Video Text Detection using Image Edge and Sparse Representation
Text is a very concise and accurate clue for describing the contents in video frames, in this paper, we propose a novel classification method, which utilize the discriminative nature of sparse representation to perform classification for text detection task. At first, the edge map is obtained by using gradient information. And then the potential text regions are located by morphology operation and projection profile analysis on binirized edge image. Finally discriminative sparse representation is introduced to refine those potential regions, which are achieved by casting the classification problem as finding the sparsest representation of individual potential text region in terms of two different dictionaries. Experimental results show that the proposed text detection method is robust to text fonts,sizes,colors and appearances,and also achieves improvements in terms of detection accuracy.
text detection classification discriminative nature edge image sparse representation
Xiaoyan Qi Shutao Li
College of Electrical and Information Engineering Hunan University Changsha, China
国际会议
The 10th International Conference on Intelligent Technologies(第十届智慧科技国际会议 InTech09)
桂林
英文
415-419
2009-12-12（万方平台首次上网日期，不代表论文的发表时间）