A method of content based fax retrieval

According to the characteristics of fax paper, a method of content-based fax retrieval is put forward. The wavelet decomposition is used in the fax image de-noising, and the wavelet coefficients are used for regional division, and then combined with optical character recognition technology to extract the semantic features of the text region and the shape features of the logo for the fax retrieval. The accuracy of retrieval is improved notably. At the same time, the algorithm of features fusion for retrieval is given. Experiments show that the method has high precision and recall, which is more effective than retrieval methods based on manual annotated semantic.