Edge detection based salient region detection for accurate image forgery detection

Image Forgery is a real time issue in this present era and that causes inimical effects in the society. This is more common due to the availability of numerous image modifications software. The ultimate aim of this project is to provide a best optimal solution to this existing problem. The forgery in images includes object removal, object addition, unusual color modifications. There are many existing techniques available to overcome this problem but they have some limitations such as lack of accuracy in saliency detection, larger hash values, deficient in detecting small area tampering. In this proposed method the image is preprocessed to fixed size so that generated hash value of fixed size. The resized image is converted into YCbCr image and is passed into Gaussian filter, blurred image is obtained. The saliency regions are accurately detected by using edge detection concepts, the global and local features are extracted from this detected region and the image is processed to remove the noise. The final image is obtained from which the object is extracted and edge pixels are detected and mapped to the original image, sensitive hash is constructed for those detected regions. This proposed method outperforms the existing system by accurately detecting saliency regions, reducing hash length and increases the sensitivity of the hash, so that even the small area tampering can be detected accurately.