Self-adversarial coding type data enhancement method for artillery inner bore flaw detection

The invention discloses a self-adversarial coding type data enhancement method for artillery inner bore defect detection, and belongs to the technical field of detection control. The invention aims to provide a self-adversarial encoding type data enhancement method for artillery bore defect detection, which is based on a self-adversarial encoder generation model combined with a self-encoder and a generative adversarial network and is used for deep hole part inner wall defect enhancement data detection. The prior distribution after coding in the VAE is close to standard normal distribution, a decoder samples the prior distribution and learns the prior distribution into a generated sample similar to an input deep hole inner wall flaw image, the generated sample is compared with an original sample so as to train a more vivid sample image, the VAE is used as a GAN generator to generate similar samples, and the generation accuracy of the deep hole inner wall flaw image is improved. And the GAN is used as a discrim