Complex Context Of The Color Images Embedded In Text Analysis And Extraction

With the development of multimedia technology, the new standard MPEG-7 semantic content-based retrieval. It refers to the semantic object in accordance with the media, the characteristics of search. As the video is a very rich amount of information data, content-based video retrieval research has become the current hot spots. Therefore the text extraction and recognition from color image has a more practical value: with current retrieval technology, it is possible for the content-based retrieval to reach a conceptual level.The text extraction and recognition from color image under the complex background is the quite more difficult than that under sample background. As the position of text and the background image are indefinite, it directly influences the effect of image process.The thesis includes: Summing-up the research results on text extraction in home and abroad in recent years. Carrying on the analysis and contrasting to each concrete algorithm. And based on gradation picture edge examination Robert operator, the full area length is used to consider various components of each pixel color value comprehensively. Then, adding the determination by statistics, the real text block will be identified. Finaly, based on the text block spatial frequency characteristic, the work that text extraction and recognition from color image under the complex background will be realized.The radon images have been chosen to define that the algorithm the thesis proposed performs well the text extraction and recognition from color image under the complex background.