The Image Encryption Method Based on Chaotic Systems and Ghost Imaging with Deep Learning Decryption

The paper introduces a deep decryption method based on chaotic ghost imaging, termed Chaotic Ghost Imaging with Deep Decryption (CGI-DD). Utilizing chaotic sequences generated by the Hénon system as encryption keys, combined with ghost imaging technology, the intensity information of an object is acquired using a single-pixel detector to complete image encryption. To improve the quality of decrypted images, this paper introduces a deep learning model based on deep image prior technology. This model can effectively remove noise and restore high-quality decrypted images even at low sampling rates. Experimental results show that the CGI-DD method outperforms traditional ghost imaging methods in decryption effectiveness across various sampling rates, particularly demonstrating accurate reconstruction of target images even under conditions of low sampling rates.