An Upgraded Rubber Tree Seed Clones Identification Model with Wavelet Coefficient

The paper presents an upgrading process of rubber tree seed clones identification model using image processing techniques. Sample of rubber tree seeds are captured using digital camera where the RGB color image are processed involving segmentation algorithm which includes thresholding and morphological technique. Texture patterns from seed clones images are then analysed through wavelet´s Daubechies D4 algorithm which produced discrete frequency coefficients representing the extracted features. Previous work only utilized three statistical parameters representing these coefficients such as mean, variance and standard deviation as the inputs for designing an intelligent identification model for various rubber tree seed clones. However, the accuracy was not that convincing. This work has proposed to use seven input parameter in order to improve the model´s accuracy. In this work, 285 sample images representing three types of rubber tree seed clones are used to train Artificial Neural Network (ANN) with Levenberg Marquardt algorithm. Two models are being designed, known as Model 1 and Model 2, to identify seed clones RRIM2005 and RRIM2009 respectively. The outcomes have shown that both models´ accuracy has improved but not that substantial.