Article Per Year (5 Year)
p-Index From 2020 - 2025
0
P-Index
This Author published in this journals
All Journal Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Rizka Husnun Zakiyyah
Fakultas Ilmu Komputer, Universitas Brawijaya
Computer Science & IT Control & Systems Engineering Education Electrical & Electronics Engineering Engineering

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Pengenalan Citra Tanda Tangan Off-Line dengan Pemanfaatan Ciri Centroid Distance Function Rizka Husnun Zakiyyah; Agus Wahyu Widodo; Fitri Utaminingrum
Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer Vol 1 No 10 (2017): Oktober 2017
Publisher : Fakultas Ilmu Komputer (FILKOM), Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1060.86 KB)

Abstract

A person's signature is one of the most valid proof that shows ownership of documents and transactions that contain their most important data. However, the process of analizing its authenticity is still done manually. To resolve this problem, an image recognition system for signature will be developed by applying characteristic centroid distance function. This Image recognition process begins with preprocessing, such as binerisasi, filtering, cropping, resizing, and thinning. Next the position of pixels will be searched to store all the foreground pixels and centroid pixels of the image. All pixels stored distance will be calculated using centroid function and grouped according to the amount of features that were selected so that each group has the same amount of data. The average of centroid distance function will be counted on every group so that each group will generate one feature. The results of feature extraction will be processed with the k-nearest neighbor classification method. On the research that has been done the highest accuracy obtained from extraction characteristics of centroid distance function uses 20 class is 88.5% obtained from 20 features and k= 1 with the amount of 10 and 14 training data for each class. The highest accuracy to 50 class is 67.4% obtained from 15 features and k= 3 with 10 and 14 training data for each class.
Co-Authors Agus Wahyu Widodo Utaminingrum, Fitri