Circle Fitting Based Image Segmentation and Multi-Scale Block Local Binary Pattern Based Distinction of Ring Rot and...

Area of Science:

• Agricultural science
• Computer vision
• Plant pathology

Background:

• Ring rot (Botryosphaeria dothidea) and anthracnose (Colletotrichum gloeosporioides) are significant apple fruit diseases.
• Accurate and timely diagnosis is crucial for effective apple disease management and quality assurance.

Purpose of the Study:

• To investigate an automated method for distinguishing between apple ring rot and anthracnose using image processing.
• To develop and evaluate machine learning models for high-accuracy disease classification.

Main Methods:

• Image preprocessing included scaling, contrast stretching, and morphological operations.
• Lesion segmentation was performed using a circle fitting method, with the optimal method selected based on comparison with manual segmentation.
• Features were extracted using multi-scale block local binary pattern (LBP) operators across RGB, HSI, and L*a*b* color spaces, followed by classification using Support Vector Machine (SVM) and Random Forest models.

Main Results:

• The optimal SVM model achieved 100% accuracy on the training set and 95.12% on the testing set.
• The optimal Random Forest model achieved 100% accuracy on the training set and 90.24% on the testing set.
• The proposed image processing method demonstrated high accuracy in distinguishing between ring rot and anthracnose.

Conclusions:

• An automated image processing technique can effectively distinguish between apple ring rot and anthracnose.
• The developed method provides a valuable tool for apple disease management and quality control, enabling accurate diagnosis.