SEGMENTASI DOKUMEN DENGAN MENGGUNAKAN BACKPROPAGATION NEURAL NETWORK

Abstract

The development of knowledge in the field of artificial intelligence is highly developed, such as research on segmentation of documents. Segmentation of documents makes the user becomes easy to obtain the required information. Segmentation is to divide an image into several regions based on the similarity of the form or object. Segmentation can also means, an image recognized into small blocks and recognized into a kind whether text or not. The design of this application that have utility as a maker of computer programs that can automatically segmenting types contained in the mixture and recognize text document or image automatically as well, make the document easier to be taken and analyzed its contents. Segmentation this document is based on a block by using backpropagation and histogram smoothing for smoothing the histogram. Input form document image obtained with the help of the scanner with JPEG and BMP file formats. After testing by using the Automatic Cropping, it is known that the accuracy rate of 82.1% for the first size is 1048x195 and 58.7% for the second size is 1048x52. While testing using Windowing process, known accuracy rate of 21%. The introduction of more accurate to type text rather than images.

Key words

Backpropagation, Histogram, Segmentation, Smoothing