Improvement in Purity of Healthy Tomato Seeds Using an Image-Based One-Class Classification Method

Area of Science:

• Agricultural Science
• Computer Vision
• Seed Technology

Background:

• Seed health is crucial for germination and crop yield.
• Contaminated or diseased seeds reduce germination rates.
• Manual seed quality assessment is labor-intensive and subjective.

Purpose of the Study:

• To investigate the feasibility of a color machine vision technique for tomato seed quality assessment.
• To develop a system for segregating healthy tomato seeds from diseased, infected, foreign, or broken seeds.
• To correlate image analysis results with actual seed germination rates.

Main Methods:

• A custom-built machine vision system with a color camera and LED light source was used for image acquisition.
• Feature extraction was performed on acquired seed images.
• A one-class classification method was employed to identify healthy seeds.
• Seed germination tests were conducted using organic growing media.

Main Results:

• A significant difference in features was observed between healthy and infected tomato seeds.
• The machine vision system achieved over 97% accuracy in classifying tomato seeds.
• Infected seeds showed significantly lower germination rates (<10%) compared to healthy seeds.

Conclusions:

• Color machine vision with one-class classification is a feasible and accurate method for tomato seed quality assessment.
• The developed system can effectively discriminate between healthy and poor-quality tomato seeds in real time.
• This technology has the potential to improve seed quality control and enhance agricultural productivity.