Inverse Halftoning Algorithm Based on SLIC Superpixels and DBSCAN Clustering

Abstract

Halftone technology is widely used in the printing industry. This paper proposes an inverse halftoning algorithm based on SLIC (Simple Linear Iterative Clustering) superpixels and DBSCAN (density-based spatial clustering of applications with noise) clustering. Firstly, halftoning image is segmented by SLIC superpixels algorithm. Then the boundaries region of image is tracked by DBSCAN clustering algorithm and the boundaries of image is vectored. Secondly, the remaining part of halftoning image that boundaries have been extracted is smoothed by linear and nonlinear smoothing filters. Finally the vector boundaries and the smooth background is combined together to get the inverse halftoning image. Experimental results show that the proposed method can effectively remove halftone patterns while retains boundaries information.