Recognising Planes in a Single Image

摘要: We present a novel method to recognise planar structures in single image and estimate their 3D orientation. This is done by exploiting the relationship between appearance structure, using machine learning methods with supervised training data. As such, does not require specific features or use geometric cues, such as vanishing points. employ general feature representations based on spatiograms of gradients colour, coupled relevance vector machines for classification regression. first show that hand-labelled data, we are able classify pre-segmented regions being not, then incorporate into segmentation algorithm detect multiple from previously unseen image.