Research on Oracle Recognition Based on Convolutional Neural Network Modeling

This paper is dedicated to solving the interference and mis-segmentation problems in oracle bone image segmentation. Through data preprocessing, feature extraction and model building, we propose methods such as threshold segmentation, median filter model, global threshold segmentation model, combined with Gaussian filtering and deep learning technology, to realize accurate segmentation and text restoration of oracle bone topography images. Through mathematical modeling and image processing techniques, we effectively deal with the interference factors of the oracle bone itself and the challenges posed by variant characters, and finally propose a convolutional neural network deep learning model that can automatically identify variant characters, which provides new methods and ideas for the study of oracle bone and the construction of digitization projects. These methods not only improve the accuracy and efficiency of oracle bone image processing, but also provide strong support for the further development of oracle bone research.