The success of space rendezvous missions hinges on the accuracy of target detection. To address the challenges of target detection in distant non-coplanar rendezvous, including similarities between space targets and stars, minimal distinctive features, random imaging trajectories, and slow stellar motion, a space target detection algorithm based on star correlation and trajectory matching is proposed. Initially, multi-frame star map images undergo preprocessing, including noise reduction, threshold segmentation, centroid localization, and image registration. Subsequently, a multi-frame star association is performed using the proposed correlation function and association rules, filtering out stray noise points while extracting the inter-frame motion trajectories of star points. Finally, trajectory matching is conducted using three indicators：target size, velocity, and direction, to detect the space target. Simulation results demonstrate that the algorithm meets the detection requirements of non-coplanar rendezvous processes In the presence of a certain degree of frame mismatch, the spatial target detection rate can reach 99.8%, and the false alarm rate is 0.04%.