A Coverless Image Steganography Technique Based on Multi-Object Mapping Rules

With the rapid development of multimedia technology and computer networks, confidential and sensitive digital information is frequently transmitted over the Internet. To pre vent this information from being intentionally damaged or forged, ensuring its security during transmission has become an important research topic. Coverless image steganography can hide secret messages within images for transmission without leaving any traces, making it one of the main methods considered for securely conveying secret information. However, existing cover less image steganography techniques overlook the possibility that the number of images in the dataset may be insufficient to con struct a complete image sequence index, rendering them impractical for data hiding applications. In light of this, this paper pro poses a novel coverless image steganography technique based on multi-object detection. By creating indexes based primarily on object labels contained within images, it establishes a flexible mapping relationship between secret message segments and im age feature sequences, thereby reducing the number of images required to construct a complete sequence index. Experimental results demonstrate that this method can effectively perform the task of hiding secret messages in real image datasets. Further-more, our method significantly outperforms previous research in terms of steganographic capacity and robustness, showing substantial improvements of 27.65% and 16.07%, respectively.