FOOD SCIENCE ›› 2021, Vol. 42 ›› Issue (18): 232-239.doi: 10.7506/spkx1002-6630-20200719-255

• Safety Detection • Previous Articles     Next Articles

Non-destructive Firmness Detection and Grading of Bunches of Red Globe Grapes Based on Machine Vision

SHI Hang, WANG Qiaohua, GU Wei, WANG Xianbo, GAO Sheng   

  1. (1. College of Engineering, Huazhong Agricultural University, Wuhan 430070, China; 2. Key Laboratory of Agricultural Equipment in Mid-Lower Yangtze River, Ministry of Agriculture and Rural Affairs, Wuhan 430070, China)
  • Published:2021-09-29

Abstract: A machine vision-based method for the non-destructive firmness detection and grading of whole bunches of Red Globe grapes was proposed in this study. Red, green and blue (RGB) and near infrared images of the samples were acquired with an industrial camera. The three color channels in the RGB images were extracted, and the normalized GB color difference method was applied to locate the fruit stems based on the RGB images. Meanwhile, the center of mass of each fruit in a bunch was located by morphological reconstruction based on local maximum brightness and the center of mass of the whole bunch was extracted. The area of the fruit stalk, the ratio of the number of grapes to the area of the whole bunch of grapes, and the ratio between the sum of distance from the surface of each grape to the center of mass of the whole bunch of grapes and the number of grapes were selected as characteristic parameters to establish a classification model for predicting the firmness of grape bunches using linear discriminant analysis (LDA), integrated learning algorithm (ILA) or support vector machine (SVM). The results of validation showed that the SVM model had the best classification performance. This model was applied to 130 bunches of Red Globe grapes with a grading accuracy of 94.6%. These results will contribute to predicting grape quality and yield in the future.

Key words: Red Globe grape; compactness; normalized GB color difference method; morphological reconstruction

CLC Number: