Cone-beam computed tomography (CT) has the notable features, viz high efficiency and high precision, and is widely used in the areas such as medical imaging and industrial non-destructive testing, but the presence of image lag reduces the quality of CT images. By referencing the multi-exponential decay model for the image lag and combining with the actual decay rule of the flat panel detector output signal, a new decay modeling and correction method for the image lag based on multi-exponential fitting is proposed. Firstly, an imaging experiment using cone-beam CT based on flat panel detector is carried out; the results show that the image lag decay of the pixels in the flat panel detector has a good consistency, and is irrelevant to the initial gray value. Then, the rapid image lag correction is achieved according to the image lag decay model, and the comparison of image quality of the projected images and slice images before and after image lag correction indicates that the edge sharpness of the part has been significantly improved after the lag correction. This method does not need to obtain the scintillation compositions and the decay time constants of the detector, and is easily applied to the practical cone-beam CT imaging systems for image lag detection and correction.