Content-Based Feature Extraction: Image Binarization
Rik Kamal Kumar Das · 2020
Image binarization is a trusted technique for extraction of robust content-based descriptors from image data. Binarizing an image will attempt to differentiate its background from its foreground by determining a threshold value. This is significantly useful in identifying the region of interest for feature extraction. This chapter demonstrates different threshold selection techniques to facilitate efficient binarization process for feature extraction. It demonstrates all the implementations with MATLAB codes for better understanding. Each of the extracted feature vectors are compared against each other using diverse classification metrics and are evaluated for statistical significance of contribution in enhancing performances of classification.