Content Based Document Image Retrieval with Support Vectors Clustering

The goal of this paper is representing a suitable approach to content based document image retrieval. in proposed algorithm a feature vector is extracted with wavelet transform for sub-words. then based on this features, sub-words are clustered with support vector clustering (SVC) algorithm, then this approach is used for searching based on keyword in content based document retrieval problem. The retrieval algorithm achieved a precision of 81.6% at a recall rate of 74.2% on a database of 20 pages.