k-means-implementation

The k-means method is a classic classification tool which allows a data set to be divided into k homogeneous classes. Most of the images (photos, 2D vector drawings, 3D syntheses, etc.) locally verify properties of homogeneity, in particular in terms of light intensity. The k-means algorithm therefore makes it possible to provide a solution to the segmentation of images. The K-Means algorithm can be used to segment an image that has areas of relatively uniform color. We represent all the pixels of the image in a three-dimensional space based on their Red / Green / Blue components. We thus obtain a point cloud on which we apply the k-means algorithm. To illustrate the use of k-means, we use a synthetic image made up of two zones with clearly distinct colors.