Impurity detecting method in medicine detection robot based on time domain features of sequence images

The present invention discloses a method for detecting impurities in a characteristic sequence region based upon an image of a pharmaceutical detecting robot, according to the sequence domain features image pixels, wherein the light background areas and dark areas to the same feature map feature space, and eliminate the background light the difference between the dark background and of the target and background classified by the neural network, to achieve the target and background segmentation, feature motion trajectory according to the foreign matter, the foreign matter is detected. An effective solution to the background light can not be solved sequence frame difference method, a dark background image segmentation, the foreign object detection missing rate, a lower detection error rate; the process of the invention compared to prior sequence difference method is faster, and is not a sequence of frames the number of impact, especially in the above sequence of image processing 20, the effect is more pronounced; BP neural network, the learning sample by the method of automatically establishing classification criteria, enhancing the ability to adapt to the complex background, features meet more samples effective extraction.