Impact of Distance Measures in Shepards Interpolation Algorithm on Natural Images Corrupted by Outliers

The impact of Distance Measures in an Interpolation Algorithm on Natural images corrupted by Outliers is proposed. The Adaptive Inverse Distance Interpolation (Shepards algorithm) is modified with various distances for increasing distances. The Window size increases adaptively when three non noisy pixels does not exist in the current processing window. It was found that L1 Norm is robust than L2 Norm on Natural images corrupted by high density outliers noise. The algorithm requires a minimum of three non noisy pixels to perform L1 norm based Sheppard algorithm. Exhaustive testing were done on images of standard Database The use of L1 Norm based as distance measure between the processed pixel and non noisy pixel had a better impact on the performance of the algorithm in the presence of outliers with information preservation.