摘要
在工业产品检测中,传统模板匹配方法不适用于元件标志的旋转情况。针对该问题提出基于圆投影与矢量和的旋转不变量检测方案,采用构造模板的方法,乘以模板中的系数实现灰度图像圆周投影统计和一维特征向量矢量圆求和的功能。实验结果证明,该方案可以实现目标发生旋转后的标志匹配,其抗噪能力和处理速度高于SIFT算法。
Traditional template matching methods are not proper for the rotation situation of component logo in the industrial production detection. Aiming at this problem, this paper proposes a rotation invariance detection scheme based on circular projection and vector sum. Through making mask of which coefficient is multiplied by the corresponding pixel, this method realizes function of circular projection statistics of grayscale image and 1 D eigenvector vector circularity summing. Experimental results demonstrate that this method can realize symbol matching after the target is rotated and has higher noise suppression ability and detection speed than SIFT algorithm.
出处
《计算机工程》
CAS
CSCD
北大核心
2009年第21期181-182,185,共3页
Computer Engineering
基金
国家自然科学基金资助项目"基于二值结构光的芯片凸点三维检测中的编码策略优化"(60806050)
关键词
圆投影
矢量和
旋转不变量
模板匹配
circular projection
vector sum
rotation invariance
template matching
