Red-Green-Blue (RGB) colour density as a non-destructive marker in sorting deteriorated lentil (Lens culinaris Medik.) seeds

Progressive seed coat browning in lentil (Lens culinaris Medik.) seeds may indicate decreasing quality with advancing deterioration. A computer - aided image analysis system has been developed to analyse Red-Green-Blue (RGB) colour components in two dimensional lentil seed images. The method is based on measuring the RGB values and the corresponding calculated medium RGB index (i.e., the average of the three distinct component values) of intact seeds, which are then separated into three fractions each having a different RGB range. Sorted seeds with medium RGB values below 98 had the lowest germination percentage with increasing abnormal seedling percentage at early ageing stages, whereas sorted seeds having RGB values over 125 had the highest germination and increasing abnormal seedlings only at the end of the deterioration period. Seeds sorted in the range between 98 and 125 values of medium RGB index showed an intermediate decreasing germination. This trend was confirmed also by the time courses of mean germination time and leachate electrical conductivity. Linear regression lines between medium RGB index values and seed quality parameters of non sorted and sorted aged seeds have high correlations coefficients, suggesting that the progressive and continuous deterioration process can be monitored with colour marker change. The operative easiness, low cost of commercial computer technology and non-destructive seed analysis and sorting highlight the potential of this technique for application in a seed laboratory.