Fruit Recognition Using Surface and Geometric Information

Authors

  • Joel C. De Goma School of Information Technology, Mapúa University, Makati 333 Sen. Gil Puyat Ave., Makati City 1200, Philippines.
  • Cyrus Arlan M. Quilas School of Information Technology, Mapúa University, Makati 333 Sen. Gil Puyat Ave., Makati City 1200, Philippines.
  • Maurice Albert B. Valerio School of Information Technology, Mapúa University, Makati 333 Sen. Gil Puyat Ave., Makati City 1200, Philippines.
  • Jon Jake P. Young School of Information Technology, Mapúa University, Makati 333 Sen. Gil Puyat Ave., Makati City 1200, Philippines.
  • Zaliman Sauli School of Microelectronic Engineering, Universiti Malaysia Perlis, Pauh Putra Campus, 02600 Arau, Perlis, Malaysia.

Keywords:

Fruit Recognition, Co-Occurrence Features, Pre-Processing, Feature Extraction, K- Nearest Neighbor,

Abstract

One of the interesting topics in image processing and computer vision is Fruit Recognition. The computer vision strategies used to recognise fruits rely on four basic features which are colour, texture, size and shape. In fruit recognition, unrecognised fruit images are caused by different factors. These factors are different illuminations, specular reflections, and different poses of each fruit, variability on the number of elements, and cropping or occlusions. This paper proposes and aims an efficient and effective way to recognise fruits regardless of the said factors by combining the four basic features of the fruit. Fruit recognition involves different processes which are pre-processing, feature extraction, recognition and testing. The recognition is done using the K-Nearest Neighbor based on statistical values of the colour moments, Gray Level Cooccurrence Matrix (GLCM) features, area by pixels for the size and shape roundness. The fruit images comprised of 2633 fruit images from 15 different kinds of fruits. The authors tested different classifiers which are KNN, Naïve Bayes, Decision Tree, and bagging to know what best fits for the images. After testing the classifiers based on the 2633 images, results showed that KNN outperformed the other classifiers. The result showed that combining all the features namely colour, texture, size and shape, the overall recognition rate for all classifiers has increased and it has shown the best output.

Downloads

Download data is not yet available.

Downloads

Published

2018-05-30

How to Cite

De Goma, J. C., Quilas, C. A. M., Valerio, M. A. B., Young, J. J. P., & Sauli, Z. (2018). Fruit Recognition Using Surface and Geometric Information. Journal of Telecommunication, Electronic and Computer Engineering (JTEC), 10(1-15), 39–42. Retrieved from https://jtec.utem.edu.my/jtec/article/view/4043