Research On Fax Classification And Retrieval Based On User Telephone Number

The paper belongs to a research topic. With deep research on new theories and technologies of locating telephone number regions and designing the fax classification and retrieval system based on user telephone number, the purpose is to meet the urgent need of some agents, improve the efficiency and accuracy, with important theoretical meaning and application values.Firstly, the paper presents the study and implementation on the region location algorithm of user telephone number, based on the key words region features. A labeling algorithm based on line-labeling and region-growing is adopted; features of horizontal traversing times and spatial distribution of connected components are extracted. The judgment algorithms on multi-class Support Vectors Machine are studied and the telephone number regions location is carried out based on the key words region features and normal decision tree, which demonstrates good robustness on facsimile layout and picture quality by experiments.Secondly, the paper presents the location algorithm's study and implementation based on key words recognition. Recursive projection multi-tree is adopted to implement layout segmentation effectively; a character segmentation strategy based on word fragment recognition is introduced to overcome the character sticking influence on key words recognition. The paper realizes the location of telephone number regions by means of feature extraction of projection on characters, character recognition and key words search, which demonstrates to own comparatively good location performance when the fax is in good picture quality by experiments.Finally, an amalgamation algorithm based on character segmentation characteristic to locate the telephone number regions is introduced, according to analysis on the adaptability of two location algorithms. In addition, through analyzing the form of telephone numbers in faxes, the paper presents a judgment algorithm of fax numbers based on special punctuation detection, with realization of integral telephone number recognition. A fax classification and retrieval system based on user telephone number is designed, achieving comparatively ideal performance.