Early Detection of Seed Infection using Motion History Image based Laser Biospeckle Technique

Abstract

Early detection of seed infection is important, as often, infected seeds appear asymptomatic and result in severe losses in crop production. Traditional methods used for detection of infected seeds are complex, costly, time-consuming, require mycological skills, and not sufficiently sensitive to early stage of pathogen development. To circumvent these drawbacks, this work presents a feasible alternative to the routine methods based on laser biospeckle analysis and motion history image (MHI) based processing strategy to detect the seed infection in early stage of its development. Visual results showed high activity regions in infected seeds as compared to healthy one, even at early stage of infection. Numerical index of biospeckle activity increased significantly (p<0.01) with time for infected seeds, whereas for control seeds it showed almost constant response. The obtained results demonstrate that biospeckle can be used as an efficient, automated, fast, and non-destructive tool for early detection of infection in seeds. © 2020 IEEE.