Hadamard defense mechanisms for rice grain image privacy

This paper proposes a novel Hadamard Transformation-based image defense methodology tailored for rice grain images, targeting enhanced security and computational efficiency. The pipeline includes halftoning, share generation, secret sharing via Hadamard transformation, and quality-enhanced reconstruction. Privacy-sensitive regions are isolated and resized to reduce processing overhead. The method is evaluated using metrics such as pixel-wise difference, entropy, histogram analysis, radar charts (PSNR/SSIM), and featural integrity, and compared with benchmark techniques. The proposed model achieves 97.01% reconstruction accuracy, 45 ms processing time, 60 KB memory usage, and demonstrates strong robustness (score: 0.89), scalability (60 images/s), and energy efficiency (0.035 J/image), outperforming existing approaches in security and performance.