Character Recognition Based on PCANet

The character recognition is an important issue, which has been pursued in recent year. In the paper, we used PCANet(principal component analysis network) to learn the character features. We verified the influence of these parameters on the performance of PCANet by modifying the key parameters of the experiment. Then we made a handwritten dataset to do the experiment and to verify whether the PCANet can also be used to identify. And the results not only was okey but also the recognition rate can reach 95.56%. At last, we compared the experimental results with the experimental results on Caffe.