贴装元件的方差环形投影特征匹配识别

为了准确定位贴片后印刷电路板（PCB）上电子元件的位置，提出了一种基于方差环形投影的图像快速匹配算法。先将输入图像用小波变换分解为不同尺度的低频图像，之后利用方差环形投影特征做序贯相似性检测得到一系列可能的匹配点，最后在这些点上作相关匹配运算，获得其准确的位置。实验证明该算法大大减少了计算量又保持了匹配的精度，而且对图像的旋转不敏感。

Matching recognition of mounted components based on variance ring projection feature

To locate the mounted electricity components in the Printed Circuit Board(PCB),a fast image matching algorithm based on variance ring projection is proposed. Firstly, an input image is decomposed into different low levels sub-image with the walvelet-transform.Then, after getting the variance ring projection feature,the Sequential Similarity Detection Algorithm(SSDA) is exploited to represent some potential matching points. Finally, based on these points, a precisely locating phase using the correlation is introduced to get the right position. Experimental results show this method significantly reduces the computational burden and is rotation-invariant, while the precision remained fine.