Elaeis Guineensis leaf image segmentation: A comparative study and analysis

The main intention of the research is to identify and segment the diseased-pattern section that comprised of colour and texture, apart from the background which is also known as region of interest (ROI). Hence, a comparative study related to segmentation of Elaeis Guineensis (oil palm) leaf images extracted from the oil palm will be evaluated and validated. The database consists of images of leaf suffering from nutrition deficiency namely nitrogen, potassium and magnesium. Three different techniques of segmentation are investigated specifically the Otsu global threshold, local threshold and global threshold with tophat. Next, the segmentation algorithms have been developed to be capable to perform segmentation process using the leaf images that were exposed to varying illumination. Initial findings showed that Otsu global threshold is the best segmentation based on the tested images.