摘要
由高速线阵CCD相机基于交汇原理组成的弹丸立靶坐标、射击密集度测试系统,其测试精度与弹丸图像的边缘位置密切相关,弹丸穿过大靶面的边缘区域时成像模糊,其灰度值呈连续变化状态,常用的图像边缘检测算子已经不能适应这种情况。对8m×8m的靶面,一个像素就会在物方最大引起±10mm的测试误差。将Sobel边缘检测算子的粗检测定位与三次样条函数差值的细检测相结合,先用Sobel算子粗检测,再构建三次样条函数,利用三次样条插值对弹丸图像边缘进行亚像素级细分。实弹射击试验结果证明,这种方法能够将系统的最大测试误差降低到±2.5mm。
The test system of projectile target coordinate and shooting intensity based on high speed linear CCD intersecting principle, its testing precision and edge position of projectile image are closely interrelated. When the projectile passes through an edge region of big target, the image is fuzzy and the gray value display continuous change state. So the commonly operator for image edge detection is unsuitable to the situation. For 8 m × 8 m target, it can make ± 10 mm detection error on one pixel. Approximately orienting in Sobel edge detection operator combines with detailed detection on three-order spline interpolation. Firstly, the image is detected approximately by Sobel operator, and then made up three-order spline function, the three-order spline interpolation can make subpixel subdivision for the edge of projectile image. The effect is made clear that the max-detection error is ± 2.52 mm.
出处
《科学技术与工程》
2008年第3期634-637,642,共5页
Science Technology and Engineering
关键词
亚像素细分
三次样条插值
多尺度边缘检测
线阵CCD光靶
sub-pixel subdivision CCD optical target three-order spline interpolation muhiscale edge detection linear
