Development of an artificial cloud lighting condition system using machine vision for strawberry powdery mildew disease detection

•Factors affecting machine vision parameters were optimised.•The artificial cloud lighting condition was outperformed than natural lighting condition.•Luminance colour plane was not able to increase detection accuracy.•Image acquisition speeds (1 and 1.5 km h−1) and working depth (300 mm) showed promising results. Strawberry plants have been facing a significant proportion of diseases during cultivation, scattered throughout the field, emphasizing the need for proper diseases management. Powdery mildew is one of the major fungal strawberry disease which is typically responsible for approximately 30–70% loss of yields. The aim of this study was to develop a machine vision based artificial cloud lighting condition system for detecting strawberry powdery mildew leaf disease. The artificial cloud lighting condition system was developed consisting of custom software, two µEye colour cameras, a black cloth cover, real time kinematics-global positioning system and a ruggedized laptop computer and mounted on a mobile platform. The custom software was developed in C# programming language. The colour co-occurrence matrix based texture analysis was used to extract image features and discriminant analysis (quadratic) for classification. The study proposed mobile platform of artificial cloud lighting condition for image acquisition is beneficial. It showed higher detection accuracies of 95.26%, 95.45% and 95.37% for recall, precision and F-measure, respectively compared to 81.54%, 72% and 75.95% of recall, precision and F-measure, respectively with acquired images at natural cloud lighting condition. The feature selection results suggested the PM_GHSI feature model was best fit for this study. This study also revealed that the image acquisition speed (1.5 km h−1) and working depth (300 mm) are suitable for strawberry powdery mildew disease detection in real-time field condition.