双目视觉用于鱼苗尺寸测量

为了有效地提高鱼苗尺寸测量的效率及精度,搭建了平行式双目立体视觉系统。根据双目视觉原理,首先基于改进后的由多组同心圆构成的标定板对双目视觉测量系统进行定标;然后利用Harris角点提取算法获取鱼苗图像的关键特征点,并基于归一化互相关(Normalized Cross Correlation,NCC)立体匹配算法对关键特征点进行匹配,提取坐标数据;最后根据坐标数据进行计算,得到鱼苗的空间坐标,实现对鱼苗尺寸的精确测量。结果表明,由所搭建的平行式双目立体视觉系统测量的相对误差在8%以内,为双目视觉应用于鱼类养殖业的可行性提供了依据。

Stereo vision for measuring the size of fish

To improve the efficiency and accuracy of measuring the size of fish,a parallel binocular stereo vision system is established to get the size of fish.According to the principle of binocular vision,first of all,to make use of the synthetic images of concentric circles as the calibration board,the location of the cameras is got.Next based on Harris corner detection,the key feature points of the image of fish are obtained.Then using the Normalized Cross Correlation (NCC) stereo matching algorithm,matching the feature points are got,and coordinate data of the matching points are ex tracted.In the end,the three-dimensional coordinate values are calculated,which is from the matching points,and the size of fish is got.The results show that the accuracy of measurement methods is about 8％.The evidence is provide for applving stereo vision to the fish farming industry.