Surface color distribution analysis by computer vision compared to sensory testing: Vacuum fried fruits as a case...

Area of Science:

• Food Science
• Sensory Science
• Colorimetry

Background:

• Food product quality perception is heavily influenced by color.
• Food surfaces exhibit heterogeneity across multiple scales, with colors changing during processing.
• Analyzing diverse color distributions offers greater insight than average color measurements.

Purpose of the Study:

• To compare sensory testing, Hunterlab colorimetry, and two computer vision systems (CVS) for analyzing color heterogeneity.
• To evaluate the ability of each technique to differentiate samples and quantify color variations.
• To assess the flexibility and limitations of each color analysis method.

Main Methods:

• Sensory testing by a trained panel.
• Color measurement using a Hunterlab colorimeter.
• Analysis with a commercial computer vision system (IRIS-Alphasoft).
• Analysis with a custom-made computer vision system (Canon-CVS).
• Application to nine different vacuum-fried fruit samples.

Main Results:

• Sensory testing effectively described color heterogeneity but suffered from panelist subjectivity.
• Hunterlab colorimeter provided accurate measurements for homogeneous samples but lacked color distribution data.
• IRIS-Alphasoft offered quick color distribution analysis but operated within a closed system.
• The custom Canon-CVS demonstrated flexibility in assessing color heterogeneity and sample discrimination.

Conclusions:

• Objective color distribution analysis holds significant potential for overcoming traditional color analysis limitations.
• Detailed color distribution data is crucial for a comprehensive understanding of overall food product quality.
• Computer vision systems, particularly flexible custom-built ones, offer advanced capabilities for food color analysis.