Image Analysis

Bread structure is a determinant of loaf volume and resilience, as well as the sensation perceived during eating, and its analysis comprises the study of the fluid phase (air), also named as void, cells or pores, and the solid phase of bread crumb (cell wall material). Acquisition of two-dimensional images by flatbed scanning is the most commonly employed method to perform image analysis of bread as it is fast, easy to use, economical, robust, independent of the external light conditions, and accurate. To enhance the contrast between the two phases, once acquired, the image is subjected to cell segmentation, a process that separates or classifies objects of interest from its background, typically yielding a binary image, set at a specific threshold. Thresholding is a critical step in ensuring a successful partition of crumb cells from the background and assumes that the object and background pixels can be distinguished by selection of an optimal gray level value. Hundreds of segmentation techniques are described in the literature, yet the Otsu method, based on an algorithm that minimizes the intraclass variance of the segmented region, or manual thresholding, is those most reported. Once the binary image is obtained, the software processes it, and crumb grain properties such as number of cells, number of cells/cm2, mean cell area, cell-total area ratio, and cell wall thickness can be calculated.