关键词: 铝箔密封性检测,   ,  热像图,    , Gabor变换,   ,  极限学习机(ELM),   ,  分类识别

Abstract:         After the research on the distribution characteristic of temperature field for aluminum seal , a kind of inspection method for the sealing based on Gabor transformation and extreme learning machine (ELM) was proposed. The Gabor transformation was conducted on various thermal images ，and the textural features are extracted to train the ELM neural network. Then the training results are used to classify and identify the thermal images， and the sealing of aluminum seal could be judged through the classification features of the thermal images. Compared and analysed with BP neural network which extracts color features，it is found that the algorithm based on Gabor transformation and ELM has the advantages of strong generalization，fast response speed，high precision，etc.

Key words: Aluminum foil sealing detection,    , Thermal image,    , Gabor transformation,    , Extreme learning machine (ELM),    , Classification recognition