A+: Adjusted Anchored Neighborhood Regression for Fast Super-Resolution

A+: Adjusted Anchored Neighborhood Regression for Fast Super-Resolution

| Radu Timofte, Vincent De Smet, Luc Van Gool
The paper introduces A+, an improved variant of Anchored Neighborhood Regression (ANR) for single-image super-resolution (SR). A+ combines the strengths of ANR and Simple Functions (SF) by leveraging the features and anchored regressors from ANR while using the full training material, similar to SF. This approach enhances both the quality and speed of SR, achieving improved performance (0.2-0.7dB PSNR better than ANR) and excellent time complexity, making A+ the most efficient dictionary-based SR method to date. The method is validated on standard datasets and compared with state-of-the-art methods, demonstrating superior results in both quality and speed.The paper introduces A+, an improved variant of Anchored Neighborhood Regression (ANR) for single-image super-resolution (SR). A+ combines the strengths of ANR and Simple Functions (SF) by leveraging the features and anchored regressors from ANR while using the full training material, similar to SF. This approach enhances both the quality and speed of SR, achieving improved performance (0.2-0.7dB PSNR better than ANR) and excellent time complexity, making A+ the most efficient dictionary-based SR method to date. The method is validated on standard datasets and compared with state-of-the-art methods, demonstrating superior results in both quality and speed.
Reach us at info@study.space
Understanding A%2B%3A Adjusted Anchored Neighborhood Regression for Fast Super-Resolution