A review on crop and weed segmentation based on digital images

Abstract

Apparently weed is a major menace in crop production as it competes with crops for nutrients, moisture, space and light which resulting in poor growth and development of the crop and finally yield. Yield loss accounts for even more than 70 % when crops grown under unweeded condition with severe weed infestation. There are several weed control measures being practiced in crop production, they are physical, mechanical, biological and chemical methods. Weed Management plays vital role in agriculture and horticulture production and economic benefits derived by agricultural industry. Weed is controlled mainly by application of herbicides. Weeds are not uniformly distributed in the crop and uncropped fields and mostly they are found in patches. With the help of Color and growth parameters, the weeds and crops may not be distinguished in the fields for the reasons of imbalance in availability of nutrients, water and other environmental resources. Weed control need to be done at the early stage of the crop growth. The management of weeds with in the field is imperative. Weed management practices using chemical tools propose to apply herbicide in the dosage strictly necessary based on weed infestation and location or position. Currently research is carried out relating to identification of weed spices and the location of the weed occurrence with the aims to allow accurate weeding and apply herbicides based on the weed density. Machine vision system, remote sensing and aerial imaging techniques are used for control weeds. Sensor attached electromagnetic system, imaging spectra radiometer and spectrometer can also be used to identify weeds for effective weed control. Almost all the existing weed detection methods process the captured image by segmentation of vegetation against background (soil), detection of weed vegetation pixels. Further, classification of feature extraction of weeds is done by color, shape and texture. The various methods studied and concepts used for crop and weed discrimination by the various researchers are discussed in this paper. © 2013 Springer.