Noname manuscript No. (will be inserted by the editor) Shape controllable geometry completion for point cloud models Long Yang 1, 2 · Qingan Yan 1 · Chunxia Xiao 1* Received: date / Accepted: date Abstract Geometry completion is an important operation for generating a complete model. In this paper, we present a novel geometry completion algorithm for point cloud models, which is capable of filling holes on either smooth models or surfaces with sharp features. Our method is built on the physical diffusion pattern. We first decompose each pass hole-boundary contraction into two steps, namely normal propagation and position sampling. Then the normal dissimilarity constraint is incorporated into these two steps to fill holes with sharp features. Our algorithm implements these two steps alternately and terminates until generating no new hole-boundary. Experimental results demonstrate its feasibility and validity of recovering the potential geometry shapes. Keywords point cloud model · geometry completion · sharp features · normal propagation · position sampling 1 Introduction Benefiting from its simple representation, point cloud model has been widely used in the last two decades [5, 13, 21]. Although capture devices have been improved substantially, the scanned data still contain deficient holes in certain situations. Moreover, we often confront abraded surfaces and damaged models with different deficiencies. All these holes need to be completed appropriately. Many techniques have been proposed to deal with this ill-posed problem. The existing methods, such as [3, 9, 15, Long Yang, Qingan Yan, Chunxia Xiao(Corresponding author) E-mail: {yanglong,yanqingan,cxxiao}@whu.edu.cn 1 Computer school, Wuhan University, China 2 College of information engineering, Northwest A&F University, China 23, 24, 28, 33], are designed to fill holes for polygonal mesh models. Most of these methods define geometry completion operators by utilizing connection topology and generate ro- bust hole-filling results. More details about mesh completion please refer to the surveys of [2, 6, 18]. Filling holes directly on point cloud models currently turns to be an essential requirement for many practical applications. Early work [8] presents an overall pipeline of geometry completion for point cloud models. Although many techniques [7, 19, 20, 25, 29, 30, 32] work on point cloud models, most of them only generate smooth hole- filling results. In contrast to smooth hole-filling, sometimes it makes special sense to complete a hole by preserving sharp features (i.e. edge / corner / apex) or using the least materials, see figure 1(e) and 1(b). Since general smooth hole-filling methods cannot complete the protruding features plausibly, it is still a difficult task to recover the potential sharp features on a deficient point cloud surface. Our work is inspired by the observation that geometry completion should reasonably provide potential shape op- tions for deficient regions. Therefore, in this paper, we pro- pose a novel hole-filling approach for point cloud models. Our algorithm simulates energy diffusion process and progressively contracts a hole-boundary until the hole is closed. Unlike the existing boundary propagating method [10], it could control the propagating process effectively. The main contributions of this paper are as follows: – Presenting a unified geometry completion algorithm that recovers both smooth and feature preserved holes for point cloud models. Sharp features are reproduced by controlling the hole-boundary contracting process. – Developing a new position sampling operator based on elastic force to generate the filling points. It avoids local