Honey characterization using computer vision system and artificial neural networks
Sahameh Shafiee1, Saeid Minaei1, Nasrollah Moghaddam-Charkari2
1Department of Agricultural Machinery Engineering, Tarbiat Modares University, Tehran, Iran.
Food Chemistry
|April 29, 2014
Summary
A new computer vision system (CVS) non-destructively analyzes honey quality using color. This system accurately predicts chemical attributes like antioxidant activity (AA) and total phenolic content (TPC) from honey images.
Area of Science:
- Food Science
- Analytical Chemistry
- Computer Vision
Background:
- Traditional honey characterization methods can be destructive and time-consuming.
- Objective quality assessment is crucial for the honey industry.
Purpose of the Study:
- To develop a non-destructive computer vision system (CVS) for honey characterization.
- To correlate honey color with key chemical attributes: ash content (AC), antioxidant activity (AA), and total phenolic content (TPC).
Main Methods:
- Utilized artificial neural network (ANN) models to process honey images.
- Transformed RGB color values to CIE L*a*b* colorimetric measurements.
- Predicted AC, AA, and TPC using color features extracted from images.
Main Results:
- Achieved low generalization error (1.01±0.99) in converting RGB to CIE L*a*b* values.
- Demonstrated high prediction accuracy for AC (R²=0.99), AA (R²=0.98), and TPC (R²=0.87) based on color.
- Validated the effectiveness of the ANN models for colorimetric measurements and chemical attribute prediction.
Conclusions:
- The developed CVS provides an effective, non-destructive method for honey quality assessment.
- The system demonstrates the potential for industrial application in honey characterization.
- Color analysis via CVS offers a rapid and reliable alternative to traditional methods.
