摘要
为满足智能寻位加工中对零件图像配准速度和配准精度的高要求,提出了一种采用伪极快速傅里叶变换(PPFFT)和小世界-克隆选择算法(SWCSA)的图像配准方法(PPFFT-SWCSA).首先对图像进行伪极快速傅里叶变换,然后利用变换后获取的频谱特征信息设计优化算法的代价函数,最后采用SWCSA算法得到2幅图像间的配准参数.采用PPFFT降低了运算复杂性,提高了运算速度;采用SWCSA算法,克服了图像配准中常用的相位相关法无法检测到图像间较小偏移量的缺点.实验结果表明,PPFFT-SWCSA方法的配准角度精度可以达到空间0.2°,在对图幅为256×256像素的图像配准实验中,PPFFT-SWCSA方法的配准速度比基于离散极坐标傅里叶变换和相位相关法的配准速度快2倍.
Owing to the demanding accuracy and speed for image registration in intelligent searching machining system, a registration method based on pseudo-polar fast Fourier transform （PPFFT） and small world clonal selection algorithm （SWCSA） is proposed. Firstly, PPFFT is done for the image. Secondly, a suitable fitness function is selected by using the magnitude of PPFFT. Lastly, registration parameters are obtained by SWCSA. With PPFFT, the complexity of computation is significantly lowered and the speed is accelerated. With SWCSA, the disadvantage that small translation can not he detected by phase correlation is eliminated. Experimental results demonstrate that the registration accuracy of rotation is up to 0.2°. In the 256 × 256 pixels image registration experiment, the speed of the proposed method is two times faster than that of discrete polar Fourier transform and phase correlation algorithm.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2009年第6期38-42,共5页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(50705073)
陕西省自然科学基金资助项目(2007E224)
江苏省自然科学基金资助项目(BK2008184).
关键词
图像配准
伪极快速傅里叶变换
克隆选择
小世界
image registration
pseudo-polar fast Fourier transform
clone selection
small world
