Copy move forgery detection using low complexity feature extraction

The significance of blind forensic is to detect a forgery without a support of any digital water marking or signature. A common type of forgery is the copy move forgery. In this kind of forgery a portion from the same image is copied and pasted on the object which is undesirable. Such a forgery is generally carried out to conceal a portion or to enhance the image content. The method proposed in this paper is a new and simple approach for detecting copy move forgery in digital images, which extract features based on color information of the forged image. The features are extracted using Auto Color Correlogram, Color Moments and HSV color space. The experimental results show that the proposed features can detect a forged portion with considerable accuracy while maintaining simplicity in terms of features considered resulting in very low computational time. A comparative analysis between the results from above three techniques is presented in terms of accuracy and computational time. Output results indicate increase in accuracy with increasing values of features taken into consideration for forgery detection. Highest Accuracy of 90% is observed with Auto Color Correlogram which takes most computation time of 11.8 seconds while examining 8192 values of image block features which is also highest amongst three types of techniques considered.