Numerical analysis of scale morphology to discriminate between atlantic salmon stocks

The use of an image analysis microsystem with a specially developed software enabled scale image processing, outline extraction and computation of features which are not available in the conventional approach of scale analysis. Those parameters (shape factors, moment invariants and elliptic Fourier coefficients) were introduced into a discriminant analysis process to test their usefulness in stock identification. The method was applied to European stocks originating from Norway (Etne river) and France (Elorn river). The low misclassification rate obtained indicates that this approach is particularly promising.