Ensemble Model of Lanczos and Bicubic Interpolation with Neural Network and Resampling for Image Enhancement

Abstract

In the field of image processing, enhancing image resolution through upscaling and downscaling poses significant challenges. This study investigates the efficacy of an ensemble model combining Lanczos and Bicubic interpolation methods, augmented by neural network techniques, for image enhancement. We applied an innovative ensemble model, integrating Lanczos and Bicubic algorithms, to a variety of images. The model's performance was rigorously evaluated using Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) across different scenarios, including varying image scales and device origins. The results revealed that while each algorithm has unique strengths, the ensemble model demonstrated superior performance in certain aspects, particularly in the context of image origin and scaling. These findings suggest the potential of ensemble models in image enhancement and underscore the need for further exploration into advanced interpolation techniques for improved image processing. Future research should focus on expanding testing parameters and exploring alternative algorithms to further advance the field of image enhancement.