Abstract

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.

Dataset

We build up a new texture dataset ETHZ Synthesizability, which contains 21,302 texture samples, and their synthesized results by four standard methods. All textures of the dataset are annotated according to their synthesizability: good, acceptable, and bad. The best synthesis method for each texture sample is also recorded. See Fig.2 for examples of such annotation.

Figure 2. Three texture examples from our dataset with their annotations of synthesizability. Left: texture exemplars; right: synthesized textures.

Features

We defined four texture features: Textureness , Homogeneity , Repetitiveness and Irregularity. See our paper for their definitions and corresponding methods. Below comes some examples.

Results

Synthesizability can be predicted with high precision (Table.1) and the prediction is largely consistent with human perception (Fig.6).