Associating targets detected by heterogeneous imaging sensors is a key issue in Target recognition based on multi-sensor image fusion. Traditional algorithms often use a single type of information from the image to calculate the cost matrix between sensor-detected targets, such as attribute information or position information. When multiple targets have similar information, these methods often lead to incorrect associations and are susceptible to sensor system errors and environmental factors. However, as technology advances, the information that sensors can detect is becoming more diverse and abundant. Thus, this paper proposes an algorithm that combines grey relational analysis based on target attribute information and DBSCAN clustering analysis based on position information. Using Dempster-Shafer evidence theory, the algorithm fuses the two types of information to achieve target association for heterogeneous imaging sensors. According to Monte Carlo simulation experiments using ships as targets, comparative experiments were conducted with grey relational analysis based on attribute information, bias mapping clustering based on position information, and the weighted bipartite graph optimal solution algorithm that uses both attribute and position information as features. The experimental results indicate that the algorithm proposed in this paper can overcome the limitations of single-information target association. It effectively mitigates the impacts of false alarms and positional distribution, thereby improving the accuracy of target association for heterogeneous imaging sensors.