Summary of stages in the developed epidermis segmentation algorithm. Figure 2 summarizes the main stages in the algorithm to segment the epidermis from a digital image of an H&E stained skin section. The initial steps of sample segmentation, image cropping and colour normalization process the image so that subsequent steps are more efficient and robust. The colourspace conversion, contrast enhancement and linear combination steps enhance the contrast between the epidermis and the rest of the tissue; this makes the thresholding step that follows more effective. Morphological processing fine tunes the segmentation of the binary image created by thresholding. The object classification step removes objects in the image which do not satisfy criteria set out for objects that are part of the epidermis, the user interaction step allows a human operator to change the classification of objects performed in the previous step.