Method and device for selecting optimal frame of fiber nasopharyngoscope video based on CNN (Convolutional Neural Network) and Transform

The invention provides a fiber nasopharyngeal endoscope video optimal frame selection method and device based on CNN and Transform. The method comprises the following steps: acquiring a fiber nasopharyngeal endoscope video of a single object; the fiber nasopharyngoscope video comprises at least one respiratory cycle of the object; splitting the fiber nasopharyngeal endoscope video into a plurality of image frames which are continuously arranged; selecting one image frame from the image frames based on a pre-trained frame selection network model to serve as an optimal frame of the fiber nasopharyngeal endoscope video; the frame selection network model at least comprises a CNN (Convolutional Neural Network) structure and a Transform structure. According to the method, the CNN structure and the Transform structure are utilized to identify the image frames, so that the optimal frame is selected from the image frames at the split positions, and the problem of accurate selection of the optimal frame of the fiber na