Arabic character recognition using particle swarm optimization with selected and weighted moment invariants
Muhammad Sarfraz, Ali Taleb Al-Awami · 2007
A new Arabic character recognition system has been proposed using moments as features. The proposed scheme works in such a way that the features are selected as well as weighted using a swarm-based optimization technique. For the sake of simplicity, it has been assumed that the Arabic text has already been preprocessed and segmented. Recognition results have been achieved up to 82% of accuracy. Authors believe that the 82% of accuracy is mainly due to not using very effective segmentation technique, otherwise the results could be above 95% as has been observed in the case of object recognition in an earlier paper of the authors.