Premium
The Potential of Double K ‐Means Clustering for Banana Image Segmentation
Author(s) -
Hu Menghan,
Dong Qingli,
Liu Baolin,
Malakar Pradeep K.
Publication year - 2014
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.12054
Subject(s) - cluster analysis , segmentation , artificial intelligence , computer science , image segmentation , pattern recognition (psychology) , region growing , automation , image (mathematics) , k means clustering , computer vision , scale space segmentation , engineering , mechanical engineering
A two‐step k‐means clustering technique was used to segment banana images in this study. The first k‐means clustering image segmentation procedure could segment the contours of a banana finger and a banana hand from the background image. Adding the second k‐means clustering could quantify the damage lesions and senescent spots on the banana surface. The result of the validation test showed that the algorithm was suitable for the flaw extraction of banana finger, and the human visual evaluation of comparison among the original, manual separated and automatic segmented images of banana hand demonstrated the potential of this algorithm for banana hand segmentation. Furthermore, the influences of the other special factors, i.e., the specular reflection and the blurry phenomenon, on the segmentation of various banana images were also discussed in this study. Practical Applications The algorithm based on the double k‐means clustering has a potential for image segmentation of banana fingers and hand under nearly all postharvest situations. This automatic image segmentation algorithm will be beneficial for achieving a comprehensive automation of banana industry.