Color quantization using principal components for initialization of Kohonen SOFM
Dimitris Mavridis, Nikos Papamarkos · 2009
A new method is proposed for initializing Kohonen's self-organizing feature maps (SOFM) of fixed zero neighborhood radius for use in color quantization. The method employs the two largest principal components of the input image so that the initial weights of a number of neurons approach the input image color distribution. The rest of the neurons are initialized using the smallest principal component of the input image. Namely, standard SOFM is applied to the projection of the input image pixels onto the plane spanned by the two largest principal components and to pixels of the original image defined by the smallest principal component. The neuron values which emerge initialize the final SOFM of fixed zero neighborhood radius that performs the color quantization of the original image. Experimental results show that the proposed method can often produce smaller quantization errors than standard SOFM and other color quantization methods.