Classification of Broken Maize Kernels Using Artificial Neural Network-Assisted Image-Processing Approach
2023
Pareek, C.M. | Singh, N. | Tewari, V.K. | Dhruw, L.K. | Singh, H.D.
PURPOSE: The mechanical damage that occurs during harvesting, handling, and processing operations deteriorates the quality of seeds and reduces their commercial value. Therefore, the classification of mechanically damaged seeds in the seed lot is one of the most important seed quality assessment tasks in the seed industries. The conventional method of seed quality assessment in terms of mechanical damage is time-consuming and susceptible to human error. Therefore, a fast and effective seed quality assessment technique needs to be developed. METHODS: In the present study, an intelligent seed quality assessment method based on the artificial neural network-assisted image-processing approach was proposed for the real-time classification of the broken maize kernels from the whole maize kernels. Initially, the images of seed samples were captured, and the image-processing operations were performed to extract the morphological features of maize kernels. Afterward, the ant colony optimization (ACO) algorithm was applied to select the superior features. Finally, the multilayer perceptron neural network (MLPNN)-based classifier was used to identify the broken and whole maize kernels. RESULTS: The best classification performance was obtained with a 5–10-5–1 MLPNN classifier. The testing results revealed that the classification accuracy of 91.85% with an average processing time of 0.14 s per kernel was achieved with the proposed image-processing-assisted ANN classifier. CONCLUSION: Hence, the proposed method appears promising and could be employed to design the real-time automated seed grading or sorting machines.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par Korea Agricultural Science Digital Library
Découvrez la collection de ce fournisseur de données dans AGRIS