摘要
基于边界链码方法,将二值图像中的目标轮廓边界链码变换为有序数组,并对有序数组进行频域变换,获得边界有序数组的幅度谱分布,得出了边界有序数组频域幅度谱与边界链码起点选取的无关性。当目标边界比例变化及含有噪声干扰时,讨论了边界数组的幅度谱分布情况。试验结果表明,当轮廓边界链码起点不定及变形目标外形轮廓变化不大情况下,仍可以利用幅度谱相似性对目标外形进行有效的识别。
Based on the boundary chain-code, we convert the chain-code to a sequence array and then put FFT on it. Changing the start point of the chain-code, it shows the amplitude distributions keep invariant. The feature of the amplitude distribution under this state is introduced, along with changing of the boundary scale and affecting of the noises. The experiences have been given to illustrate the validity on the objects recognition with the characteristic of the amplitude distributions under the state of altering start point on the chain-code and small changing on the object contour.
出处
《计算机仿真》
CSCD
2004年第8期149-151,144,共4页
Computer Simulation
关键词
边界链码
有序数组
幅度谱
目标识别
Boundary chain-code
Sequence array
FFT Spectrum
Objects recognition
