From Color Quantization to Image Segmentation

A new technique is presented for color image segmentation. Five processes are accomplished that are respectively dealing with color image quantization, noisy regions removal, removal of thin regions, color-based region merging, and area-based region merging. Some parameters involved in the method are automatically computed, others are fixed depending on the specific application. Thus, the method is characterized by some flexibility that makes it useful for different applications. The method has been checked on color images from publicly available repositories. The performance of the method has been evaluated in terms of Precision, Recall and F-measure. The obtained results are satisfactory from both a qualitative and a quantitative point of view.