On the basis of the cardinalized probability hypothesis density (CPHD), track-before-detect (TBD) algorithm is able to effectively solve the detection and tracking of weak point target with unknown target number. A detailed study of the CPHD algorithm which starts from the standard CPHD filter to the practicalities of TBD is presented. The updated expression for calculating particle weight of CPHD-TBD algorithm was deduced. Meanwhile, according to the physical means of the target distribution of CPHD, its update calculation in TBD has been implemented. Ultimately the combination of the CPHD and TBD has been achieved. The method to use it was introduced. The CPHD-TBD algorithm changes the way of target number estimation essentially compared with the PHD-TBD, resulting in accurate information of target distributions. Simulation results demonstrated that the proposed algorithm can estimate the number and states of targets more stability and accurately than the existing PHD-TBD algorithm.