Summary: | (A-C) Morphological analyses were conducted on SEM backscatter images. BMU size and coverage were analyzed in the same image (A,C), taken after oxidation of the organic matrix by immersion in 10.5 vol% H 2 O 2 for 20 minutes (image A magnified for visual clarity). Images for pore morphometry (B), in contrast, were taken in a polished and chemically untreated shell slab. (D-F) Binary images (objects of interest = white; remainder = black) used for the automated image segmentation process. Individual BMUs (D; here shown in various colors to allow discrimination of the individual entities) and pores (E, white) were recognized by the machine learning–based image segmentation process and separated from the remainder of the images (black). (F) For the calculation of BMU coverage a threshold based on the average gray value of the image series was applied. Values above this threshold were assigned to the crystalline (white) phase, those below the threshold to the inter-crystalline phase (black).
|