基于机器视觉的沙糖橘果皮破裂和缺陷检测

Detection of Broken Peel and Imperfection of Shatangju Based on Machine Vision

在对果皮质量进行检测分级时,破裂果、机械损伤和硬疤是沙糖橘表面常见的主要缺陷。传统的自动检测系统通常不能准确地识别这些表面缺陷。为了解决这一问题,设计了一种硬件和软件相结合的计算机视觉检测分类系统。该系统采用单CCD和LED环形光源,通过计算机协作,利用计算机视觉系统提取沙糖橘果皮的正面图像,构建了有效的图像采集方法、预处理方法、颜色模型和分割方法,采用傅里叶变换、高频滤波、形态学(方案)和分类树等方法对沙糖橘的表面缺陷进行研究,并为实际的自动化应用找到更准确和更合适的方法。结果表明,该方法的可靠性和稳定性优于传统的单一形态学的识别方法。

When grading and detecting fruit quality,the broken fruit,mechanical damage and hard scar are the common main defects on the surface of sugar orange.The traditional automatic inspect system often cannot accurately distinguish these surface defects.To solve this problem,we designed a computer vision detection and classification system which combined hardware and software.The system used a single CCD and a ring LED light source through computer cooperation,used computer vision system to extract the front image of sugar orange peel,and constructed an effective image acquisition method,preprocessing method,color module and segmentation method.We used the method(scheme)of Fourier transform,high-frequency filtering and morphology and classification trees and so on to study the surface defects of sugar orange,and found a more accurate and suitable method for the practical application of automation.The results showed that the reliability and stability of this peel recognition method were better than the traditional recognition method using a single morphology.