Study of micro-vessel recognizing using sequence image spatiotemporal relationship

In this paper we employ spatiotemporal relationship among sequential images to deal with micro vessel segmentation. The first stage, a morphological filter bank was applied to remove the background noise, smooth silhouettes and fill gaps in the images. The second stage, a centroid-based boundary search was used to segment forehead object, micro vessel from non-uniform background. Finally centroid-based silhouette search was carried out and pilot trial showed that boundary of micro vessel can be picked up though further smoothing task should be employed in future work.