Supervised genetic image segmentation
Christophe Rosenberger, Kacem Chehdi · 2003
We present a supervised image segmentation method using a local ground truth to determine the level of precision of the final result. The segmentation of an image is realized by optimizing two criteria with a genetic algorithm. The first is unsupervised and measures the quality of a segmentation result. The second computes the good classification rate on a local ground truth set by the user. The optimization process takes into account the nature of regions. We show the efficiency of the method through experimental results on several images.