Research Of Image Text Positioning And Extraction Technology

With the rapid development of information technology and network technology,multimedia has become an important way in information bearing and sharing.Digital image character recognition technology is one of the most important content in image information retrieval and analysis currently,and has been widely applied in the field of robot vision,license plate recognition,network filtering,automatic real-time processing of bills,etc.In this thesis,the image text positioning technology and extraction technology is analysis and researched on the base of collation and analysis of relevant technical and theoretical in image character recognition,and the corresponding image positioning and text extraction algorithm is also designed and checked under Matlab 5.0 in this thesis.Firstly,this thesis analyzes the color space model theory,digital image processing technologies and the basic processes,and the technical difficulty of image character recognition.Secondly,on the basis of theoretical analysis,this thesis designed the image text localization algorithms,including edge detection,linearization processing,morphological processing and noise removing,etc.Thirdly,on the basis of the positioning of text in the image,this thesis designed images and text extraction and recognition algorithms,including text tilt correction,character segmentation,normalization,and text feature extraction and recognition.Finally,the algorithms designed in this thesis are realized under the Matlab5.0 environment.And from the analysis of the test results of the algorithms designed in this thesis,they are able to respond to the needs of conventional digital image text recognition application environment.