| ARTIFICIAL INTELLIGENCE| COMPUTER VISION| PREPROCESSING|

A practical guide to morphological image processing

simple but powerful operations to analyze images

Image from Aquaryus15 at unsplash.com

A guide about which and what are morphological operations, how to implement them in Python, and why are important in digital image processing.

Here is the previous article of the series:

A practical guide to neighborhood image processing

Introduction to morphology

Photo by CDC on Unsplash

Morphology happens over time. It’s not necessarily a bad thing. — Jimenez Lai

Morphology (sometimes referred also as a mathematical morphology) can be considered a branch of neighborhood processing. It was developed in 1964 by Georges Matheron and Jean Serra to quantify characteristics of mineral cross-sections but proved to be valuable in many other applications. In general, morphology can be used to remove the noise originated by a first thresholding step (which often happens in images where the exposition is not uniform). In fact, morphology works very well with the binary images obtained by thresholding (but you can also use it on grayscale images).

Just as an example, a few cases can happen after thresholding and can be solved by morphology.

image by the author

Here I have applied hit or fit operations, dilation, and erosion, which will discuss below.

Morphology has different interesting applications, for instance, is used as preprocessing step in optical character recognition (OCR), detecting barcodes and license plates. Morphology operations are simple and computation not expensive and can be combined…