Improving inverse distance weighting method for single-image super-resolution

Superresolution Reconstruction in Automatic Thai Sign Language Feature Extraction Using Adaptive Triangulation Interpolation

“…The interpolation method offers several SR image processes, all of which use point data as an input. It can be used in inverse distance weighting [26], natural neighbors [27] or kriging [1] to generate images via the 3D analyst user interface. The reconstruction process, Trend, requires programming or instruction to interpolate the value.…”

“…These properties are essential for image fusion [22]. [23] There are three common SR-type methods: interpolation-based methods, reconstruction-based methods, and learning-based methods [5,26,27]. In general, SR assesses unknown HR pixels using neighborhood pixels with reconstruction properties.…”

“…The inverse distance weighting [26] interpolates values by randomly selecting sampling points that affect the target cell, and the effect gradually decreases according to the distance between the target cell and the sample. Thus, the point located nearest to the target cell is calculated with a weight greater than the farther point.…”

“…This measured the effect of improving blurry images with more clarity. This value was calculated from equation 6 and equation 7 [26]. 𝑃 𝑐 𝑗=1 𝑃 𝑟 𝑖=1 (7) Where b is the number of bits in one pixel, Pr and Pc are the numbers of pixels in rows and columns, respectively, X(i,j) is the value of the pixel image at i,j of the original image, X'(i,j) is the value of the pixel image at i,j of the SR image.…”

“…SSIM measurements were initiated with two sets of n intensities from the linked windows in the original and compared images. The mean μx and μy were calculated using equation 8 [26].…”