Huang, Hua Yu," Text detection and restoration in natural scene images", J. Vis. Commun. Image R, vol. 18, 2007, pp. 504–513.
10. A. K. Jain and B. Yu, "Automatic Text Location in Images and Video Frames", Pattern Recognition, Vol. 31, 1998, pp. 2055-2076.
11. P Shivakumara, Weihua Huang, TrungQuyPhan, Chew Lim Tan," Accurate video text detection through classification of low and high contrast images", Pattern Recognition, vol. 43, 2010, pp. 2165–2185.
12. Qixiang Yea,*, QingmingHuangb, Wen Gao, Debin Zhao," Fast and robust text detection in images and video frames"', Image and Vision Computing , vol. 23 , 2005, pp. 565–576.
13. B H Shekar and M SharmilaKumari"Text Detection in Video Frames: An integrated approach based on weber's local descriptor and differential excitation difference"
14. V. Y. Mariano and R. Kasturi, "Locating Uniform-Colored Text in Video Frames", 15th ICPR , Vol. 4, 2000, pp 539-542.
15. Datong Chen?, Jean-Marc Odobez, Herv/ e Bourlard," Text detection and recognition in images and video frames", Pattern Recognition, vol. 37, 2004, pp. 595–608.

Laplacian Operator Corner Response Multi-scale Edge. Text Detection Text Localization