【摘要】 讨论在识别目标前不对图像进行几何变形的预处理,而是在图像上对小车和球的位置识别成功后,通过插值算法把目标坐标映射为实际场地的位置坐标。该方法避免了把场地内所有的点都进行坐标变换的图像预处理过程,从而极大地减少了计算量。更多还原

【Abstract】 This paper describes an interpolation method that can acquire the target coordinates as soon as cars and ball are identified.This method does not need image aberration emendation and avoids coordinate transformation for all the image pixels,so that it significantly reduces the calculation.更多还原