Biospeckle image processing algorithms for non-destructive differentiation between maturity and ripe stages of Indian climacteric fruits and evaluation of their ripening period

The proposed work deals with biospeckle image processing for non-destructive differentiation between maturity and ripe stages along with a ripening period evaluation of three Indian climacteric fruits (namely mango, guava and banana). We use existing algorithms such as histograms, RGB and HSB colour space, autocovariance, inertia moment (IM), absolute value difference, granulometric size distribution, gray level co-occurrence matrix and three new proposed algorithms, namely parameterized geometric mean of generalized difference (GD), image sequencing mean of parameterized GD, and squared temporal difference (STD). Histogram plots and spectral maps have been used for qualitative analysis whereas RGB and HSB values, speckle grain size plots, mean activity plots, GSD plots and textural features have been used for quantitative analysis. The experimental results reveal that IM is the best among the existing algorithms for differentiation between the maturity and ripe stages of the fruits quantitatively with the highest biospeckle activity (BA) difference, and STD is the best among the proposed algorithms for finding the qualitative as well as the quantitative difference and evaluation of ripening periods of the fruits. The average ripening periods of the three fruits were found to be approximately 86.86  ±  3.00 h, 73.93  ±  4.16 h, and 58.47  ±  2.23 h, respectively. It is further concluded that both the BA difference and one of the textural feature variations between the maturity and ripe stage are highest for the banana using IM (1562.32) and contrast variation (73.72), respectively. (paper)