基于内容的图像检索是图像处理研究的重点,而相似性度量是其核心问题。基于near集的tNM(Tolerance Nearness Measure)方法在仅提取图像的灰度值特征时比IRM(Integrated Region Matching)检索结果更好。基于tNM与人类视觉近似的特点,将...基于内容的图像检索是图像处理研究的重点,而相似性度量是其核心问题。基于near集的tNM(Tolerance Nearness Measure)方法在仅提取图像的灰度值特征时比IRM(Integrated Region Matching)检索结果更好。基于tNM与人类视觉近似的特点,将灰度值替换为面向用户视觉的HSV(Hue,Saturation,Value)颜色空间,分别提取图像的灰度值(Grey)+纹理(Texture)、HSV+纹理两组特征。使用IRM和tNM算法对10类图像进行检索,对其检索结果进行比较分析,结果表明使用tNM算法提取的图像的HSV+纹理特征与人类视觉更加近似,效果更佳。展开更多
根据CSLBP(center-symmetric local binary pattern)和Uniform LBP(local binary pattern)特征描述行人局部纹理互补性的特点,提出将二者级联的组合特征用于行人检测:基于灰度图像的纹理特征(hybrid local binary pattern,HLBP)和基于...根据CSLBP(center-symmetric local binary pattern)和Uniform LBP(local binary pattern)特征描述行人局部纹理互补性的特点,提出将二者级联的组合特征用于行人检测:基于灰度图像的纹理特征(hybrid local binary pattern,HLBP)和基于颜色空间的纹理特征(color based hybrid local binary pattern,CHLBP)。实验结果表明,当FPPW=10–4时,HLBP特征的检测率为93.96%,与Uniform LBP和CSLBP特征相比分别提高3.46%和9.68%,基于颜色空间L′C′C′与HIKSVM分类器结合时的检测率高达98.58%。与传统的纹理特征检测方法相比,该特征提高了行人检测精度,降低了误检率,检测性能得到较大幅度的提升。展开更多
针对复杂工况环境下异常条形码的识别问题,提出了一种新的基于特征提取与BP神经网络协同作用的异常条形码判别方法。首先,为有效对条形码图像特征进行表征,从图像histogram of oriented gradient(HOG)特征、曲线特征、纹理粗糙度、纹理...针对复杂工况环境下异常条形码的识别问题,提出了一种新的基于特征提取与BP神经网络协同作用的异常条形码判别方法。首先,为有效对条形码图像特征进行表征,从图像histogram of oriented gradient(HOG)特征、曲线特征、纹理粗糙度、纹理灰度特征着手,建立条形码识别的特征库;在此基础上,建立以LM-BP神经网络为核心的辨识框架对条形码特征进行训练和辨识;最后,通过模拟国网新疆电力有限公司电力科学研究院计量生产自动化系统现场的条形码图像验证了算法的合理性。实验结果表明:基于特征提取与LM-BP神经网络协同辨识的方法能有效对条形码状态进行识别,其识别精度可达88. 29%。展开更多
This paper presents a method to characterize asphalt pavement macrotexture using the gray-tone difference matrix （GTDM）and discusses the potentials of the GTDM indicators for skid resistance evaluation.There are 37 ...This paper presents a method to characterize asphalt pavement macrotexture using the gray-tone difference matrix （GTDM）and discusses the potentials of the GTDM indicators for skid resistance evaluation.There are 37 field sites included in the data collection,which cover 6 types of asphalt pavement surfaces. The mean profile depth derived from 3-D macrotexture measurements （MPD3 ） has a significant relationship with the mean texture depth （MTD ）,which can be described by a logarithm model with R2 of 0.962.There is no significant linear relationship between the friction coefficient at a speed of 60 km/h （DFT60 ）and macrotexture indicators.A nonlinear model with British pendulum number （BPN ） incorporated can relate DFT60 to MTD or indicator fcon .A comparison with MTD shows that GTDM-based fcon has a potential to be a macrotexture indicator for skid resistance evaluation,which describes the general height difference and the average local height difference of pavement macrotexture. A relatively high fcon is helpful for improving asphalt pavement skid resistance.展开更多
文摘基于内容的图像检索是图像处理研究的重点,而相似性度量是其核心问题。基于near集的tNM(Tolerance Nearness Measure)方法在仅提取图像的灰度值特征时比IRM(Integrated Region Matching)检索结果更好。基于tNM与人类视觉近似的特点,将灰度值替换为面向用户视觉的HSV(Hue,Saturation,Value)颜色空间,分别提取图像的灰度值(Grey)+纹理(Texture)、HSV+纹理两组特征。使用IRM和tNM算法对10类图像进行检索,对其检索结果进行比较分析,结果表明使用tNM算法提取的图像的HSV+纹理特征与人类视觉更加近似,效果更佳。
文摘根据CSLBP(center-symmetric local binary pattern)和Uniform LBP(local binary pattern)特征描述行人局部纹理互补性的特点,提出将二者级联的组合特征用于行人检测:基于灰度图像的纹理特征(hybrid local binary pattern,HLBP)和基于颜色空间的纹理特征(color based hybrid local binary pattern,CHLBP)。实验结果表明,当FPPW=10–4时,HLBP特征的检测率为93.96%,与Uniform LBP和CSLBP特征相比分别提高3.46%和9.68%,基于颜色空间L′C′C′与HIKSVM分类器结合时的检测率高达98.58%。与传统的纹理特征检测方法相比,该特征提高了行人检测精度,降低了误检率,检测性能得到较大幅度的提升。
文摘针对复杂工况环境下异常条形码的识别问题,提出了一种新的基于特征提取与BP神经网络协同作用的异常条形码判别方法。首先,为有效对条形码图像特征进行表征,从图像histogram of oriented gradient(HOG)特征、曲线特征、纹理粗糙度、纹理灰度特征着手,建立条形码识别的特征库;在此基础上,建立以LM-BP神经网络为核心的辨识框架对条形码特征进行训练和辨识;最后,通过模拟国网新疆电力有限公司电力科学研究院计量生产自动化系统现场的条形码图像验证了算法的合理性。实验结果表明:基于特征提取与LM-BP神经网络协同辨识的方法能有效对条形码状态进行识别,其识别精度可达88. 29%。
基金The National Natural Science Foundation of China(No.50908004,51178013)
文摘This paper presents a method to characterize asphalt pavement macrotexture using the gray-tone difference matrix （GTDM）and discusses the potentials of the GTDM indicators for skid resistance evaluation.There are 37 field sites included in the data collection,which cover 6 types of asphalt pavement surfaces. The mean profile depth derived from 3-D macrotexture measurements （MPD3 ） has a significant relationship with the mean texture depth （MTD ）,which can be described by a logarithm model with R2 of 0.962.There is no significant linear relationship between the friction coefficient at a speed of 60 km/h （DFT60 ）and macrotexture indicators.A nonlinear model with British pendulum number （BPN ） incorporated can relate DFT60 to MTD or indicator fcon .A comparison with MTD shows that GTDM-based fcon has a potential to be a macrotexture indicator for skid resistance evaluation,which describes the general height difference and the average local height difference of pavement macrotexture. A relatively high fcon is helpful for improving asphalt pavement skid resistance.
