Handwritten Document Security System with Inner Product and Shape based Feature Extraction Method

Security of handwritten document in present is one of important things, because many crimes against the falsification of document is growing. For example, in the case of signature forgery or falsification of land certificate. This researche was conducted with the aim of designing an application system which can recognize handwritten of the owner of the document through the characteristics of shape, so that the falsification of documents can be prevented. The method applied in this study consisted of writing stage analog to digital conversion, pre-processing, automatic segmentation into the size of 500x200 pixels every data word by word and 10 grids, feature extraction stage, and the percentage calculation of similarity through the similarity measures and inner product method. From the research that has been conducted on 100 documents of 20 owners handwriting, 72 documents identified the owner managed appropriately through matching the features of the 12 words in each document, namely “The”, “You”, “Will”, “ To”, “He”, “And”, “It”, “Is”, “Are”, “His”, “Have”, “For”. So that the percentage of accuration, precision, and recall obtained against the document security system that is equal to 72%, 72.8%, and 67.4%. General Terms Data Mining and Pattern Recognition.