Hands-On Image Processing with Python

SLIC

The SLIC algorithm simply performs k-means clustering (we will explore more on this clustering algorithm in Chapter 9, Classical Machine Learning Methods in Image Processing) in the five-dimensional space of color space (RGB or Lab) and image location (that is, pixel coordinates: x, y). This algorithm is very efficient, since the clustering method is simpler. To obtain good results with this algorithm, it is essential to work in Lab color space. The algorithm has quickly gained momentum and is now widely used. The compactness parameter trades off color-similarity and proximity, while n_segments parameter chooses the number of centers for k-means. The next code block demonstrates how to implement this algorithm using the scikit-image ...