Defect sorting of dry dates by image analysis

Spec tral characteristics were obtained for good quality and defective fruit of two dry date varieties, 'Majhul' and 'Zahidi'. An image process ing algorithm was developed to sort good and defective fruit using simple thresholding operations. To eliminate the effects of varying system characteristics, defective date spectral characteristics were normalised based on an average 'good date' reference spectrum. Thresholding parameters were determined by integrating the normal ized reflectance curves over the wavelength range in which defect/good date contrast was maximised. The reflectance data were also used to select optical filters to enhance the features of interest in the images. Preliminary results indicate that the method proposed offers a promising means of rapidly detecting defective fruit. Further research is needed to estimate sampling errors associated with the technique and to investigate real-time sorting speed.