Klasifikasi kelompok usia berdasarkan citra wajah menggunakan algoritma neural network dengan fitur face anthropometry dan kedalam kerutan
| Main Author: | Hayatin, Nur |
|---|---|
| Format: | Article PeerReviewed Book |
| Bahasa: | eng |
| Terbitan: |
Fakultas Teknik Unipdu Kompleks Ponpes Darul 'Ulum Peterongan Jombang
, 2016
|
| Subjects: | |
| Online Access: |
http://eprints.umm.ac.id/58185/19/Peer%20review%20-%20Hayatin%20-%20Age%20Prediction%20Face%20Ratio%20Neural%20Network%20Wrinkle.pdf http://eprints.umm.ac.id/58185/18/Similarity%20-%20Hayatin%20-%20Age%20Prediction%20Face%20Ratio%20Neural%20Network%20Wrinkle.pdf http://eprints.umm.ac.id/58185/20/Hayatin%20-%20Age%20Prediction%20Face%20Ratio%20Neural%20Network%20Wrinkle.pdf http://eprints.umm.ac.id/58185/ http://www.journal.unipdu.ac.id/index.php/teknologi/article/view/577 |
Daftar Isi:
- Age classification is one of the research topics related to the prediction of age based on facial image. The problems associated with age groupings based on the image of the face is how to choose the right facial features, that will affect the final result grouping. This study aims to classify age based on facial image by using the important features, that is face anthropometry and wrinkles. Wherein the wrinkles features that used are wrinkles density and the depth of wrinkles. The research methodology consists of four stages: preprocessing, identification of the face point location , feature extraction and classification. The face point is identified based on facial symmetry and the difference of pixel intensities. While wrinkles was obtained from the combined edge detection method using Sobel operator and histogram equalization. The algorithm used for the classification process is a Neural Network (NN) algorithm that would classify the input image data into three classes, there are children, young and old. The final results of test-ing show that the proposed method was able to categorize age based on facial image fairly well with the results of the test accuracy of 65% with epochs = 1000, and the error rate = 0.0095, 100 iterations