基于机器视觉的蓝莓果实分级检测算法

Blueberries grading detection algorithm based on machine vision

为了实现蓝莓果实的无损检测和自动分级,提出一种基于机器视觉的蓝莓果实分级检测算法。应用Open CV,Visual Studio 2015开发环境编写算法程序,并搭建了一套基于机器视觉的蓝莓果实自动分级检测实验装置。通过实验装置的CCD摄像头获取蓝莓果实图像后,应用图像处理技术对其进行降噪、HSV加权运算、开运算、二值化处理及边缘检测,获取蓝莓果实图像的边缘像素点集。对边缘像素点集进行椭圆拟合后,应用最小二乘法确定蓝莓果实的实际值与计算机测量值之间的数学关系,求取蓝莓果实直径。为了验证算法的可行性,使用数字式游标卡尺对蓝莓果实的测量值与检测装置的测量值进行实验对比分析。实验结果表明,测量误差低于3.1%,验证了算法的有效性。

A new blueberries grading detection algorithm based on machine vision is proposed to realize the non-destructive detection and automatic grading of blueberries,and a set of experiment device for it is established. The Open CV and Visual Studio 2015 development environment are used to compile the algorithm program. The blueberries image is acquired by means of CCD camera of the experiment device,for which the denoising,HSV weighting operation,open operation,binarization and edge detection are performed to obtain the edge pixel point set of blueberries image. The ellipse fitting is conducted for edge pixel point set,and then the least square method is adopted to determine the mathematic relation between the actual value and computer measured value of the blueberries to get the diameter of blueberries. The sizes of blueberries measured by digital vernier caliper and detection device were compared and analyzed in experiment to verify the feasibility of the algorithm. The experimental results show that the measurement error is less than 3.1%,which can meet the practical requirement.