Classification of bitter and sweet olives using image processing and artificial neural networks during curing process in brine and water environments

Figures & data

Figure 1. Image of the various stages in the two samples test environment (a. Water curing and b. Brine curing).

Figure 2. Schematic diagram of computer vision system for colour extraction and classification.

Figure 3. Changes in L*a*b* values using image analysis of olive during storage for nine day (a: L* Intensity, b: A* Intensity, c: b* Intensity).

Table 1. Mean comparison of storage day and colour spaces RGB and L*a*b* (brine curing).

	Mean Square
Day	R	G	B	L*	a*	b*
0	108.790^a	119.761^a	49.783^b	48.05737^a	−14.98258^d	36.37087^a
3	70.805^bc	79.199^b	37.284^d	30.68225^b	−8.36469^c	21.71777^b
5	69.084^bc	71.847^b	43.738^c	29.58737^b	−6.34233^c	16.50306^c
7	64.624^c	58.214^c	57.324^a	25.12172^c	2.62667^b	1.54879^d
9	71.247^b	57.382^c	60.311^a	25.71569^c	6.72621^a	0.51152^d

Means in each column followed by similar letter (s) are not significantly different at 5% probability level using Duncan Multiple Rang Test

Table 2. Mean comparison of storage day and colour spaces RGB and L*a*b*(water curing).

	Mean Square
Days	R	G	B	L*	a*	b*
0	108.790^a	119.761^a	49.783^c	48.05719^a	−14.9826^d	36.3708^a
3	77.826^cd	85.442^b	35.318^e	34.60091^b	−11.3723^c	27.7684^b
5	73.242^d	77.657^c	45.524^c	31.6948^c	−8.9447^b	24.1862^c
7	82.505^bc	86.234^b	55.570^b	35.5658^b	−7.8123^b	17.3342^d
9	87.706^b	86.444^b	60.761^a	36.207^b	−4.0907^a	15.1474^d

Means in each column followed by similar letter (s) are not significantly different at 5% probability level using Duncan Multiple Rang Test

Figure 4. Changes in RGB values using image analysis of olive during storage for nine day (a: Red Intensity, b: Green Intensity, c: Blue Intensity).

Table 3. Mean comparison of storage day and sensory parameters (Water and brine curing).

	MS (Water Curing)				MS (brine Curing)
Days	Colour	Taste and aroma	Texture	General acceptance	Colour	Taste and aroma	Texture	General acceptance
0	8.2^a	1.52^d	1.57^d	2.87^d	8.18^a	1.52^e	1.57^e	2.87^d
3	5.36^d	8.52^a	8.37^a	7.7^a	7.34^b	5.7^d	5.97^d	6.14^c
5	6.6^c	7.48^b	7.47^b	7.1^b	6.36^c	7.5^c	7.57^c	7.31^b
7	7.34^b	5.79^c	5.97^c	6.1^c	5.6^d	8.41^b	8.27^b	7.84^a
9	8.17^a	1.52^d	1.57^d	2.86^d	4.41^e	8.88^a	8.86^a	7.79^a

Means in each column followed by similar letter (s) are not significantly different at 5% probability level using Duncan Multiple Rang Test.

Table 4. Correlation between colour parameters and sensory test in water curing environment.

	L*	a*	b*	R	G	B	Colour	Taste and aroma	Texture	General acceptance
L*	1
a*	−0.58*	1
b*	0.67*	−0.85*	1
R	0.96*	−0.36^ns	0.57*	1
G	0.99*	0.65*	0.73*	0.94*	1
B	0.22^ns	0.45^ns	−0.51^ns	0. 33^ns	0.16^ns	1
colour	0.49^ns	0.16^ns	−0.13^ns	0.57*	0.44^ns	0.78*	1
Taste and aroma	−0.65*	−0.04^ns	−0.04^ns	−0.73*	−0.61*	0.66*	−0.86*	1
Texture	0.65*-	−0.02^ns	−0.05^ns	−0.72*	−0.61*	−0.65^ns	−0.85*	0.99*	1
General acceptance	−0.66*	−0.019^ns	−0.07^ns	−0.73*	−0.61*	−0.62*	−0.83*	0.99*	0.99*	1

* and ^ns Significant at 5% and not significant, respectively.

Table 5. Correlation between colour parameters and sensory test in brine curing environment.

	L*	a*	b*	R	G	B	Colour	Taste and aroma	Texture	General acceptance
L*	1
a*	−0.81*	1
b*	0.90*	−0.94*	1
R	0.96*	−0.61*	0.78*	1
G	0.99*	−0.85*	0.92*	0.93*	1
B	−0.17^ns	0.63*	−0.57*	0.02^ns	−0.23^ns	1
Colour	0.69*	−0.83*	−0.84*	0.57*	0.73*	−0.61*	1
Taste and aroma	−0. 94*	0.85*	−0.91*	−0.87*	−0.95*	−0.31^ns	−0.83*	1
Texture	−0.93*	0.84*	−0.90*	−0.86*	−0.95*	−0.29^ns	−0. 82*	0.99*	1
General acceptance	−0.95*	0.83*	−0.89*	−0.89*	−0.96*	0.25^ns	0.77*-	0.99*	0.99*	1

* and ^ns Significant at 5% and not significant, respectively.

Figure 5. Classification results from the artificial neural network analysis of bitter and sweet olive groups.