An algorithm based on cardinality probability hypothesis density (CPHD) filter for in-group target estimation is proposed, aiming at the state estimation and in-group target number estimation of dense group targets under clutter conditions. The algorithm constructs multiple hypotheses on number of the unknown targets corresponding to unresolved measurements through the amplitude information. On this basis, a new measurement likelihood function with amplitude information and position information is established. Then, based on the CPHD filter framework, to enhance the precision of the in-group targets’ number estimation, a new update procedure is derived. Furthermore, the algorithm distinguishes between the real targets and clutter through using ellipsoidal gating method to decrease the computational amount. The simulation experiment’s findings show that the algorithm significantly improved the estimation accuracy for both the number of targets within the group and the group target states.