Handwriting Recognition Using Eccentricity and Metric Feature Extraction Based on K-Nearest Neighbors

The process of handwriting recognition aims to digitize handwriting without having to retype handwriting. This study proposes the extraction of Eccentricity and roundness features based on K-Nearest Neighbors (KNN) to recognize handwriting. To maximize the process of recognition is done several stag...

Full description

Saved in:
Bibliographic Details
Published in2018 International Seminar on Application for Technology of Information and Communication pp. 411 - 416
Main Authors Hari Rachmawanto, Eko, Rambu Anarqi, Galang, Moses Setiadi, De Rosal Ignatius, Atika Sari, Christy
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.09.2018
Subjects
Classification
Classification algorithms
Eccentricity
Feature extraction
Handwriting recognition
Image segmentation
KNN
Measurement
Roundness
Testing
Training data
Online AccessGet full text
DOI10.1109/ISEMANTIC.2018.8549804

Cover

More Information
Summary:The process of handwriting recognition aims to digitize handwriting without having to retype handwriting. This study proposes the extraction of Eccentricity and roundness features based on K-Nearest Neighbors (KNN) to recognize handwriting. To maximize the process of recognition is done several stages of preprocessing stages such as thresholding, noise filtering, and cropping stage to take every character on handwriting image. Each image that has been through the preprocessing process is further extracted features. Eccentricity and metric are the features of the extracted form of each image to be recognized. After obtained the two features are then carried out the training process to group the feature form on each type of letter. Next, KNN is used to classify each letter character. Based on the results of testing the proposed method for recognition of handwriting characters obtained an accuracy of 85.38%.
DOI:10.1109/ISEMANTIC.2018.8549804