Enlarging and shrinking a region of interest

A very common technique for refining a region of interest derived using thresholding is either enlarging or shrinking it to fit our target size. This can be accomplished by the morphological operations called dilation and erosion, respectively. These operations can be implemented in MATLAB using their respective functions, intuitively named imdilate and imerode.

Explaining and analyzing the mathematical properties of dilation and erosion lie beyond the scope of this book. We will instead explain their significance using practical examples that demonstrate their importance for image processing. The basic idea that you have to understand before we start, is that the two operations can be used for enlarging or shrinking an ROI (denoted by the instances of 1 in the image) using a structuring element. Structuring elements can be small binary images generated by the user either arbitrarily (placing the instances of 1 and 0 in a small image), or by using the strel function. The choice of a structuring element should be made following two simple rules:

Let's dive right in, to understand the physical meaning of all these concepts in practice.