In order to solve the defect such as low forgery detection precision and difficult to detect small tampered area induced by color interpolation noise which resulting in low resolution because of using the color filter array (CFA) in current image forgery location technologies, the copy-move image forgery detection algorithm based on pixel prediction error coupled likelihood mapping was proposed. The green component was extracted from the image based on analysis of CFA. The prediction error and its local weighted variance model were constructed by embedding the weight factor, and the function of false feature statistics was defined by prediction error and Bayes theory to identify the feature value tending to zero. The likelihood ratio of this feature function was built by the feature statistical model to output forgery mapping for finishing the detection. Simulation results show that this algorithm has better robustness to identify small false pixels, as well as higher AUC value and ideal ROC curve.