Blind detection based on color SIFT

Blind detection algorithm based on keypoints has a small amount of computation and faster speed, however, its accuracy is lower, comparing with blind detection based on block. One factor is that keypoints are extracted in a grayscale image, which leads to loss of the color information of the image. In order to solve it, this paper proposes a novel blind detection algorithm based on key point image. Firstly, an image is changed from RGB space to HSV space, the color local binary patterns feature of each pixel is computed based on the HSV image. Then keypoints are extracted from the color local binary patterns image. which also carry color information. Then the nearest neighbors are used to find matching keypoints in all keypoints. Finally, wrong matching is filtered by the angle between matching points and morphology operations are used to come into being a forgery detection map to locate the tampered regions. Experimental results show that the proposed method based on color scale-invariant feature transform keypoint outperforms these methods based on traditional scale-invariant feature transform keypoint in the reliability of detection and the block-based method in efficiency.