Engineering characterization of kinnow and tomato for grading using image processing
Loading...
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Punjab Agricultural University, Ludhiana
Abstract
The present study was undertaken for characterization of kinnow and tomato based on
engineering parameters using digital image processing. The images of kinnow and tomato of
different grades were acquired using a digital camera. An algorithm using MATLAB was
developed to quantify and process these digital images. The geometric parameters such as
axial dimensions, mass, volume, density, sphericity, aspect ratio and ellipsoid ratio were
recorded. The size of kinnow varied between 62.50-87.41 mm while tomato had size varying
from 30.00-77.62 mm. Results of image analysis showed variation of major axis length from
1597.21 pixels to 1987.1 pixels for kinnow. The bulk density was found to be maximum (0.40
g/cc) for kinnow having size 72-74mm whereas it was highest (0.26 g/cc) for tomato of size
57-66mm. The sphericity varied in the range of 0.92-0.94 for kinnow while tomatoes of
variety Punjab Chhuhara were found to be oblong. The results obtained by digital image
analysis were compared and verified with the experimental observation recorded by vernier
caliper. The volume computed from the images of longitudinal view and lateral view of
kinnow and tomato of all the grades showed good correlation with the true volume measured
by water displacement method. The Bland-Altman approach was used to plot the agreement
between the values measured by both the methods and differences between two methods were
normally distributed and were estimated to lie between M - 1.96SD and M + 1.96SD, known
as 95% limits of agreement. The paired samples t-test results showed that parameters
determined with image processing method was not significantly (P>0.05) different from the
same parameters measured with vernier caliper. A linear relationship between mass of the
kinnow and tomato and the projected area, volume and axial dimensions was also developed
for prediction of mass using image processing.
Description
Keywords
null