Energy‐Efficient Sparse Kogge‐Stone Based Approximate Adder and Improved Baugh‐Wooley Multiplier Integration for Accelerated Image Enhancement in Signal Processing

ABSTRACT

Recent advances in approximate and optimized arithmetic architectures have significantly improved the power efficiency, propagation delay, and area utilization of digital processing systems. However, many existing designs primarily optimize either addition or multiplication operations independently, whereas several approximation‐based approaches introduce computational inaccuracies, routing complexity, or scalability limitations that may affect image quality and real‐time processing performance. To address these challenges, this paper proposes an energy‐efficient arithmetic architecture that integrates a Sparse Kogge–Stone Adder (SKSA) with an Improved Baugh–Wooley Multiplier (IBWM) for accelerated image‐enhancement applications. The proposed framework employs sparse carry propagation, optimized signed multiplication, selective approximation, and fault‐tolerant logic partitioning to reduce switching activity and arithmetic critical‐path delay while preserving computational accuracy. By jointly optimizing addition and multiplication operations within a unified arithmetic pipeline, the architecture achieves an effective balance among power consumption, hardware area, processing speed, and image fidelity. Experimental synthesis and simulation results demonstrate that the proposed SKSA–IBWM architecture achieves up to 31% lower power consumption, 14% reduction in silicon area, and 8% improvement in operating speed compared with recent optimized arithmetic architectures. Image‐quality evaluation further confirms the effectiveness of the proposed design, achieving a peak signal‐to‐noise ratio (PSNR) of 46.93 dB and a structural similarity index (SSIM) of 0.9374. The obtained results demonstrate that the proposed architecture provides a scalable, high‐performance, and energy‐efficient arithmetic solution for real‐time image processing and resource‐constrained embedded vision applications.