A performance impact of an edge kernel for the high-frequency image prediction reconstruction

As a rule, the performance of almost digital image processing (DIP) algorithms and these applications directly depends on the spatial resolution of observed input images. Unfortunately, from the current image sensor technology, it is hard to take sufficient high spatial resolution images from commercial devices therefore the fantastic research attempts and, consequently, simple digital image resolution enhancements have been boosted in the last decade. The high-frequency image prediction reconstruction is the simple and effective algorithm for enhancing the image resolution however this algorithm is strongly depends on the edge detection kernel and M₀ parameter. Therefore, this paper studies a performance impact of an edge detection kernel such as Roberts kernel, Prewitt Kernel, Sobel Kernel, Laplacian Kernel and Laplacian of Gaussian (LOG) Kernel for the high-frequency image prediction reconstruction. This paper presents three experimental performance studies under a noiseless environment, several blurred environments at different blurred variance and several noisy environments at different noise power levels. The first performance study is an empirical exhaustive study of an optimal edge detection kernel and the study of optimal M₀ value is experimentally determined under this environment. The second performance study is an empirical exhaustive study of an optimal edge detection kernel and the study of optimal M₀ value is experimentally determined under these environments. Finally, the last performance study is an empirical exhaustive study of an optimal edge detection kernel and the study of optimal M₀ value is experimentally determined under these environments.