作　　者： ; ; ; ;

机构地区： 深圳大学信息工程学院

出　　处： 《中国科学：信息科学》 2011年第12期1423-1435,共13页

摘　　要： 采用Clifford代数理论处理多光谱图像不同于传统代数的处理方法,其可有效利用光谱层之间的联系信息.本文通过分析Clifford微分,提出了多光谱图像的Clifford拟微分算子,丰富了多光谱图像处理的Clifford代数理论体系.在边缘识别仿真实验中,与最大熵法相比较,本文提出的以Clifford拟微分算子理论为核心的边缘检测算法识别模糊边界能力更强.在医学CT影像辅助诊断边缘检测实验中,新算法能有效凸显病灶,提高了诊疗准确性和诊断效率. Different from the classic algebra processing on multi-spectral images, Clifford algebra can effectively utilize the information associated with different spectrum layers. In this paper, according to analyzing Clifford differentiation, a systematic Clifford pseudo-differential operator theory is proposed, which enriches the theory of Clifford algebra of multi-spectral image processing. Compared with the maximum entropy edge detection algo- rithm, the simulation results indicate that the proposed method in term of Clifford pseudo-differential operator theory is better at identifying fuzzy edge. Furthermore, the experiment of CT images accessory diagnostic shows that the new algorithm can obviously reveal the lesion, which leading to the more correct and effective diagnosis.

关 键 词： 多光谱图像 代数 拟微分 边缘检测

领　　域： [自动化与计算机技术] [自动化与计算机技术]