摘要
在目标的识别中,归一化的傅立叶描述子具有旋转,平移和尺度变换的不变性。在传统傅立叶算子的基础上,通过边界离散过程的采样点进行近似量级的归一化,来达到简化计算和消除大尺度变换离散误差的目的,并运用于军机模型的识别上,取得比较好的效果。
The normalized fourier descriptors are invariant in rotation, shift and scale transform in target recognition. The sampling point is approximately normalized by boundary scatter process in order to simplify calculations and avoid discrete error of large scale transformation. The new method is used to recognize aircraft models and some good results are achieved.
出处
《湖南工业大学学报》
2007年第5期95-98,共4页
Journal of Hunan University of Technology
关键词
边缘检测
边界跟踪
近似归一化
傅立叶算子
edge detection
boundary tracking
approximately normalization
fourier descriptor
