The invention discloses a kernel K-means track association method based on KMDL criterion criterion, comprising the following steps: Step 1: constructing a typical track association scene; Step 2: using KMDL criterion criterion to determine the number of target tracks; Step 3: Correlate with the kernel K-means algorithm for the track observation scene. The disclosed kernel K-means track correlation method based on the KMDL criterion criterion of the present invention, based on the target state information, combines the KMDL criterion criterion and the kernel K-means algorithm to solve the complex environment (intensive clutter, close to the target, The multi-target track association problem with unknown number of targets). This method makes full use of the target's motion state information, effectively improving the correlation accuracy rate, the correlation criterion is simple and easy to implement, the calculation amount is small, the correlation accuracy rate is high, and it is not sensitive to target crossing, so it is suitable for navigation in dense and cross-target environments. Trace correlation, suitable for engineering implementation.