An Images Analysis Technique for Recognition of Brown Leaf Spot Disease in Cassava

Abstract

An automated computer vision system to monitor epidemic development of plant pathogens is important in site-specific crop protection. The objective of this study was to develop an image analysis technique for the detection of brown leaf spot (BLS) disease caused by Cercosporidium henningsii Allesch in cassava (Manihot esculenta Crantz). Color images of healthy and diseased cassava leaves were captured in fields with a resolution of 640×480 pixels and further cropped into blocks of 80×80 pixels. Several color indices including red, green, and blue chromatic coordinates (rgb), contrast indices r – g, g – b, (g – b)/r – g and 2g – r – b, and hue, saturation, and intensity (HSI) were used as color descriptors. An artificial neural network was used in classification between the healthy and BLS-infected plants. The algorithm correctly recognized 79.23% of diseased leaves and 89.92% of healthy plants. Influence of the neural network architecture on the identification accuracy was also observed.