Research On Texture Synthesis Method Based On Optimum Match

Texturesythesisbased on sample is inordertosynthesizethe similar image whichsizeissatisfied user’s requirements.Thistechnologyusesmalltextureimages asinputpictures.Basedonthestudyofsample-basedtexture sythesis,this papera new methodbasedontheexistingalgorithms,themainworkincludes:Introduces currentsituation,problemsandapplicationsofthesample-basedtexturesynthesis,thenintroducetherelatedconceptsoftexturesynthesisbasedontheMRFmodel,andsketchesoutsomeofthepoint-basedalgorithmsand patch-basedalgorithms.Anewalgorithm is presented based on patch-basedalgorithms.Itweighsthesimilarityofthecolorinformation and gradient construction.The colorsandgradientsisasameasureofthesimilarityofbetweenthetwomatching patches,theextent ofthesimilarityisimproved.By comparingthesimilarityofhistogramsof thetextureblocksandhistogramsofinputsampletextureblocks,getthebestsizeoftextureblock.Improvedalgorithmhasadjustedtheselectionstrategyoftextureblocksinoriginal texture image sothattheoutputoftextures canbettermaintainthestructural characteristicsof thesampletextures.Aiming to repairing relatively big area, we put forward themethod which combined theimage inpainting and texture synthesis method. Onlyusing texture synthesis to repair theimage, there will be an unstablematching in matching texture. In order to remove this unstablephenomenon,before using texture synthesize, we estimate the basic color and gradientinformation byusing improved image inpainting. Based on this, we using synthesizingtorepair the damaged image.