Climacteric Fruit Skin Defect Region Segmentation by Boundary Identification Using Morphologically Enhanced Watershed

Climacteric fruits are perishable due to their ripening nature even after being harvested. Reaped fruits are to be checked before marketing to decide their self-life. Manual checking of harvested fruits consumes more time and cost. Harsh handling and manual tactile methods lead to the spoilage of more than fifty percent of fruits which affect the livelihood of the farmers. Image processing methods are used to avoid spoilage by harsh handling and to check accurately to elevate a grower’s economy. The image of the fruits is captured through a digital device and pre-processed followed by the segmentation of the defective peel to segregate the affected fruits from the marketable gives a better result than manual checking. In this proposed work a combination filtering by MBG filter followed by segmentation using Morphologically Enhanced Watershed (MEWS) boundary identification was implemented. The MBG filter removes noise and enhances the image to select texture and edge features in different orientations that improves the segmentation process. The segmentation of defective fruit peel is checked with morphologically enhanced watershed, grab cut, k-means (3 clusters), and canny methods. Even with the improper lighting while capturing the images, in other segmentation methods, shadows are taken into consideration. On the contrary, the novel approach of using Morphologically Enhanced watershed method segregates the defective region of the harvested climacteric fruits more effectively than the other methods with the segmentation accuracy of 93.9% when the images are captured in an uneven light source.