Flowering estimation in apple orchards by image analysis

Tree-specific management practice related to individual tree physiological condition is necessary for higher quality and quantity in apple fruit production. Detection of apple flowering abundance based on analysis of HSL (hue, saturation, luminance) images was used to estimate the number of flower clusters (FC) of individual trees in a high density apple orchard. The image acquisition was performed with a still camera and an industrial color camera during the day and night. The FC estimation algorithm included HSL thresholding with parameter optimization. Three hypothetical, tree-specific management practices (sprayings) were assumed, using >25, >50 and >100 FC thresholds to carry out the practice. When an industrial camera was used for image acquisition during the daytime and hypothetical spraying was done by on/off criterion >100 FC per tree, 10 % incorrect executions were identified. Comparable FC counting performance was achieved by using a still camera or an industrial camera.