A Hybrid Method for the Classification of Paddy Varieties Based on Image Segmentation

In Seed Processing Plant, different paddy varieties are identified and classified manually by visual inspection which is a tedious and less accuracy process. To overcome this, we developed an automated system for identifying and classifying the five different varieties such as ADT-38, ADT-39, ADT-43, ASD-16, and TKM-9 based on their morphological features.  In this paper, three phases are involved. In First Phase, Image of paddy seeds are acquired by a digital camera, which is collected from seed unit and the acquired image is stored in JPEG format. The stored image is given to wiener filter for Image pre-processing such as edge detection, image denoising and image enhancement. In Second Phase, Image Segmentation is performed using Dual Tree Complex Wavelet Transform (DTCWT). In Third Phase, from segmented image features are extracted and given to Neural Network to classify the paddy varieties according to the extracted color features, morphological features and shape factors of each paddy grain.