摘要
研究和分析了伪Zernike矩的位移及尺度不变特性。利用此特性,对图像识别中的一种情况,即不同位移及不同尺度下同一目标的识别提出了新的方法。通过计算目标图像的高阶伪Zernike矩模值,再利用聚类方法进行分类。实验结果表明,伪Zernike矩对于不同尺度目标的识别具有很好的作用。
Translation invariance and scale invariance of pseudo-Zernike moments are studied. Using this feature, a new method for identifying the same kind of objects under different scale or translation is proposed. Firstly, the magnitude values of high order pseudo-Zernike moments are calculated for object images, and then they are classified by clustering methods. Experiment results show that pseudo-Zernike moments play a good role in identifying objects under different scale.
出处
《计算机工程与设计》
CSCD
2004年第11期1964-1965,共2页
Computer Engineering and Design
