Study of temporal variations of color and temperature on bruised apples from thermal and visible images

High percentage of orchard products, such as apples, is wasted due to mechanical damages that cause fruit quality loss. Damages due to static or dynamic pressure or impact are among very common mechanical damages that begin to bruise fruits. Post-harvest bruise damage is a major cause for the loss in fruit quality. Bruising means damaging fruit tissue and consequently physical changes resulting in fruit color and chemical changes resulting in fruit tastes (Xing and Baerdemaeker, 2005). Most research projects conducted on apple bruising have focused on the use of image processing techniques for detecting apple surface defects from images. In addition to images taken in visible spectral range, thermal images have been also used for this purpose. Having reviewed the literature and research gaps in this area, we set two hypotheses for this research project: first, the color characteristics of bruised tissue would change over time and these changes would be detectable on the images taken from the affected fruits. Second, there would be a significant difference between the surface temperature of bruised and sound tissues. The distribution of temperature on an impact-caused bruised tissue would change over time in a different manner compared with that for a sound tissue. The color and temperature variation is particularly related to the intensity of impact caused bruising and where the impact is applied on apples.
Therefore, the first objective of this paper was to study the color changes on the tissues bruised from the impacts with three energy levels applied on three locations on apple surface with different curvatures: top, middle and bottom. The second objective was to investigate the temperature variation on the surface of the bruised apples and to examine the capability of visible and thermal imaging in detecting bruised tissues at different times after bruising occurred.
For these purposes, the experiments were conducted on sixty apples of Golden Delicious variety. From sixty samples, five apples were used for determining apples ripeness index and five apples were used for determining emissivity factor which was used later in calibrating fruit surface temperatures on thermal images. Bruising was simulated by an impact pendulum. Bruising was conducted at three impact energy levels of 200, 700, 1200 mJ and applied at three locations on apple surfaces: top, middle and bottom. The samples affected by bruising-simulated impacting device were kept in a refrigerator at 5°C and were individually imaged in a regular basis until 624 hours after impact application. At the time of imaging, both visible and thermal images were taken from each sample. Samples visible images were taken in an imaging box with uniform controlled lighting. Thermal images were taken while samples were placed in a box that was thermally insulated from surrounding temperature.
A newly defined color factor, named excessive yellow index (EYI) was extracted from visible color images. The EYI index formula is EYI = 1.5r1.5g where r, g and b are red, green and blue color values, respectively. Factorial experiment was conducted for the assessment of EYI. This experimental design looked at the effects of three factors of time, impact energy and impact landing location on EYI.
Result and The results showed that time passed after impact and the location of impact application had significant effect on EYI at 95% confidence interval. The apples EYI index decreased until 15 days after impact application and started increasing thereafter.
Surface temperatures were extracted from the thermal images of samples. The results of processing thermal images showed that the bruised tissue was cooler than the sound tissue until 48 hours after impact application. Both tissues had the same temperature from 56 to 96 hours and then the bruised tissue started becoming warmer by 0.5-1°C after 96 hours.
The color variation of bruised region was not detectable from visible images within the first 48 hours after impact application, while these regions were cooler than undamaged region and detectable from thermal images. The bruised regions started to turn dark brown at 48 hours after impact application. However, there was no temperature difference between bruised and sound regions on fruit surface for the period of 56-96 hours. The bruised brown regions paled after 360 hours. As a result, this reduced the capability of visible images for discriminating bruised apples from sound ones.
The results of this research show that both the visible spectrum and thermal imaging systems can record the changes in color and temperature at different times after the bruising in apples. Therefore, these methods can be used as an efficient methods for grading apples.