Removal of specular reflections in tooth color image by perceptron neural nets

Abstract

This study presents the removal algorithm of specular reflections in a tooth color image to eliminate the specularities which degrades the performance of color image segmentation algorithms. Our proposed methodology includes two tasks: (i) automated detection of specular reflections by Perceptron neural nets and (ii) recursive corrections of the specularities by applying a smoothing spatial filter on the target pixels (specular regions) based on the decision of Perceptron.