DAMAGE FRUITS INSPECTING SYSTEM BY MACHINE VISION

在水果包装车间，水果的采后处理和包装已广泛自动化，但分拣操作仍然是人工操作。因此，水果缺陷分拣的自动化除了降低包装车间的人工成本外，还具有提高产品质量的潜力。本文研究了水果表面缺陷的光谱反射特性，为水果缺陷的机器视觉分选系统的开发奠定了基础。果实缺陷的类型是瘀伤，割伤，褐腐。压缩和冲击损伤。在可见波长区域（290 780），桃表面的颜色变化使桃缺陷的检测条件复杂化。而桃红和底色曲线的光谱反射率基本相同，可以明显区分桃红表面的正常和损伤。在损伤水果检测研究中，尝试开发一种基于彩色图像的NTSC r、g、b色度和红外图像灰度的图像分析算法来识别缺陷。利用判别分析方法和分层算法，建立了基于正常和缺陷灰度值差异的桃缺陷检测算法。研究了两种检测缺陷的方法，如蠕虫、疤痕、擦伤和裂口。任何人都可以准确地从桃子图像中分离出缺陷。

In the fruits packinghousep the postharvest handling and packaging of fruits has been extensively automated, with the exception of the sorting operation, which continues to be a manual effort. Consequently, automation of the fruit defect sorting has potential for improving product quality, in addition to reducing packinghouse labor costs.This study dealt with the spectral reflectance characteristics of fruitsurface defects in order to the development of a machine vision sorting system for fruit defects. And the types of fruit defects were bruises, cut, brown rots. copressed and impacted damages.Detecting condition of peach defects in the visible wavelength region(290 780) were complicated by the variation in color over the surface of the peach. But blush and ground color curves had about same values of spectral reflectance drew the clear distinction between normal and damage of peach surface.The study on damage fruits inspection dealt with a trial of developing an image analysis algolithms based on NTSC r, g, b chromatictity in color image and the gray level of infrared image to identify defects.The detecting algolism of defects of peach was developed on the difference of the gray-level between normal and defects, utilizing the method of discriminat analysis and superlices algolithm. Two method had to be investigated in detecting the defects such as worm, scar, burises and aracle. Any one can exactl split the defects from the peach image.