A real-time apple grading system using multicolor space

Area of Science:

• Computer Vision
• Agricultural Technology
• Image Processing

Background:

• Accurate apple classification is crucial for quality control in the fruit industry.
• Traditional methods often lack precision in assessing both color and size simultaneously.
• Multicolor space analysis offers enhanced differentiation of fruit characteristics.

Purpose of the Study:

• To design and evaluate a real-time machine vision system for classifying apples based on color and size.
• To investigate the effectiveness of different color spaces for apple image analysis.
• To achieve high accuracy in classifying various apple cultivars.

Main Methods:

• Development of a real-time machine vision system.
• Utilized multicolor space analysis, including CIE L*a*b* and CIE XYZ color spaces.
• Classified apples into four categories based on color and size parameters.
• Tested the system on three apple varieties: Golden, Starking, and Jonagold.

Main Results:

• Achieved 97% identification success for red apple fields using the 'a' parameter of CIE L*a*b* color space.
• Obtained 94% identification success for yellow apple fields using the 'y' parameter of CIE XYZ color space.
• Attained an overall success rate of 99% for 595 analyzed apples.

Conclusions:

• The developed machine vision system effectively classifies apples by color and size.
• Multicolor space analysis, particularly CIE L*a*b* and CIE XYZ, significantly enhances classification accuracy.
• The system demonstrates high potential for practical application in automated fruit sorting and quality assessment.