A new image content forgery detection algorithm by means of energy constraints and structure similarity clustering mechanism is proposed. Firstly, with the help of Hessian operator, a box filter is utilized to generate row Hessian formulation to accurately detect image features. Then, by calculating the Haar wavelet value of the image, the direction information of the image is obtained to construct the neighborhood window of the image features. Then the curvature information of the pixels in the neighborhood window is calculated to form a robust eigenvector. Finally, Euclidean distance measurement is adopted to measure image features, and regional energy features are combined to complete the constraints of measurement results, so as to achieve accurate matching of image features. Structural Similarity Index(SSIM) function is utilized to identify forgery areas by clustering matching results, and accurate detection is realized. The simulation results show that, compared with current content detection technology, the proposed algorithm has higher detection accuracy and robustness under the interference of various geometric transformations.