Heron and Alpha-Centroidal Mean Filters with Exponential factors for Deep Space Images

Abstract

In this article, we discuss empirical methods that can be add-on to enhance spatial image filters. Discussing on the implementation of such methods for Deep-Sky images from space telescopes. Spatial image filtering houses mean filters that use kernel method to filter a digital image in the spatial domain. To improve the performance of the existing filters and aid the new upcoming filter, a technique is developed that can be added to the filter to improve their performance. This study gives a detailed analysis of an empirical weight factor and exponential factoring methods for the mean filters that are proposed and implemented. The exponential factoring is applied to the recent mean filters Heron, Centroidal and Inverse-Contraharmonic. Performance analysis of both methods and comparison to existing mean filtering performance are represented. The images are classified into white or black based background and analyzed for performance parameters, comparing the results from the filters and filters with the exponential factoring. The empirical methods were proven to improve the efficiency in denoising of images of the existing mean filters and the new Heron and Centroidal mean filters, hence can be used as add-ons to existing spatial mean filters. One of the empirical techniques, exponential factoring is promising in denoising high density noised images. This novel method is implemented to filter noise from Deep Space images taken by space telescopes. Ultra-high-resolution images are filtered and implemented on Single Board Computers for remote handling, with testing done on MATLAB.