An Analysis of Segmentation Techniques on Diseased Leaf Images

Agriculture plays a vital role in the Indian Economy. In India over 70 percent of the rural household depend on agriculture. Insect pests, bacteria, virus, fungi caused plant diseases and weeds inflict enormous losses to the potential agriculture production. Anecdotal evidences also indicate rise in the losses, despite increasing use of chemical pesticides. At the same time, there is a rising public concern about the potential adverse effects of chemical pesticides on the human health, environment and biodiversity. The greenhouse staffs periodically observe and search the defect leaves manually, it is very time consuming. The advanced image processing techniques are used to detect the leaf diseases automatically. The various segmentation techniques can be applied to detect plant leaf disease caused by bacteria, virus and fungi. The segmentation techniques are used to subdivide an image into components. It distinguishes objects of interest from background. The comparison of various segmentation techniques such as adaptive threshold, multilevel threshold, ostu threshold, k –means color based segmentation and watershed segmentation are done. The values of PSNR and MSE are used to compare and analyze the image quality after segmentation.