Halftone Feature Based Classification of Commercial White Print Paper Using BP-MLP

摘要

The paper presents a machine vision approach for classification of the plain surface commercially available printing paper. In the presented method the halftone features are extracted from the images of plain surface papers that are difficult to distinguish due to flat white appearance. The extracted features have been used to train the back-propagation multi-layer perception (BP-MLP) classifier of artificial neural network (ANN). The presented method has been tested with different commercially available A4 size printing paper and found to be promising in terms of classification performance. The paper also reveals that half toning which is conventionally used for display of continuous tone images may be potentially useful for classification tasks, particularly for materials with flat surfaces.