Digital images are made up of pixels, each of which may have one of 256 levels of intensity. The SEM images are monochrome as they reflect the electron or X-ray flux at each pixel. They can be pseudo-colored to improve visual discrimination of constituents through changes in image intensity. Image analysis is actually composed of two tasks; (1) image processing to enhance details, feature differences, and isolate constituent phases and, (2) image analysis to perform measurements on those features. Measurements may include area fraction, surface area, size, and distribution.
Fig. 6 shows high-magnification BE and XR images of polished cement particles embedded in an epoxy. The individual element density maps for Ca, Si, AI, Fe, S, K, Mg, are displayed with the backscattered electron image in the lower-right corner.
The first step in processing uses filtering operations to reduce random noise in the X-ray images. The median filter is useful to accomplish this while retaining edge details. The BE image noise is generally low enough that random noise poses fewer problems for segmentation. Given this, emphasis for phase segmentation is given to this image and the X-ray images are used only when necessary.
Fig. 6. SEM, X-ray and BE (lower right) images of a cement. Field width, 150 µm.