Research on glass classification and identification based on support vector machine

Abstract

Ancient glass is very susceptible to the environment of buried environment and weathering, in the soil its internal elements and the environment a large number of exchange, so that the proportion of glass components change, in order to identify the type of glass artifacts, we established a glass classification and identification model based on support vector machine (SVM). Firstly, 67 valid sample points of known categories were classified and sorted, and the data were integrated in the order of lead-barium glass, high-potassium glass, and unknown cultural relics. Use matlab to bring the sorted data into the SVM classification model to obtain the type of unknown cultural relics. Finally, the sensitivity analysis of classification was carried out by reducing the number of known types of cultural relics and adjusting the calcium oxide (CaO) content of lead-barium glass and high-potassium glass. The results showed that the glass artifacts numbered A6 and A7 were classified as high potassium weathering. A1 is high potassium unweathered; A2 and A5 are lead-barium weathering; A3, A4 and A8 are lead barium unweathered. Finally, sensitivity analysis shows that the overall stability of the model is good.