IDENTIFYING UNDAMAGED AND DAMAGED PEANUT KERNELS USING TRISTIMULUS VALUES AND SPECTRAL REFLECTANCE

Peanut kernels from grade samples were sorted into damaged and undamaged categories based measurements from four optical sensing methods. A monochromatic machine vision system, a contact colorimeter, a non-contact colorimeter and a spectrophotometer correctly classified 62.7%, 79.1%, 85.8%, and 94.1%, respectively, of the damaged kernels while correctly classifying 100% of the undamaged kernels. Further analysis of the spectro-photometer data resulted in 99% correct classification of damaged kernels when critical spectral reflectance ranges and line slopes at specific wavelengths were used.