[发明专利]一种基于圆盘靶标的角点检测方法

权利要求书

1.一种基于圆盘靶标的角点检测方法，其特征在于，包括以下步骤：

步骤一：对摄像机获得的一帧图像进行滤波处理去除噪声干扰，获得预处理图像I(i,j)；

步骤二：用预处理图像I(i,j)内每个像素点的灰度值分别与摄像机获得的黑白圆盘靶标的纵向对称算子H、横向对称算子H^T、倾斜45°的纵向对称算子S、倾斜45°的横向对称算子S^T进行卷积，获得每个像素点在所述四种对称算子下的响应值，即纵向对称算子H下的响应值R_H(i,j)、横向对称算子H^T下的响应值倾斜45°的纵向对称算子S下的响应值R_S(i,j)、倾斜45°的横向对称算子S^T下的响应值

步骤三：根据步骤二获得的响应值，分别用四种选取条件选取符合各自条件的候选角点，求取候选角点数量最多条件下的所有候选角点坐标的平均值，将该平均值作为检测到的最终角点的坐标，完成该帧图像的角点检测，所述四种选取条件为：

条件一：RH(i,j)≥ThmaxandRHT(i,j)≤Thmin,]]>

条件二：RS(i,j)≥ThmaxandRST(i,j)≤Thmin,]]>

条件三：RH(i,j)≤ThminandRHT(i,j)≥Thmax,]]>

条件四：RS(i,j)≤ThminandRST(i,j)≥Thmax,]]>

其中，Th_max为四个对称算子中任意对称算子所检测的全部白色像素点的灰度值总和，Th_min为四个对称算子中任意对称算子所检测的全部黑色像素点的灰度值总和；

步骤四：以步骤三检测到的最终角点为中心建立大小为K*K的窗口邻域，比较该窗口邻域内白色像素点数量与黑像素点数量之间的差值，如果该差值小于预先设定的检测阈值，则步骤三检测到的最终角点是该帧图像的准确角点，否则，步骤三检测到的最终角点不是该帧图像的准确角点，并将其剔除，K的取值根据摄像机获得的圆盘靶标的图像大小确定。

2.如权利要求1所述的基于圆盘靶标的角点检测方法，其特征在于，所述步骤二中，

纵向对称算子H=······111···············1·······································0·······································1···············111······,]]>

预处理图像I(i,j)内每个像素点的灰度值在纵向对称算子H下的响应值

RH(i,j)=I(i,j)*H=Σm=-NNΣn=-NNI(i+m,j+n)·H(m+N+1,n+N+1);]]>

横向对称算子HT=··········································1···············111···0···111···············1··········································,]]>

预处理图像I(i,j)内每个像素点的灰度值在横向对称算子HT下的响应值

RHT(i,j)=I(i,j)*HT=Σm=-NNΣn=-NNI(i+m,j+n)·HT(m+N+1,n+N+1);]]>

倾斜45°的纵向对称算子S=···············11···············11······························0······························11···············11···············,]]>

预处理图像I(i,j)内每个像素点的灰度值在倾斜45°的纵向对称算子S下的响应值

RS(i,j)=I(i,j)*S=Σm=-NNΣn=-NNI(i+m,j+n)·S(m+N+1,n+N+1);]]>

倾斜45°的横向对称算子ST=11···············11·············································0·············································11···············11,]]>

预处理图像I(i,j)内每个像素点的灰度值在倾斜45°的横向对称算子S^T下的响应值

RST(i,j)=I(i,j)*ST=Σm=-NNΣn=-NNI(i+m,j+n)·ST(m+N+1,n+N+1).]]>