Computer Vision Classification of Barley Flour Based on Spatial Pyramid Partition Ensemble

Area of Science:

• Agricultural Science
• Food Science
• Computer Science

Background:

• Barley flour quality is crucial for the food industry, with malting varieties facing limitations due to color and husk fragments.
• Naked barley varieties offer superior visual and nutritional quality for human consumption.
• Automated quality control in food processing benefits from precise sample classification methods.

Purpose of the Study:

• To develop and evaluate a Computer Vision System (CVS) combined with Spatial Pyramid Partition ensemble (SPPe) for classifying barley flour varieties.
• To differentiate between malting and naked barley flour types using image analysis and machine learning.
• To assess the accuracy of different machine learning algorithms for barley flour classification.

Main Methods:

• A Computer Vision System (CVS) integrated with the Spatial Pyramid Partition ensemble (SPPe) technique was employed.
• Fifty-five image features were extracted from twenty-two barley flour varieties.
• Machine learning algorithms including Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), J48 decision tree, and Random Forest (RF) were utilized for classification.

Main Results:

• The combined CVS and SPPe approach achieved high accuracy in classifying barley flour samples.
• Classification accuracy ranged from 75.00% (k-NN) to 100.00% (J48) across the tested machine learning algorithms.
• The J48 decision tree algorithm demonstrated superior performance with 100.00% accuracy.

Conclusions:

• The integration of CVS with SPPe offers a highly accurate and automated method for barley flour classification.
• This technique has the potential to significantly enhance quality control processes in the food industry.
• Distinguishing between malting and naked barley flour varieties can be effectively achieved using this image-based approach.