|
چکیده
|
Nutritional properties and simple accessibility caused to pay attention to medicinal plants for long time. Therefore, the present study has been carried out to recognize and classify Falcaria vulgaris, Pelargonium sidoides, Trigonella foenum-graecum, Origanum vulgare, Rumex acetosa as most popular medicinal plants Iran. The plant images were acquired by smartphone vision system. A novel robust automatic image processing algorithm was designed for fast identification of medicinal plants under controlled illumination condition. The algorithm was implemented to extract texture, color, and shape features from the acquired images. Artificial neural networks were applied to classify various groups of the studied medicinal plants and the efficient classifier was selected based on error, correlation, and accuracy. The efficient features were feed to the model and the optimum classifier model was obtained with 28− 10-6 structure. The accuracy of the model was 100 % with correlation coefficient and mean square error of 1.00 and 2.35 × 10− 12, respectively. The proposed algorithm has enough potential for ease and accurate classification of the medicinal plants.
|