Example-Based Facade Texture Synthesis

Dai, Dengxin; Riemenschneider, Hayko; Schmitt, Gerhard; Van, Luc

doi:10.1109/iccv.2013.136

Cited by 24 publications

(23 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A parametric model of building facades as waveforms, based on Fourier series, is used to approximate the facade structure. (Dai, 2013) also details another interesting method for synthesizing realistic façade textures from a simple example. The sample image is parsed and described as semantic components that can be tilled to generate some new facade textures with different dimensions or with occluded parts inpainted.…”

Section: Context and Objectivesmentioning

confidence: 99%

Adaptive Texture Synthesis for Large Scale City Modeling

Despine¹,

Colleu²

2015

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:Large scale city models textured with aerial images are well suited for bird-eye navigation but generally the image resolution does not allow pedestrian navigation. One solution to face this problem is to use high resolution terrestrial photos but it requires huge amount of manual work to remove occlusions. Another solution is to synthesize generic textures with a set of procedural rules and elementary patterns like bricks, roof tiles, doors and windows. This solution may give realistic textures but with no correlation to the ground truth. Instead of using pure procedural modelling we present a method to extract information from aerial images and adapt the texture synthesis to each building. We describe a workflow allowing the user to drive the information extraction and to select the appropriate texture patterns. We also emphasize the importance to organize the knowledge about elementary pattern in a texture catalogue allowing attaching physical information, semantic attributes and to execute selection requests. Roofs are processed according to the detected building material. Façades are first described in terms of principal colours, then opening positions are detected and some window features are computed. These features allow selecting the most appropriate patterns from the texture catalogue. We experimented this workflow on two samples with 20 cm and 5 cm resolution images. The roof texture synthesis and opening detection were successfully conducted on hundreds of buildings. The window characterization is still sensitive to the distortions inherent to the projection of aerial images onto the facades.

show abstract

Section: Context and Objectivesmentioning

confidence: 99%

Adaptive Texture Synthesis for Large Scale City Modeling

Despine¹,

Colleu²

2015

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…This stream of methods are very efficient once the tiles are estimated. However, identifying these tiles is non-trivial: [24] handled this problem by estimating the translation symmetries and [5] through semantic labeling.…”

Section: Related Workmentioning

confidence: 99%

“…This strand is theoretically well-founded, but is computationally expensive. The third group generate textures by copying pixels or patches from the exemplar inputs [8,7,16,15,5]. Unlike the first two groups they do not provide cues for texture analysis, but are often more efficient and tend to work for a larger variety of textures.…”

Section: Related Workmentioning

confidence: 99%

“…Techniques of example-based texture synthesis can be broadly categorized into four categories: feature-oriented synthesis [13,6,33,9], Markov Random Fields (MRFs) methods [32,40,39], neighborhood-based methods [7,16,15,5], and tile-based methods [4,23]. The first group learns the statistics of carefully designed features and leads the synthetic images to have/achieve similar values, e.g.…”

Section: Related Workmentioning

confidence: 99%

“…A substantial amount of work has been devoted to synthesising textures from examples [13,33,7,16,19,27,5]. We will refer to such example-based texture synthesis as 'ETS' in the remainder of this paper.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

The Synthesizability of Texture Examples

Dai

Riemenschneider

Gool

2014

2014 IEEE Conference on Computer Vision and Pattern Recognition

Self Cite

View full text Add to dashboard Cite

Example-based texture synthesis (ETS) has been widely used to generate high quality textures of desired sizes from a small example. However, not all textures are equally well reproducible that way. We predict how synthesizable a particular texture is by ETS. We introduce a dataset (21, 302 textures) of which all images have been annotated in terms of their synthesizability. We design a set of texture features, such as 'textureness', homogeneity, repetitiveness, and irregularity, and train a predictor using these features on the data collection. This work is the first attempt to quantify this image property, and we find that texture synthesizability can be learned and predicted. We use this insight to trim images to parts that are more synthesizable. Also we suggest which texture synthesis method is best suited to synthesise a given texture. Our approach can be seen as 'winner-uses-all': picking one method among several alternatives, ending up with an overall superior ETS method. Such strategy could also be considered for other vision tasks: rather than building an even stronger method, choose from existing methods based on some simple preprocessing.

show abstract