Aiming at the problem of information differentiation of heterogeneous sensors and particle poverty in detection and tracking of standard particle flow filter, an RPF-based tracking before detection algorithm for heterogeneous sensors is proposed. Due to the low particle size of standard particle filters, it is impossible to perform limited search detection on the detection space. Therefore, the RPF filter is introduced to solve the problem of particle deficiencies in particle filter resembling, and the number of tracking and searching particles is unchanged under the promise of ensuring tracking accuracy. At the same time, by using the spatial distribution characteristics of particles, the space transformation and registration are realized by means of spatial transformation, so as to realize the consistent representation of heterogeneous sensors in the probability space. The algorithm is simulated and the results show that the heterogeneous tracking algorithm is better than single sensor detection and tracking performance.