Online detection and grading system for citrus full-surface color

Abstract

Objective: This study aimed to realize the fast and accurate sorting of citrus color. Methods: An online color detection and grading system of citrus based on machine vision was designed in this study. The system is composed of feeding unit, chain conveying mechanism, image acquisition system and fruit dividing unit. The industrial camera which was combined with the rolling mechanism was used to uniformly capture 50~60 frames of images for obtaining the complete surface information of citrus. The non-destructive testing software preprocessed each frame of images acquired in real time to obtain the two-dimensional coloring ratio, which is dynamically tracked and stored. The arithmetic mean value of two-dimensional coloring ratio was taken to reduce the influence of the repeated areas on the coloring rate calculation of citrus surface, and finally the calculated coloring rate was discriminated and graded. Results: The experimental results showed that when the transmission speed was 6 s-1, the maximum error of citrus coloring proportion calculated by the system was 5%, and the sorting accuracy rate was 90.54%. Conclusion: The system can meet the needs of fast and accurate sorting of citrus color.

Keywords

• machine vision
• color grading
• full surface information detection
• citrus