Journal ID : TRKU-10-03-2020-10491
[This article belongs to Volume - 62, Issue - 02]
Total View : 174

Title : Handwritten Security Modeling based on Cosine and Inner Product Method

Abstract :

Research aimed to identify the characteristics of the owner of the handwriting, so that handwriting can be recognized and able to prevent forgery of handwriting through two methods, namely the inner product and cosine. The increased use of handwriting as a means for authentication and authorization from accessing an account or specific media, it takes a technological advancement in terms of security handwriting. Both method will be compared to the level of accuracy in recognizing the characteristic and the owner of the handwriting. Steps being taken to 100 consists of digitized handwritten documents, pre-processing such as color conversion and thresholding, and gridding stage, feature extraction, and pattern recognition through the inner product and cosine. From the test results which consists of 60 training data (in database) and 40 test data through methods cosine, 9 data misidentified, 10 unknown data, and 21 data has correctly identified the owner of the handwriting, so accuracy is obtained only by 52.5%.While testing with the same data through inner methods, produced 27 recognized correctly managed data, 13 data is not in accordance with the owner's handwriting, so accuracy is obtained from the inner product method amounted to 67.5%. Overall it can be concluded from this study, the use of methods of cosines and inner product in identifying the owner of the handwriting can produce fairly good accuracy, where accuracy inner product value is still higher than the cosine

Full article