Visual analysis of sea buckthorn fruit moisture content based on deep image processing technology

Area of Science:

• Agricultural Engineering
• Food Science
• Computer Vision

Background:

• Monitoring sea buckthorn (Hippophae rhamnoides) during drying is crucial for quality.
• Traditional methods for assessing moisture content can be time-consuming and destructive.

Purpose of the Study:

• To investigate the impact of moisture content changes on sea buckthorn appearance during drying.
• To develop a non-destructive method for real-time moisture content detection using computer vision.
• To predict sea buckthorn appearance using Long Short-Term Memory (LSTM) networks.

Main Methods:

• Sea buckthorn was dried at 65°C with Level 1 wind speed.
• Images were captured every 30 minutes throughout the drying process.
• Computer vision and image processing techniques were employed for data analysis.
• Color components were calibrated and modeled for moisture content detection.
• LSTM networks were utilized to predict fruit appearance based on moisture content.

Main Results:

• Real-time online detection of sea buckthorn moisture content was achieved through calibrated color components.
• The study demonstrated the effectiveness of image analysis in monitoring the drying process.
• LSTM modeling showed potential in predicting sea buckthorn appearance changes related to moisture content.

Conclusions:

• Color component analysis is a viable and effective method for non-destructively monitoring sea buckthorn moisture content during drying.
• Computer vision integrated with LSTM offers a promising approach for quality control in fruit processing.
• Standardized color modeling, even with limited data, can be applied to various agricultural products for moisture detection.