ABSTRACT In engineering and agriculture, morphological properties are the important characters in studying physical properties of the products. The use of image processing and machine vision technique to measure appearance and morphological properties of objects has become very common in recent years. The purpose of this study was to predict the diameters, area and perimeter of spherical objects using image processing and nonlinear multivariate modeling. The hardware part of the system included an imaging station, a back lighting source, a camera, a frame grabber and a personal computer. The distance from the objects to the camera was considered to determine the characteristics. An algorithm in MATLAB software was developed to model the characteristics based on number of pixels and the distance from the object to the camera. The results showed that the system can determine the perimeter, projected area and the major and minor diameters with high accuracy (>99%) and prediction error of 0.1113, 0.1989, 2.0721 and 0.6953%, respectively. Based on the modeling results, the procedure is a promising technique to measure the morphological properties of the fruits and vegetables.

中文翻译：

---

一种利用图像处理技术和非线性多元建模确定果蔬形态特性的新方法

摘要 在工程和农业中，形态特性是研究产品物理特性的重要特征。近年来，使用图像处理和机器视觉技术来测量物体的外观和形态特性已变得非常普遍。本研究的目的是使用图像处理和非线性多元建模来预测球形物体的直径、面积和周长。该系统的硬件部分包括一个成像站、一个背光源、一个摄像头、一个图像采集卡和一台个人电脑。考虑从物体到相机的距离来确定特征。MATLAB 软件中的算法被开发用于基于像素数和物体到相机的距离对特征进行建模。结果表明，该系统能够以较高的精度（>99%）确定周长、投影面积和长短径，预测误差分别为0.1113、0.1989、2.0721和0.6953%。基于建模结果，该程序是测量水果和蔬菜形态特性的一种很有前途的技术。