Research On Oracle Image Segmentation and Recognition Based on LBP And R-CNN
Yuxiang Ma, Suhang Zhu, Jiaxu Liao · Highlights in Science Engineering and Technology · 2025
With the development of today's science and technology, we are constantly exploring the application of computer algorithms and other technologies in the study of oracle bones. In this paper, for the original carrier image of oracle bone, first filtering and denoising to reduce or attenuate the many interference factors in the original topographic image of the oracle bone, and then the local binary mode of feature extraction on the three images to realize the feature enhancement of the image, and then use the support vector machine to optimally classify the image features, so as to achieve the accurate identification of the interference factors, and finally through the feature confusion matrix to prove the accuracy of the oracle bone. After completing the noise processing and classification of the image, the R-CNN algorithm is used to combine deep learning and target detection, iteratively learning the training set and validation set, and automatically adjusting the parameters until the optimal model is derived to realize image segmentation and recognition.