基于四元数表示的彩色图像混合噪声去噪.pdfVIP

JournalofSoutheastUniversity(EnglishEdition) Vo1．31，No．3，PP．347—352 Sept．2015 ISSN 1003--7985 JlVIl●Xed1n0l●Se rem 0Val’10r C01■0r l●m ageS usingquaternionrepresentati0n WangDingcheng TianYuhang ChenBeijing，。 ShuHuazhong (JiangsuEngineeringCenterofNetworkMonitoring，NanjingUniversityofInformationScienceandTechnology， Nnajing210044，China) (SchoolofComputernadSoftware，NanjingUniversityofInformationSciencenadTechnology，Nanjing210044，China) (JiangsuCollaborativeInnovationCenteronAtmosphericEnvironmentnadEquipmentTechnology， NanjingUniversityofInfomrationSciencenadTechnology，Nnajing210044，China) (LaboratoryofImageSciencenadTechnology，SoutheastUniversiyt，Nanjing210096，China) Abstract：In orderto effectively restore colornoisy images In ordertorestorethecolornoiseimagewith themix— withthemixtureofGaussiannoiseandimpulsenoise，anew tureofGaussian noisenadimpulsenoise．many studies algorithm is proposed using hte quatemion。based holistic havebeenpublishedduringthelastfew decades．Among processing idea for color images．First， a color image is them，Xiongeta1．[21proposedhterobustoutlyingnessra— representedbyapurequatemionmatrix．Secondly，according tio．nonlocalmeans (R0R—NLM ) algorihtm．They uti． tothedifferentcharacteristicsoftheGaussina noisenad the 1izedtherobustoutlyingnessratioalgofithm todetecthte impulsenoise．na algorithm basedon quaternion directional impulse noise pixels， nad hten applied the non—local vectororder statistics isused to detectthe impulse noise． menasfiltertwicetoremoveimpulsenoiseandGaussina Finally，thequatemionoptimalweightsnon·localmeansfilter noise． respectively． Lin eta1． presented hte sorted (QOWNLMF)forGaussiannoiseremovalisimprovedforthe quadrantmedina vectorswitchingbilateralfilter(SQMV— mi