Dense clutter in complex environment not only increases the urgency of radar point-trace fusion parallel processing algorithms, but also directly affects the efficiency and performance of traditional parallel computing methods. This paper focuses on the parallel implementation of radar point fusion under dense clutter, and proposes a parallel processing method of radar point fusion based on modified space partitioning method. Moreover, in response to the problem of continuity of boundary point traces, we specially designed the corresponding process. Simulation tests and actual measurements show that the algorithm can effectively overcome the degradation of parallel computing performance under dense clutter, improve the tracking accuracy and real-time performance of the algorithm, and meet the actual engineering application