Imaging of multiples, as a supplement to imaging of primaries, can provide a wider range of subsurface illumination. Therefore, it can provide more detailed information on subsurface structures. However, imaging of multiples suffers from crosstalk issues generated by unrelated events. Many strategies have been proposed to attenuate crosstalk, among which the angle domain Radon crosstalk attenuation algorithm achieves good application effect. In the angle domain, the true imaging is flat, while the crosstalk events have moveouts. Therefore, it is convenient to identify the crosstalk in angle gathers using the Radon transform. However, the conventional Radon transform lacks a quantitative description for crosstalk in angle gathers, which would affect the accuracy of crosstalk attenuation. In this paper, residual moveout kernels are derived with a Radon transform to attenuate crosstalk in angle gathers for imaging of multiples. First, two types of residual moveout (RMO) equations are derived based on the causality of crosstalk. A three-layer model is used to verify the correctness of the analytical solutions. Then, based on the derived equations, the two types of crosstalk can be attenuated respectively in the Radon domain. Synthetic experiments demonstrate that the derived RMO equations can effectively attenuate the crosstalk events in imaging of multiples.