A Streamlined GAN Based Real-Time Image Information Hiding for Networked Visual Servo Systems

The existing image information hiding methods based tampering detection cannot satisfy real-time requirement of networked visual servo systems (NVSSs). To solve the problem, a novel streamlined generative adversarial network based real-time image information hiding (SGAN-RIIH) method is proposed by reducing the layer number of convolutional neural network in the generator and selecting random frame to carry out image information hiding, where the consumed time of SGAN-RIIH method is decreased importantly and the image hiding performance is still satisfactory. Specially, the generative adversarial network (GAN) trained by COCO datasets can be implemented to other datasets. The proposed SGAN-RIIH method is verified for tampering detection on NVSSs.The experimental results show that the SGAN-RIIH method can assist in tampering detection.