Adaptive chaos compression sensing algorithm based on target detection

With the rapid proliferation of image information, the challenges associated with image storage and transmission have become increasingly prominent. Image compression a crucial strategy to address these issues by reducing redundant information, thereby lowering storage requirements and transmission bandwidth without compromising visual quality. Compression-aware techniques further enhance the efficiency of image storage and transmission by integrating sampling and compression processes.

This paper proposes an image compression scheme based on compressed sensing, which effectively alleviates storage and transmission burdens while preserving the visual quality of critical regions.

The proposed approach employs target detection techniques to segment images into important and non-important regions, applying distinct compression levels accordingly. Leveraging the superior pseudo-randomness of chaotic systems, the scheme generates measurement matrices and random coordinates, and further implements a random coordinate replacement strategy by embedding pixels from important regions into non-important regions.

Experimental results demonstrate that the proposed scheme efficiently compresses non-important regions while maintaining the visual fidelity of important regions, thereby achieving an optimal balance between compression ratio and image quality.

By partitioning image regions and adopting a differentiated compression strategy, the proposed scheme effectively reduces image file sizes while ensuring the clarity and recoverability of critical information. Furthermore, the introduction of a partial Hadamard matrix generated via a chaotic system, combined with the random coordinate replacement technique, not only alleviates the substantial pressure on image storage and transmission in network environments but also significantly enhances the security of information transmission.