果实缺陷面积的计算机视觉测定研究

Computer Vision Method for Measuring Defect Area of SphereFruits

摘要: 以提高球形果实表面缺陷面积的计算机视觉测定精度为目的，分析了利用投影图像直接测定球形果实表面缺陷存在的问题；提出并建立了从投影图像恢复球形果实表面几何特征的象素点变换法和边界变换法，并通过试验进行验证。结果表明，建立的两种方法可使测定相对误差减小３５％左右。

Abstract: The purpose of this paper is to improve the measuring accuracy of surface defect area of spherefruit by computer vision method.The problem of determining the defect area by umbriferous image was analysed. The methods of pixel transform and boundary transform that can recover the geometrical feature of spherefruit surface defect from umbriferous image were established,and verified by experiment.The results showed that both methods can reduce the relative errors by about 35 %.