Character Recognition System Based on Depth Neural Network

Deep neural networks are increasingly important tools for machine learning and artificial intelligence research, because they can learn very complex tasks without explicit programming. They are also flexible because they can be easily adapted to solve new problems. Character recognition involves the post perception process and is affected by many factors, such as the amount of visual information available and cognitive impairment. This method assumes that there are two types of representations in character memory: abstract forms, which are determined by font attributes, may vary with each letter, and specific forms or “embodiment” representations. The detection part adopts the horizontal text line detection method to detect the text, and in the text part, the system is analyzed and introduced in detail from the production of the model to the design and implementation of the neural network.