Automatic fruits identification system using hybrid technique

• Main Authors: Aibinu, Abiodun Musa, Salami, Momoh Jimoh Emiyoka, Shafie, Amir Akramin, Hazali, Norazlanshah, Termidzi, N.
• Format: Conference or Workshop Item
• Language: English
• Published: 2011
• Subjects: S Agriculture (General)
• Online Access: http://irep.iium.edu.my/1762/1/Automatic_Fruits_Identification_System_Using_Hybrid_Technique.pdf; http://irep.iium.edu.my/1762/; http://ieeexplore.ieee.org/search/srchabstract.jsp?tp=&arnumber=5729570&queryText%3Daibinu%26openedRefinements%3D*%26filter%3DAND%28NOT%284283010803%29%29%26searchField%3DSearch+All
• Institution: Universiti Islam Antarabangsa Malaysia

In this work, a combination of artificial neural network (ANN), Fourier descriptors (FD) and spatial domain analysis (SDA) has been proposed for the development of an automatic fruits identification and sorting system. Fruits images are captured using digital camera inclined at different angles to the horizontal. Segmentation is used for the classification of the preprocessed images into two non-overlapping clusters from which shape boundary and signatures are estimated using FD and SDA technique. Furthermore, color information obtained from the extracted red-green-blue color components of the fruits images during ANN training process is used in accurately detecting the color of such a fruit. The two independent paths are then combined for fruits sorting and identification purposes. The performance of the developed hybrid system has been evaluated at three different angles of camera inclination from which an accuracy of 99.1% was obtained.