I Want to Change My Floor: Dominant Plane Recognition from a Single Image to Augment the Scene
J. A. de Jesús Osuna-Coutiño, Claudia Cruz-Martínez, José Martínez-Carranza, Miguel Arias-Estrada, Walterio W. Mayol-Cuevas · 2016
Augmented reality combines real footage taken of a scene with virtual elements. However, most current methods rely on camera localisation and 3D reconstruction or point cloud generation in order to integrate augmented reality to the footage. In contrast, in this work we present a novel method to augment virtual reality to the scene based on the recognition of dominant planes in interior scenes. Our method uses a rule system to select predefined decisions on a set of variables in order to infer dominant planes in the scene. For this, we propose to combine information from texture features, a measure of blurring in the dominant planes, and a segmentation of regions in a scene based on superpixels. The rule system infers the regions corresponding to the dominant planes, whilst light intensity in the scene is inferred from segmented regions. We also propose an approach to remove regions misclassified as dominant planes. Finally, the floor in the scene is recognised as the most dominant plane and then replaced with an augmented texture. We demonstrate our approach in a video sequence where our method is applied in a frame-to-frame basis, thus, for each single image in the video sequence, the floor is automatically recognised as the most dominant plane and then replaced with a virtual texture and furthermore, whose appearance is modified according to our coarse light model inferred from our approach.