Interpolation Techniques in Image Resampling

Abstract

The procedure of converting a sampled image from any coordinate structure to other structure is called Image Resampling. When forgeries are introduced in digital images, generally the operations like rotation, resizing, skewing etc., are included to make it relational with respect to adjacent original area.  So there is a recognizable loss in the quality of the image and it will become an important signature of manipulated images. Hence resampling is the default interpretation present in most of the tamped image. Resampling detection is an attractive standard tool in digital image forensics. Generally resampling objects are not visible to human eye in an altered images but periodic relationships get familiarized in image pixels. Because of it the changes is going to happen in certain characteristics of the image and it is going leaving behind periodical artifacts which are used as fingerprints for the forensics. These periodic interpolation objects present in the intensities of pixels  or other format of data illustration such as DFT, wavelet are the structures which detectors look for in order to decide whether an image, or a part of an image, has endured a geometrical transformation. This serves as an evidence of manipulation or tampering. This paper gives comparative study of different resampling techniques like Cubic Splines, Nearest Neighbor, Cubic convolution and Linear Interpolation, which can be used as detectors for an altered image containing resample portions