Text Image Retrieval Based on Picture Information Measurement and Model-KNN

Various shortages to understand high-level semantic feature of data information exist in computer technology which makes it difficult for computers to retrieve document image based on semantic feature directly. The capability of document image retrieval algorithm depends much on suitable abstraction of statistical feature and the selection of classifier. To this problem, the normalized generalized picture information measurement (NPIM_K) is introduced as the statistical feature. Meanwhile, an improved KNN classifier based on model is used to identify to which species one image belongs. Experimental results show that the document image retrieval algorithm based on NPIMK and mode-KNN is effective