本文针对纹理分类提出了一种有效的本地运算符,即LocalQuantizationCode(LQC)。 Theconventionallocalbinarypatterncanberegardedasaspeciallocalquanti-.zation methodwithtwolevels,0and1.SomevariantsoftheLBPdemonstratethatincreasingthelocal.quantizationlevelcanenhancethelocaldiscriminativecapability.Hence,wepresentasimpleand.unified frameworktovalidatetheperformanceofdifferentlocalquantizationlevels.IntheproposedLQC,.pixelslocatedindifferentquantizationlevelsareseparatelycountedandtheaveragelocalgrayvalue.difference isadoptedtosetaseriesofquantizationthresholds.Extensiveexperimentsarecarriedouton.severalchallengingtexturedatabases.TheexperimentalresultsdemonstratetheLQCwithappropriate.local quantizationlevelcaneffectivelycharacterizethelocalgray-leveldistribution。
2021-04-05 12:06:04
1024KB
研究论文
1