Thresholding, Dilation, Erosion, Opening, and Closing

Thresholding

converts a grayscale image to a binary black and white image.

Erosion

removes pixels around the periphery of features in the image, making lines thinner. Because small features disappear during erosion, counting the features that have disappeared can give an approximate estimate of the size distribution.

Dilation

is the converse of erosion, and adds pixels around edges of features. This is commonly performed after an erosion or skeletonization to prevent pixels from disappearing.

Skeletonization

locates the center of features, converting a figure into something resembling a line drawing, where the lines depict centers or `skeletons' of the figures. This is useful, for instance, in metallurgic analysis. Usually must be followed by dilations to obtain solid lines.

Opening

is an erosion followed by a dilation, and opens up spaces between just-touching figures.

Closing

is an dilation followed by an erosion, and closes up breaks in features. Frequently, several repetitions of openings and closings are performed to achieve a desired effect.