Handwritten numeral recognition through superimposition onto printed form

Handwriting is, eventually, a variation of the printed forms where characters are little larger, smaller, angled and deformed than forms. The small changes in handwriting define parameters character to be recognized. Handwritten numeral recognition (HNR) poses significant challenges due deformations other variations. This study proposes new notion HNR on hypothesis that handwritten numerals distinct forms, which leads easier task with higher accuracy when superimposing images onto corresponding images. In proposed HNR, auto-encoder convolutional have been adapted for transform HNIs into PNIs, while neural network employed classification PNIs. superimposition method reduces computational overhead. Moreover, this employs simple pre-processing without feature extraction whereas traditional methods employ pre-processing, extraction, machine learning tools, add performance HNRSP has evaluated recognizing Bengali, Devanagari, English benchmark datasets system achieves 99.68%, 99.73%, 99.62% numerals, respectively.