Binary Image Processing
Binary images are results of segmentation of original image binary presentation of the phase of interest that contains features and the background. Binary representation of images allows simple analysis of features of interest while disregarding background information. There are many algorithms that operate on binary images to correct for imperfect segmentation.
- Boolean Logic
- Feature Based Boolean Logic
- 'Hole' Filling
- Erosion and Dilation
- Pruning and Convex Hull
- Watershed Separation
Boolean Operations Examples
There are plethora of uses for Boolean operators on binary images and also in combination with gray-scale images. Examples include coating thickness measurements, stereological measurements, contiguity of phases and location detection of features.
A simple way to represent Boolean logic is by using a truth table, which shows the criteria that must be fulfilled..
A or B
A X or B
A and B
A not or B
A not X or B
A not and B
A and not B
Feature Based Boolean Logic
Feature based Boolean logic operates on entire features when determining whether a feature is ON or OFF.
Feature-based logic uses artificial features, such as geometric shapes, and real features, such as grain boundaries to ascertain information about features of interest.
Hole filling is a common tool that removes internal "holes" within features. For example, one technique completely fills enclosed regions of features using feature labeling.
There is no limit how large or tortuous a shape is. The only requirement for hole filling is that the hole is completely contained within the feature.
Binary image of particles with "holes'
'Hole' filing of particles
'Hole' filling based on size
Inverted binary image
Features <10 pixels eliminated (red)
Binary image with small "holes' filled
Erosion and Dilation
Erosion/dilation is the removal and/or addition of pixels to the boundary of features based on neighborhood relationships. The basic eroded point is dilated until the edge of the dilating feature touches another dilated feature, leaving a line of separation (watershed line) between touching features.
Dilation - Counting Examples
The dilation counting technique considers near neighbors and is very useful to characterize particle dispersion. However, it does not provide local area fraction measurements, which only can be obtained using the tessellation technique. Dirichlet tessellation is the most comprehensive technique to characterize particle dispersion.
Erosion - Counting Examples
Feature count: 547 831
Feauture count: 546 350
Pruning and Convex Hull
The convex-hull can be used to fill concavities and smooth very jagged skeletons or feature peripheries. Basically, a convex-hull operation selectively dilates concave feature edges until they become convex.
Watershed transformations are iterative processes performed on images that have space-filling features such as grains. The enhancement usually starts with the basic eroded point or the last point that exists in a feature during successive erosions, often referred to as ultimate eroded point. The basic eroded point is dilated until the edge of the dilating feature touches another dilating feature, leaving a line of separation (watershed line) between touching features.
Binary Image of Particles