Alert button

Pixel-level Reconstruction and Classification for Noisy Handwritten Bangla Characters

Jun 21, 2018
Manohar Karki, Qun Liu, Robert DiBiano, Saikat Basu, Supratik Mukhopadhyay

Figure 1 for Pixel-level Reconstruction and Classification for Noisy Handwritten Bangla Characters
Figure 2 for Pixel-level Reconstruction and Classification for Noisy Handwritten Bangla Characters
Figure 3 for Pixel-level Reconstruction and Classification for Noisy Handwritten Bangla Characters
Figure 4 for Pixel-level Reconstruction and Classification for Noisy Handwritten Bangla Characters

Share this with someone who'll enjoy it:

Classification techniques for images of handwritten characters are susceptible to noise. Quadtrees can be an efficient representation for learning from sparse features. In this paper, we improve the effectiveness of probabilistic quadtrees by using a pixel level classifier to extract the character pixels and remove noise from handwritten character images. The pixel level denoiser (a deep belief network) uses the map responses obtained from a pretrained CNN as features for reconstructing the characters eliminating noise. We experimentally demonstrate the effectiveness of our approach by reconstructing and classifying a noisy version of handwritten Bangla Numeral and Basic Character datasets.

* Paper was accepted at the 16th International Conference on Frontiers in Handwriting Recognition (ICFHR 2018)  
View paper onarxiv icon

Share this with someone who'll enjoy it: