An Image Encryption Method for Object Detection Based on Chaotic System and DNA Sequence

Abstract

Existing image encryption algorithms for object detection have shortcomings such as small key space, poor anti-attack ability. Therefore, this paper proposes an encryption algorithm based on Logistic map, Chen system, and DNA sequence for object detection image encryption. The main idea is to dynamically generate a prediction frame based on the object detection network, encrypt the image block in the prediction frame for the first time, then encrypt the whole image. The key is employed to drive the Logistic map and Chen system to generate the chaotic sequences, which is used in DNA computing, scrambling, and diffusion operations. This paper describes the design of the encryption algorithm in detail and conducts security analysis, including histogram statistics, adjacent element correlation analysis, and information entropy analysis. The results show that the algorithm has good cryptographic characteristics and strong anti-attack, and can be used for object detection image encryption.