Performance evaluation of a novel sampling-based texture synthesis technique using different sized patches

Abstract

We propose a novel sampling-based texture synthesis algorithm called Multipatch, which improves on the results of previous sampling-based algorithms by using patches of different size, and by minimizing global pasting errors. A key feature of the proposed algorithm is that it always converges to a local minimum. Multipatch, the patchwork algorithm, and Wei and Levoy's multi-resolution texture synthesis algorithm, which is based on a treestructured vector quantization method, are statistically analyzed and subjectively evaluated. The results of simulations show that the patchwork algorithm yields a perceptually acceptable texture in a shorter expected running time than the other two algorithms; however, Multipatch is the most efficient in terms of obtaining a good quality texture image.

An Examplar-Based Approach for Texture Compaction Synthesis and Retrieval

“…Moreover, some analytical results based on nonparametric resampling of random fields and a variational point of view are presented in [25] and [26], respectively. One problem with the approach is the selection of the correct patch size because using different-sized patches tends to generate different textures [15]. In addition, the performance of many examplar-based algorithms has not been evaluated in terms of the probability metric.…”

“…The multipatch algorithm, shown in Fig. 4, was developed to address these issues [15]. The idea is to simply sample different-sized patches and use a combination of them to generate a new image, thereby increasing the size of the patch domain and improving the synthesized result.…”

“…Since is like a Dirac distribution and the entropy of the Dirac distribution is zero, we have in the derivation of (15). To maximum (9), we have…”

“…The examplar-based multipatch algorithm is also a random algorithm that chooses a patch at random from a candidate set to paste on to the synthesized texture. Performance comparisons of various randomized texture-synthesization algorithms are detailed in [15] and given in Subsection F below. The performance benchmark that we use to evaluate our decoder is based on a subjective test, which measures the probability of deriving acceptable and good quality textures synthesized from a compressed thumbnail texture.…”

“…In Section II, we review the multipatch algorithm [15], which is a variation of the patchwork algorithm [13], [16], and the LBP feature. The multipatch algorithm is an implementation of the examplar-based texture synthesis approach and is used to demonstrate our numerical results.…”