A performance comparison of two PWS filters in different domain for image reconstruction technique under different image types

Due to many factors that can be degraded an image quality from the desired version. Image reconstruction application is the method that aims to recover those degradations based on mathematical and statistical models. Partition-based weighted sum (PWS) filtering is one of the most effective techniques for application of an image restoration and reconstruction. In this paper, we compare two PWS filters in both frequency and spatial domain under several image types. Two PWS filters include hard partitionbased weighted sum (HPWS) filter and subspace hard partition-based weighted sum (S-HPWS) filter. Five image types are considered including aerial images, human images, miscellaneous images, object images and text images. The simulation results show that the spatial domain HPWS filter offers the best performance when we apply to restore object image, but this filter not successful in term of memory usage and complexity of computation. Frequency domain S-HPWS filter, which required less memory and computation time using PCA technique to reduce size of data, offers good performance when we attempt to restore miscellaneous image. On the other hand, text image gets poor performance from all types of filters.