PROJECT TITLE :

Practically Lossless Affine Image Transformation

ABSTRACT:

An almost lossless affine 2D picture transformation approach is introduced here. Chirp-z transform theory is extended so that affine transformations of general n-dimensional images are possible. Our zero-padding technique also greatly reduces the loss of our transform, where normal transformations generate blurring artefacts because of sub-optimal interpolation. An approximate factor of 1800 improvement in the mean squared error is achieved with the proposed strategy, and an improvement of 250 is achieved with the best competitor. This paper is supplemented by python code for 2D images, which we derive from the transform's basic concepts. We demonstrate our method's higher image quality in comparison to other methods in a series of trials. When employing a toolbox algorithm, however, the runtimes are significantly longer