As the quantity of bogus photos rises, it's more necessary than ever to check their legitimacy and correctness. The scientific community is constantly attempting to build forensic techniques that may be used to analyse, detect, and pinpoint picture modifications.Copy-move forgery is a kind of malicious tampering attack on digital images in which a portion of the image is copied and pasted within the image to hide the image's critical characteristics while leaving no visible signs of manipulation. This form of image alteration raises a serious concern about the image's legitimacy for forensics.
This paper discusses ways for detecting copy-move forgery using both linear and non-linear scale space key-point descriptors.Scale Invariant Feature Transform (SIFT) and the Speeded Up Robust Features (SURF) which use the Gaussian/linear scale space and sets of Gaussian derivatives as smoothing kernels for scale space analysis, are experimented with. Due to their linear nature both of them, smooth features and noises to the same degree without taking object boundaries into account, blurring edges and details to some amount. To make blurring locally adaptive to image data, so that noise is blurred but details and edges are not affected, AKAZE features for copy move detection is also studied in this work. The results indicate that linear scale space based key-point descriptors namely SIFT and SURF performed better when compared to non-linear scale space key-point descriptor namely AKAZE.