Approximating shapes in images with low-complexity polygons

Abstract

We present an algorithm for extracting and vectorizing objects in images with polygons. Departing from a polygonal partition that oversegments an image into convex cells, the algorithm refines the geometry of the partition while labeling its cells by a semantic class. The result is a set of polygons, each capturing an object in the image. The quality of a configuration is measured by an energy that accounts for both the fidelity to input data and the complexity of the output polygons. To efficiently explore the configuration space, we perform splitting and merging operations in tandem on the cells of the polygonal partition. The exploration mechanism is controlled by a priority queue that sorts the operations most likely to decrease the energy. We show the potential of our algorithm on different types of scenes, from organic shapes to man-made objects through floor maps, and demonstrate its efficiency compared to existing vectorization methods.

Floorplan generation from 3D point clouds: A space partitioning approach

“…Comparisons on RGBD scenes. We first compare our framework against the popular Douglas-Peucker algorithm [14], the object vectorization algorithm denoted as ASIP [10], and current state-of-the-art floorplan generation method FloorSP [6] on 100 scenes collected from panorama RGBD scans. We employ the pre-trained model trained on 433 RGBD scenes released by FloorSP to provide the pixelwise room instance labeling map.…”

“…Space partitioning data-structure is a popular tool to provide basic geometric elements for recovering polygonal rooms. After dividing 2D space into polygonal facets, rooms are segmented either using an iterative clustering method [8], a global multi-class labeling approach [9] or an efficient greedy optimization mechanism [10]. Note that all of these methods rely on room instance label of points to provide the semantic similarity between adjacent facets.…”

“…Our approach is inspired by partitioning-based methods. However, in contrast to [8], [9] and [10] which recover the floorplan graph only through a roomsegmentation step, our method relies upon a more robust two-step mechanism that first extracts the boundary polygon of the scene and then divides the inside space into rooms.…”

“…Ground truth Douglas-Peucker [14] ASIP [10] FloorSP [6] Ours of isolated facets while FloorSP and our method produce a valid connected graph. In term of geometric accuracy, floorplans of ASIP and our approach align better with input points than FloorSP and Douglas-Peucker, especially in case of non-Manhattan scenes.…”

“…This kind of methods are in general timeconsuming and weakly robust in presence of under-and over-detection of primitives. Another strategy is to partition 2D space into polygonal facets and assign a room instance label to each facet [8,9,10]. This strategy is typically more robust.…”