水果尺寸和面积的机器视觉检测方法研究

针对我国水果品质检测仍停留在靠人工感官进行识别判断的现状和机器视觉技术在水果品质检测中的广阔应用前景，研究了利用机器视觉技术精确检测水果尺寸和表面缺陷面积的方法，建立了图像中的点与被测物体上的点之间的定量关系；提出了利用物体的边界信息求出物体的形心坐标的新方法。结果是：所测水果最大横径与实际最大横径的相关系数为0.96；采用像素点变换法，实现了根据三维物体的二维投影图像恢复物体表面的真实几何面积的设想；提出了一种新的面积修正方法，进一步提高了面积检测的精度，从而为研究开发机器视觉水果品质检测系统打下了基础。

Research in method to detect size and area of fruits by machine vision.

In view of the existing situation of fruits quality detection in our country, which is still dependent on human sense organ to identify and judge the fruits, and the broad application prospect of machine vision in quality evaluation of agricultural products, the method to detect the size and area of agricultural products b y machine vision was studied. The quantitative relationship between one pixel in the image and the corresponding size and area in the real object was set up, and the new method to calculate the centroid coordinates of the object only based on the boundary information was put forward, which not only offered fast computation because visiting the whole area of the object was not needed, but also provided more accurate estimation since the image texture within the object, such as blemishes, was eliminated from the consideration. It was found that the correlation coefficient of real size versus detected size was 0.96. To decrease the relative errors, the pixel transform method was adopted to recover the geometrical feature of sphere fruit surface from umbriferous image while the area of defected surface was calculated. Moreover, a new method to revise the estimated area was advanced. These results lay a solid foundation for further developing fruit quality detection system using machine vision.