Plant leaf segmentation through connected pixel approach

Agricultural plays a significant role in human survival and it has become much more essential due to population increase and food demand, and hence the crop yield has to be produced according to the demand. However, one of the reason that quality and quantity of the crop gets compromised is the disease and in past various methodology has been proposed, however they lack on the various model metrics or the segmentation is achieved for the particular leaf,. In this paper, we have proposed a methodology named as SCPA (Segmentation through Connected Pixel Approach). The main objective of this paper is to achieve high accuracy segmentation. SCPA is the two step approach first we find the ROI(Region of Interest) of the particular leaf and in the second approach we find the instance based ROI i.e. for the whole plant, here both the step are performed simultaneously through incorporating one another. Moreover, SCPA is optimized iterative-based method and it is achieved through the approach of connected pixel approach. Connected pixels are the one where the edge of one pixel is connected to the other. When performed on the LSC dataset we achieve the accuracy of 95.10%. This methodology is compared with the various state of art model and existing system by considering the model metric such as SBD, the results shows that SCPA model performs better than the other exiting method also the pictorial comparison of segmented leaf are shown and it shows our model identify it well when compared to others.