一种采用Gabor小波的纹理特征提取方法

采用Gabor小波方法计算不同尺度和方向的能量信息,根据这些信息确定了显著峰集合。根据显著峰集合,确定了纹理特征向量,并且把显著性作为权重引入到相似性度量。72T0 XX XTo1994-2010ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.neto1994-2010ChinaAcademicJournalElectronicPublishingHouseAllrightsreservedhttp://www.cnki.net113.845230.8113.844230.78230.76113.843230.74温113842113.84123068230.66113.8400.64230.62方向(p=4)10向(p=4)436.8436.64364436温435.84358435.6435.443524310152025方向(p=3)方向(P=3)698ALa方向(P2)方向(p=2方向(p1)方向(P=1)350300制250方向(P=0)方向p=0)图5一幅颈椎间盘图像,及其在5个尺度下能量信息直方图图6一幅腰椎间盘图像,及其在5个尺度下能量信息直方图Fig 5 Energy histogram of cervical disc image at five scalesFig 6 Energy histogram of lumbar intervertebral disc at five scaleso1994-2010ChinaAcademicJournalElectronicPublishingHouse.alLrightsreservedhttp:/www.cnki.net171816161221212322122522201994-2010ChinaacademicjournalelectronicPublishingHouseAllrightsreservedhttp:www.cnkinet0.9直接 Gabor小波变换方法0.8改进 Gabor小波变换方法0.60.50.40.30.245678图像0.09直接 Gabor小波变换方法改进 Gabor小波变换方法0.07瞬回0.050.040.0I123456789图像图9针对第1组测试集的检索精度和召回率曲线Fig 9 Retrieval precision and recall for first test set0.9直接 Gabor小波变换方法改进 Gabor.小波变换方法0.70.60.40.30.20.1123456789图像0.09[-+直按 Gabor小波变换方法0.08一改进 Gabor小波变换方法0.070.060.050.040.03图10第2组测试集合0.02Fig 10 Second group of test set0.0123456789201994-2010ChinaacademicjournalelectronicPublishingHouseAllrightsreservedhttp:www.cnkinet201994-2010ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.net