The Improved Corner Detection for Video Text Positioning Algorithm

Abstract:

This article mainly aims at the problem of the video text font size, achieved a algorithm of multi-scale corner text detection, and combined with the characteristic of word usually has the same color, even more precise positions the text area with color clustering way.This algorithm not only detect the game scene, video ads and video news kind of information in words, but also be used for the natural scene of the text of the detection positioning.Finally, test a public data sets, and the test results show that the proposed method can detect and positioning video the text in the complex information.

• Periodical: Applied Mechanics and Materials (Volumes 263-266)
• Pages: 1523-1526
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.263-266.1523
• Online since: December 2012
• Authors: Yan Hai Wu*, Jia Xin Li, Fang Ni Zhang
• Keywords: Color Clustering, Corner Detection, Multi-Scale Decomposition, Text Location

References

[1] Wang Yong,Zheng Hui,Hu Deiwen, the technology of getting words from image and video[J],China journal of image and graphics, No.5(2004), p.532~538.

[2] Li Nianyong,Liang Yanmei,Zhang Shu,Yang Li, Chang Shengjiang, the text positioning of complex color image based on BP neural network[J], photon journal,Vol.38, No.10(2009), p.155~157.

[3] Wang Yuzhu，Yang Dan，ZhangXiaohong．Robust corner detection based on mult iscale curvature product in B-spline scale space[J]．Acta Automatica Sinica, Vol.23, No.4(2007), pp.414-417．

DOI: 10.1360/aas-007-0414

[4] Lucas S M．ICDAR 2005 text locating competition results[C]. International Conference on Document Analysis and Recognition (2005)．

DOI: 10.1109/icdar.2005.231