RECOGNITION OF OVERLAPPING PARTICLES IN GRANULAR PRODUCT IMAGES USINGSTATISTICS AND NEURAL NETWORKS

Image analysis can be used to characterize granular populations in man y processes in the food industry or in agricultural engineering. Eithe r global or individual parameters cart be extracted from the image. Ho wever, granular products may agglomerate on the image, bringing bias m easurements of individual parmeters: products which agglomerate have t o be recognized. This is done by a combination of image analysis (to p re-process and extract features), statistical methods (to reduce infor mation) and neural network techniques (to take decisions).