<![CDATA[The signature is an important attribute for each individual because it is often used as an identity. The use of signatures is practical and simple to make the existence of the more commonly used signature. The existence of this signature facilitate the activities of individuals and even used for the identification of individuals. It is proof that the signature is an important attribute that must be protected from those who are not responsible. Sophisticated and valid needed to provide the best solution. Various approaches have been proposed in the development of the identification of signatures aimed to minimize counterfeiting signatures. This study will discuss the identification of signatures to get authenticity. Processes that exist in this study consists of two main parts: training and testing phase. The size of the imagery used is 256x256 pixels. Training phase, the image subjected to several processes that threshold, Daubechies wavelet transformation, normalization, and then will be trained using the Artificial Neural Network (ANN) Backpropagation. Testing has the same phase as in the training phase but the end of the process will be a comparison between the image data that has been stored with the image comparison. Optimal results are obtained by using a neural network has two hidden layers, respectively 20 and 10 nodes, Daubechies 3 wavelet transformation at level 4, and the learning rate of 0.13. With the results of an accuracy of 93.33%.]]>
Copyrights © 2016
