Applied Convolutional Neural Network Algorithm in Handwritten Recognition for Digital Transformation

In the paper, we exploit a small branch of identification handwritten recognition problem. That is the recognition of letters, handwritten characters for digital transformation. The information is transmitted digitally using handwriting capture camera. Images will then be processed and put into identification for information by text. In the paper, we propose the handwriting recognition algorithms based on convolutional neural network (CNN). The solution method is to separate character from handwritten letters taken by camera, and then applying recognition algorithm to determine what is the letter. The goal of the paper is to reduce the parameter of CNN. Experimental results using MATLAB show that the proposal algorithm improves up to 90 percent when selecting suitable learning factor of neural networks. The flexibility of this design allows it to extend to other languages easily.