Diseased Leaf Segmentation from Complex Background Using Indices Based Histogram

Early detection of plant diseases is a key to avoid losses in agriculture products' quality and quantity. The study of plant diseases entails the examination of visually discernible patterns on the plant. Plant health monitoring and disease identification are important for long-term agriculture...

Full description

Saved in:
Bibliographic Details
Published in2021 6th International Conference on Communication and Electronics Systems (ICCES) pp. 1502 - 1507
Main Authors BR, Pushpa, AV, Shree Hari, Ashok, Adarsh
Format Conference Proceeding
LanguageEnglish
Published IEEE 08.07.2021
Subjects
Agriculture
Classification
CNN
Histograms
Image edge detection
Image segmentation
Indices-based Histogram
Leaf Segmentation
Mean Shift Segmentation (MS)
Plants (biology)
Productivity
Shape
Slice segmentation
Online AccessGet full text

Cover

More Information
Summary:Early detection of plant diseases is a key to avoid losses in agriculture products' quality and quantity. The study of plant diseases entails the examination of visually discernible patterns on the plant. Plant health monitoring and disease identification are important for long-term agriculture. Manually monitoring plant diseases is extremely difficult that requires a great deal of effort, knowledge of plant diseases, and an inordinate amount of time consumed. As a result, image processing methods are employed to identify plant diseases. In the disease detection process, various phases of digital image processing are incorporated. The proposed work adapted a different kind of segmentation for getting the plant disease attributes based on the leaves images. The dataset used is collected from Plant Village, Kaggle, and Mendeley datasets with different plant leave images that include different shape, margin, and texture features for identifying the disease attacked to it. The collected dataset is sliced with train and test data classifiers and processed. Segmentation performance by an indices-based histogram approach resulted in 92.06% accuracy.
DOI:10.1109/ICCES51350.2021.9489112