Quantum-Inspired Denoising via Adaptive Planck Constant and Image Patching

Traditional denoising methods face challenges in environments where images are corrupted by high noise levels, due to limitations in learning intricate patterns and capturing complex structures. To overcome these limitations, recent efforts have explored learning-based approaches and especially deep learning. However, in addition to their long training time, these approaches face the lack of labeled data in many applications area such as medical imaging. A recent appealing alternative is the integration of quantum principles into denoising techniques, showing promise in improving performance. However, current approaches often entail high computational costs and manual hyperparameter setting. Our study introduces an adaptive Planck constant formulation to quantum-inspired denoising, reducing computational overhead and streamlining the process. By incorporating this adaptive approach alongside image patching, we achieve improvements in denoising efficiency. This advancement not only eliminates the need for manual intervention in hyperparameter settings but also enhances the usability of quantum-inspired methods.