We present in this paper an automatic method for segmenting and labelling of liver its surrounding tissues in intra-operative laparoscopic images. The goal is to be able to distinguished between the different structure that compose a common intra-operative hepatic surgery scene. This will permits to improve the registration between pre-operative data and intra-operative images for task such as Augmented Reality. Our segmentation method consider the scene as a 3D structured point cloud instead of a laparoscopic images in order to exploit powerful informations such as curvature and normals, in addition to visual cues that permits to efficiently classify the scene. Our approach works well on sparse and noisy point clouds, thanks to a surface approximation stage, and unlike existing approaches, is independent of organs texture in the image. Experiements performed on challenging human hepatic surgery confirm that accurate segmentation and labelling are possible using 3D structure information and appropriate visual cues.