Zernike矩与MSA矩的对比研究

A Comparative Study of Zernike Moment and MSA Moment

Zernike矩由于其信息冗余度低且对噪声不敏感,在图像特征提取和模式识别中得到了广泛的应用,MSA矩具有尺度不变性和仿射不变性,被成功应用于单纯背景的图像模式识别领域,也被越来越多的学者所关注,但Zernike矩与MSA矩的不变性、时间效率及应用条件等还没有被对比过;同时,Zernike矩本身只具有旋转不变性,当前对于Zernike矩平移不变性、尺度不变性已经有了较好的解决方法,但对仿射变化的研究还较少.针对上述两个问题,首先提出了一种同时解决Zernike矩平移不变性、尺度不变性和仿射不变性的预处理算法,其次对Zernike矩与MSA矩的仿射不变性、时间效率等进行了详细的对比.仿真结果表明提出的预处理算法可以有效地保证Zernike矩的仿射不变性,其次,经过预处理后的Zernike矩与MSA矩的对比研究可以对以后的研究提供详实的参考.

Zernike moment due to its low redundancy and its insensitivity to noise has been widely used in image feature extraction and pattern recognition. MSA moment has scale invariant and af- fine invariant characteristics and has been successfully applied to the pure background image pat- tern recognition field,which makes MSA moment more and more attractive to associate scholars. But the invariance, time efficiency and application condition difference between Zernike moment and MSA moment have not been compared in detail by now. Meanwhile, few studies about Zernike moment＇s affine invariance have been published. According to the above two problems, firstly, a preprocessing algorithm is proposed to guarantee Zernike moment＇s translational invariance,scale invariance and affine invariance at the same time. Secondly, the invariance and time efficiency of Zernike moment and MSA moment are compared in detail. Simulation results show that the pro- posed preprocessing algorithm can effectively ensure the translational invariance, scale invariance and affine invariance of Zernike moment. Meanwhile, the comparison results of Zernike moment and MSA moment would be helpful to further research.