Wheat classification using image analysis and crush-force parameters
Abbreviated Journal Title
wheat; digital imaging; machine vision; pattern recognition; multivariate analysis; hardness; DISCRIMINATION; KERNELS; Agricultural Engineering
A study was conducted to develop methodology for wheat classes and variety identification by combination of image analysis techniques with wheat hardness physical measurements. Wheat kernel morphometrical parameters were extracted from digitized images and hardness parameters were obtained from force-deformation curves from a single kernel wheat characterization system which also provided a kernel weight. Pattern recognition methods were applied to the data base of combined parameters for wheat kernels of six classes and seventeen varieties of soft and hard wheat. Recognition rates for parameter combinations of shape, size, and hardness scores were higher than hardness or imaging alone or when combined with weight. Hard and soft recognition rates of 94% was achieved with shape and hardness of the wheat kernels A PC version of the developed algorithm was written and tested with the same data set. Satisfactory performance in the PC version confirmed the practicality of the method developed.
Transactions of the Asae
"Wheat classification using image analysis and crush-force parameters" (1996). Faculty Bibliography 1990s. 1812.