Exploration of AI security: Application of a novel chaotic map 2D- SCD in the two-dimensional spatial mirror projection algorithm

Abstract

Artificial intelligence technology has broad prospects, and there are numerous researches related to computer vision. Ensuring the security of training image data for artificial intelligence models has become a research hotspot. In view of the characteristics of continuity and large amount of data of training images, this paper designs a new lightweight Two-dimensional Chaotic Map(2D-SCD) as a key generator. While ensuring the randomness of the key, it reduces the complexity of the chaotic system. The superiority of this chaotic map is proved through a series of performance analyses. In addition, this paper also proposes a random mirror mapping image security research scheme. Firstly, combined with the proposed Quadrature Amplitude Modulation(QAM), pixels of different blocks are migrated between blocks with different random waveforms to break the correlation between adjacent pixels. Afterwards, a chaotic system is used to generate a random mirror. By decomposing the key and pixel value in a 2D space, the key constitutes a spatial mirror, and the pixel value generates a diffusion value through the projection of this spatial mirror, thereby achieving the hiding of plaintext information. Simulation outcomes demonstrate that the suggested encryption method can successfully safeguard image security in the context of artificial intelligence.