Live-wire segmentation in agriculture

It is an accepted fact that the challenge in applying image analysis to agriculture processes primarily lies in designing robust image analysis algorithms capable of coping with the combination of an unstructured environment and the inherent variability of biological objects. An example of such a challenging agriculture process could be the discrimination between crops and weed plants from images for selective application of herbicides. This paper describes the initial experiments in segmentation the shape variations of plant using image analysis which will help in solving the aforementioned challenge and problem. The shape variability of plant from a collection of images taken from real field is investigated using live-wire algorithm. The results show the less number of seed point and time needed for segmentation compared to other algorithms, how accurate and efficient was the algorithm in extraction the boundary of the plant even if it was occluded or cluttered.