Potato Defect Detection using Fuzzy C-mean Clustering based Segmentation

Potato grading is one of the evolving areas of research due to infeasibility of various algorithms and designs developed to work in real time industrial environments. Various problems which arrive in real time environment include shadow in images, high complexity and inefficient speed of algorithms. The fuzzy c-mean clustering based algorithms can be efficiently exploited for potato grading due to their low complexity. Sothis paper introduces a potato defect detection method using fuzzy c-mean clustering based segmentation. The Euclidean Distance based on RGB color space is used to segment the potato image into defected and healthy areas. An anti-shadow wings model is introduced for the process of image acquisition. The potato is graded on the basis of defects like greening, cracks and rotten. The potatoes used to prepare the database were taken from local vegetable markets of India. The algorithm segments the potato images on the basis of RGB color values of the image pixels. The experimental results show that Fuzzy c-mean clustering is very effective to segment the potato images into defected and healthy areas. The novelty of the work is represented by novel wings design to remove the shadow in the images without using any pre-processing algorithm. Along with this modified Fuzzy c-mean algorithm is efficiently used first time to grade the potatoes successfully. Combination of both image acquisition design and algorithm proves to be a novel feature to grade potatoes successfully.