The paper aims to solve the poor robustness and low detection accuracy of the current image copy-paste forgery detection algorithm. The color information of the image was introduced into the process of forgery detection. An image copy-paste forgery detection algorithm based on dual information statistical mechanism coupling gravitational clustering was proposed. First, the Hessian matrix was used to extract the feature points accurately. Then, the gradient histogram was used to describe the directional features of the image, and the color information of the image was introduced into the feature representation of the image. The double information mechanism was constructed by using the color information and gradient information of the image to obtain the feature vector of the image. An approximate measurement model was constructed by calculating Euclidean distance between feature vectors to match image features. Finally, the clustering algorithm was used to realize the clustering of image feature points and detect the content of copy-paste forgery accurately. The experimental results show that the proposed method had higher detection accuracy and better robustness than the current image copy-paste forgery detection method. The proposed scheme can accurately detect and locate the forged content. It has certain reference value in the field of image watermarking and information security.