针对低剂量计算机断层扫描(CT)重建图像时出现明显条形伪影的现象,提出一种自适应广义总变分(ATGV)降噪算法。该算法考虑了传统广义总变分(TGV)算法在降噪时模糊图像边缘信息的缺点,把可以有效区分图像平滑区和细节区的直觉模糊熵应用...针对低剂量计算机断层扫描(CT)重建图像时出现明显条形伪影的现象,提出一种自适应广义总变分(ATGV)降噪算法。该算法考虑了传统广义总变分(TGV)算法在降噪时模糊图像边缘信息的缺点,把可以有效区分图像平滑区和细节区的直觉模糊熵应用到传统TGV中,对图像的不同区域进行不同强度的去噪,从而达到保护图像细节的效果。该算法首先采用滤波反投影(FBP)算法得到低剂量CT重建图像;然后利用基于直觉模糊熵的边缘指示函数对传统TGV模型进行改进;最后用改进后的模型对重建图像进行降噪处理。采用Shepp-Logan模型和数字胸腔模型(thorax phantom)仿真低剂量CT重建图像来验证算法的有效性。实验结果表明,所提算法的归一化均方距离(NMSD)和归一化平均绝对距离(NAAD)均比总变分(TV)降噪算法和广义总变分(TGV)降噪算法小,且可分别获得26.90 d B和44.58 d B的峰值信噪比(PSNR)。该算法在去除条形伪影的同时可以较好地保持图像的边缘和细节信息。展开更多
Interferogram noise reduction is a very important processing step in Interferometric Synthetic Aperture Radar(InSAR) technique. The most difficulty for this step is to remove the noises and preserve the fringes simult...Interferogram noise reduction is a very important processing step in Interferometric Synthetic Aperture Radar(InSAR) technique. The most difficulty for this step is to remove the noises and preserve the fringes simultaneously. To solve the dilemma, a new interferogram noise reduction algorithm based on the Maximum A Posteriori(MAP) estimate is introduced in this paper. The algorithm is solved under the Total Generalized Variation(TGV) minimization assumption, which exploits the phase characteristics up to the second order differentiation. The ideal noise-free phase consisting of piecewise smooth areas is involved in this assumption, which is coincident with the natural terrain. In order to overcome the phase wraparound effect, complex plane filter is utilized in this algorithm. The simulation and real data experiments show the algorithm can reduce the noises effectively and meanwhile preserve the interferogram fringes very well.展开更多
针对多通道彩色图像放大问题,文中建立二阶TGV(Total Generalized Variation)图像放大模型,并利用交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)求解。在RGB彩色空间上,针对每个彩色通道分别进行放大处理,进而放...针对多通道彩色图像放大问题,文中建立二阶TGV(Total Generalized Variation)图像放大模型,并利用交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)求解。在RGB彩色空间上,针对每个彩色通道分别进行放大处理,进而放大彩色图像。数值结果表明,与原始对偶算法相比,无论是视觉效果还是定量比较,基于二阶TGV的ADMM算法均取得了更好的放大效果。展开更多
文摘针对低剂量计算机断层扫描(CT)重建图像时出现明显条形伪影的现象,提出一种自适应广义总变分(ATGV)降噪算法。该算法考虑了传统广义总变分(TGV)算法在降噪时模糊图像边缘信息的缺点,把可以有效区分图像平滑区和细节区的直觉模糊熵应用到传统TGV中,对图像的不同区域进行不同强度的去噪,从而达到保护图像细节的效果。该算法首先采用滤波反投影(FBP)算法得到低剂量CT重建图像;然后利用基于直觉模糊熵的边缘指示函数对传统TGV模型进行改进;最后用改进后的模型对重建图像进行降噪处理。采用Shepp-Logan模型和数字胸腔模型(thorax phantom)仿真低剂量CT重建图像来验证算法的有效性。实验结果表明,所提算法的归一化均方距离(NMSD)和归一化平均绝对距离(NAAD)均比总变分(TV)降噪算法和广义总变分(TGV)降噪算法小,且可分别获得26.90 d B和44.58 d B的峰值信噪比(PSNR)。该算法在去除条形伪影的同时可以较好地保持图像的边缘和细节信息。
文摘Interferogram noise reduction is a very important processing step in Interferometric Synthetic Aperture Radar(InSAR) technique. The most difficulty for this step is to remove the noises and preserve the fringes simultaneously. To solve the dilemma, a new interferogram noise reduction algorithm based on the Maximum A Posteriori(MAP) estimate is introduced in this paper. The algorithm is solved under the Total Generalized Variation(TGV) minimization assumption, which exploits the phase characteristics up to the second order differentiation. The ideal noise-free phase consisting of piecewise smooth areas is involved in this assumption, which is coincident with the natural terrain. In order to overcome the phase wraparound effect, complex plane filter is utilized in this algorithm. The simulation and real data experiments show the algorithm can reduce the noises effectively and meanwhile preserve the interferogram fringes very well.
文摘针对多通道彩色图像放大问题,文中建立二阶TGV(Total Generalized Variation)图像放大模型,并利用交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)求解。在RGB彩色空间上,针对每个彩色通道分别进行放大处理,进而放大彩色图像。数值结果表明,与原始对偶算法相比,无论是视觉效果还是定量比较,基于二阶TGV的ADMM算法均取得了更好的放大效果。
