Method for automatically extracting key information of paper materials

The invention discloses a method for automatically extracting key information of paper materials. The method comprises the following steps: 1) converting paper materials such as a vehicle dispatchinglist, a maintenance list and an oil fee list in vehicle work into pictures through scanning equipment; 2) normalizing the obtained pictures, removing noise, and reducing the complexity among the pictures; and 3) designing a convolutional neural network optimization algorithm, identifying characters in the paper material picture, and designing a TF-IDF optimization algorithm to extract key information in the picture characters; and automatically storing the key information in the vehicle using system. The method has higher superiority in key information extraction, and can extract the useful key information of the paper material in the working field, and the convolutional neural network algorithm can automatically optimize and adjust parameters along with continuous enrichment and expansionof the model data set, th