Image Processing Applied to Eye Segmentation in Cheese Maturation

Eyes in Swiss type cheeses are a quality attribute, therefore they should have a certain shape, size and distribution. Here we present a simple pre-processing method and the evaluation of three segmentation algorithms applied to cheese quality control by detecting and segmenting eyes in optical images of cheeses. Algorithms employed were: circle detection using Hough transform, thresholding and contour detection, blob detection and watershed. Segmentation results were contrasted with the same images, but previously segmented by an expert. Results showed that up to 75 % of eyes in cheese images were correctly detected and segmented using the proposed procedure.