Abstract: Oven food maturity identification is an important part of intelligent oven. At present, maturity identification in the industry of home appliances is usually achieved by such methods as probes and oxygen sensing, which are marked by their inherent limitation. Spectral imaging, as a kind of optical sensing technology, brings advantages of large amount of information, wide range of detection, and no contact with target. In order to study the application prospects of spectral imaging in oven, the feasibility is demonstrated in principle. After feature extracted through data analysis, machine vision model and classification model of machine learning are built, so that 3 kinds of common food ingredients and 1293 groups of spectral data are collected, analyzed and tested. The results show that spectral imaging could realize positioning of food in oven and accurately identify food maturity, and the cross validation accuracy reaches above 91.3% on average.

Key words: Spectral imaging, Oven, Maturity recognition, Artificial intelligence, Machine vision