Auxiliary Pathological Analysis Based on Classification and Recognition of Electronic Gastroscope Images

In this paper, a computer-aided lesion detection method based on gastroscope image is proposed to improve the efficiency and accuracy of conventional gastroscopy. Considering that the mainstream image processing algorithms are not suitable for the region extraction of gastroscope image, this paper adopts a simpler and faster color feature extraction method with the characteristics of large changes between pixels. At the same time, a variety of feature combinations are used to replace the single feature, and the two-level series classifier is used to perform the de-interference and lesion detection steps in series. Finally, the design and implementation of gastroscope pathology auxiliary recognition system is completed in the relevant test. The results show that the recognition method based on classification detection is significantly better than the traditional machine learning classification method in the classification and recognition of precancerous diseases, and it has a significant improvement in accuracy compared with the deep learning model