The dim and small target association of heterogeneous sensors is the first problem to be solved by cooperative detection of sensors. Even in the same field of view, the detection targets of heterogeneous sensors composed of infrared photoelectric systems and radar are not exactly the same, especially in the long-distance detection, the radar detection targets are many and dense, while the detection targets of infrared photoelectric systems are relatively few, so the target track association result has a great uncertainty. Aiming at this problem, the dim and small target association method based on multi-source data and multi-feature fusion was firstly proposed based on the multi-model estimation method, selecting the same type of targets as the potential association target, then making the rough association of the same type of targets based on the track association algorithm, and finally making fine association of targets based on the multi-feature maximum joint probability distribution. The simulation tests of infrared photoelectric systems/radar are verified by the same station site detection that this method effectively improves the accuracy of the association of dim and small targets compared with only using track for target association.