Cereal Chem 69:257-261 | VIEW
Description of the Textural Appearance of Bread Crumb by Video Image Analysis.
D. Bertrand, C. Le Guerneve, D. Marion, M. F. Devaux, and P. Robert. Copyright 1992 by the American Association of Cereal Chemists, Inc.
A method for the characterization of the apopearance of bread crumb from digital images is proposed. Experimental breads were prepared by varying the nature of the surfactants added to the flour to change the textural appearance of the crumbs. Seven treatments were tested. The trials were repeated several times (three loaves for each treatment, four slices for each loaf, two images for each slice), giving 142 digital images. The textural features were extracted from images by an original mathematical procedure based on two-dimensional Haar transform. This procedure made it possible to characterize each digital image of crumb by a vector that included 66 numbers called texture characteristics. The relevance of the method was tested by applying discriminant analysis on the matrix of texture characteristics to identify the bread treatment from the digital image of its crumb. About 80% of the images were correctly identified by discriminant analysis, both in the training and in the verification set. However, the effect of loaf-to-loaf variation in breadmaking was important. The method can be applied without modification to any collection of textured samples.