Digital image texture analysis for bread crumb grain evaluation

Abstract

Baking and milling companies and scientific laboratories evaluate crumb grain as a bread quality parameter. A digital imaging technique was used to evaluate bread crumb grain, and the potential of image texture analysis for crumb grain assessment was evaluated. Images of slices of two commercial broad brands were digitized and stored in 512 X 512 pixel format. Eighteen image texture features were extracted. Multivariate discriminant analysis using patterns of the image texture features differentiated two brad brands (BRRA and BRDI). One hundred percent of BRRA and 97.5% of BRDI subimages (128 X 128) from the middle area of a slice were correctly recognized. Technological factors and location of subimages on a slice affected the value of image texture features as crumb grain vow across a slice. Variations in the crumb grain within a slice were studied, and a ranking scale was developed for evaluating the degree of coarseness of crumb grain in boundaries of 64 X 64 pixel The developed scale has the potential to be used for quality control in a breadmaking operation. A whole slice was scored by the ranking scale, which was developed by determining the percentage of fine or coarse subimages present in a slice. Bread slices with coarse texture of crumb grain were distinguished from slices with more smooth crumb grain. The ranking of the slices by these variables coincided with visual judgment. The scale developed is flexible and can be easily adapted to meet various users' requirements