Newly added untrained radiation source detection method based on multi-discriminator model

The invention discloses a newly-added untrained radiation source detection method based on a multi-discriminator model, and the method comprises the steps: obtaining a signal image sample set of N identified radiation sources, dividing the signal image sample set into a training set and a test set, setting a generator and N + 1 discriminators, sequentially constructing the generator and each discriminator into a sub-generative adversarial network, and carrying out the detection of the N identified radiation sources. Inputting the test set and the image sample set into a trained discriminator, carrying out statistics to obtain a judgment interval and a proportion parameter, inputting a signal image of a radiation source to be discriminated into each discriminator, carrying out statistics on a proportion of an output probability value in the judgment interval, and if the proportion is greater than the proportion parameter, outputting the radiation source to be discriminated. If yes, it is judged that the to-be-