The good thing about erosion and dilation is that the fundamental principles that they operate upon are similar to image filtering. So, the basic concepts that you need to understand the nature of their operations are already known to you.
Now, when we talked about image filtering, we had mentioned that image averaging using a filter-based approach is a linear operation--the same type of operation is performed at all pixel locations and the output pixel intensity at any location is a linear combination of the intensities of its neighboring pixels. When we talk of erosion and dilation, we relax the constraints of linear combinations (the condition for the same operation to be performed at all locations still holds). In spite of the overwhelming similarity between filtering and morphological operations, there are some differences between the two. First, the kernel that is used in image thresholding is given the name of structuring element. And a change in the name always...