EKSTRAKSI CIRI FEATURE POINT UNTUK PENGENALAN TANDA TANGAN DENGAN MENGGUNAKAN BACKPROPAGATION NEURAL NETWORK

Abstrak

In this research, the main goal is to make an application to recognize people signature. To make this, we need the method to do our research. Backpropagation Neural Network that is used to train image sample and Feature Point Extraction for feature extraction. This method work with center point of the pixel image. The pre-processing are image binarization, Median filter, Thinning image with Zhang Suen algorithm, cropping, and  scaling for training image.. The method  for feature extraction is feature point extraction. This method works with counting the centeroid of the image. Number of total point centeroid is decide by depth value. The process will stop if the total point centroid is achived. After feature extraction, it will continue to train the sample image using Backpropagation Neural Network.

Key words

Backpropagation, Featurte Point Extraction, Image Binarization, Median Filter, Thinning, Threshold.