An experimental performance analysis of an image super resolve reconstruction based on the high-frequency image prediction under several blurred and noisy environments

According to tremendous demands of high spatial resolution images, the brilliant research efforts in the area of image resolution enhancement algorithms have been raised hence the simple and fast computational resolution enhancement algorithms have been very attracted in the modern digital devices such as smart phone, CCTV, digital camera, etc. Due to its performance and fast computational time, this paper empirically experimental investigates the performance of the single image super resolve reconstruction based on the high-frequency image prediction for up to 14 standard test images. This paper has three main contributions. The first contribution is an experimental comprehensive study of an optimal interpolation technique selection and an optimal number of gradient directions in single image super resolve reconstruction based on the high-frequency image prediction under a noiseless environment. Moreover, the study of optimal M₀ parameter selection is computationally explored for this environment. The second contribution is a study of an experimental performance of the reconstruction under several blurred environments at different blurred variance. Moreover, the study of optimal M₀ parameter selection is computationally analyzed for this blurred environment. Finally, the last contribution is a study of an experimental performance of the reconstruction under several noisy environments at different noise power levels and the study of the optimal M₀ parameter selection is analyzed.