Improved AP Clustering Algorithm Based on Target Segmentation

This paper describe a feasible scheme of local visual navigation, local visual navigation application scenarios often some with complex background, target species more real scenario that the obstacle avoidance is particularly important. In particular the visual navigation target segmentation in the background and the foreground objects more complex scenarios important for predicting pre-step. This visual navigation closer to the qualitative analysis of the scene, so the performance can be weak, but easy to implement, without artificial markers, but more dependent on hardware performance, available online AP Cluster is running.