Identifying Central Features of Cotton Leaves in Digital Images with Difficult Backgrounds

Many digital image processing techniques applied to agricultural problems have as main target the leaves of certain species of plants. The most basic task in such a context is to segment the leaf of interest from the rest of the scene, which is relatively straightforward when the leaf is isolated and the image is captured under controlled conditions. However, real field conditions will often imply in little control over lighting and, more importantly, the background may include several elements that make the task considerably more challenging. This is especially true if there are other leaves with similar shape, texture and color in the scene, which is often the case. This paper presents a method to identify the main node of cotton leaves (where the petiole meets the veins) and the main primary vein, giving valuable information about the position and orientation of those leaves. The only constraint to which the method is subject is that the leaf of interest be located in a central position in the image.