Joint resampling and restoration of hexagonally sampled images using adaptive Wiener filter
Date of Award
M.S. in Electrical Engineering
Department of Electrical and Computer Engineering
Advisor: Russell C. Hardie
The premier objective of this research is to study the non-uniform interpolation to resample a hexagonally sampled image to a regular rectangular grid and to show hexagonal sampled data is more efficient than rectangular sampled data using an Adaptive Wiener Filter (AWF). Image processing is very important in several applications and have been using in them very efficiently. Digital image acquisition hardware, such as digital cameras, take photos by recording images as digital data using optics and a detector array. Distortions such as blur, noise and aliasing are often present, and these degrade image quality. For such reasons, image restoration algorithms are often applied to acquired images to reduce the degradations. Normally we use a rectangular sampling to digitize a continuous scene. There could be some other approaches to use as an alternate for this. One approach is to change the sampling process from rectangular pattern to hexagonal sampling pattern, considering various advantages. There is no inconsistency in pixel connectivity and thus angular resolution is higher in this arrangement and also fewer less samples need to represent the data represented in rectangular pattern. This research gives an overview of implementation of hexagonal sampling can be done by simulating a hexagonal sampled camera and adapt AWF to hexagonal sampling. Apply new AWF to simulated data quantitatively and qualitatively evaluate performance verses a standard rectangular sampling camera.
Digital images Deconvolution, Adaptive filters, Image reconstruction, Electrical Engineering, Sampling, Hexagonal Sampling, Rectangular Sampling, Adaptive Wiener Filter, Wiener Filter
Copyright 2015, author
Burada, Ranga, "Joint resampling and restoration of hexagonally sampled images using adaptive Wiener filter" (2015). Graduate Theses and Dissertations. 1098.