Abstract: For the security inspection at airports and stations,it is urgently needed to discriminate the liquid examination.We proposed a novel method for high atomic liquid container discrimination based on global and local features.First,we extracted the combined invariant moments based on the global feature,which quickly find a large number of candidates for an efficient recognition.Then local feature identification was made as a second judgment to improve the recognition accuracy of feature detection.The container wall the bottom projection features were chosen as the local feature.Experimental results show that this method combines good global and local target information.This approach can solve the container occlusion problems with a high atomic number materials in partial,and also has high recognition accuracy.