This study analyses the effect of browning through image analysis based on colour and textural features in fresh-cut apple slices. A computer vision system (CVS) was developed for image acquisition, which consisted of a digital camera and a florescent lamp source for illumination with a contrasting background. The CVS was calibrated using standard colour values and a model was developed by artificial neural network technique. Three varieties of apples such as Honey crisp, Granny Smith, and Golden Delicious were used for the analysis. The apples were freshly cut and subjected to image acquisition. Normalized colour features (L*, browning index, hue, and colour change) and textural features (entropy, contrast, and homogeneity) were analysed from the acquired images. The varieties Honey Crisp and Granny Smith did undergo browning within 120 min, whereas Golden delicious did not brown significantly. The study concluded that colour and textural features were important decision features for detecting browning in apples through image analysis.

No abstract

No abstract

Producers often aggregate bales into stacks before transporting these bales to an outlet for consumption or delivery to industrial applications. Efficiency improvement in this infield bale logistics will be beneficial. To address this an R simulation program involving five methods for field stack location, namely field middle, middle data range, centroid, geometric median, and medoid, as well as origin a direct aggregation method to outlet, were developed. These methods were evaluated against field areas, ranging from 0.5-520 ha, for infield bale logistics (aggregation, transport, and total) using Euclidean distances. The simulation used several input field variables, laid out bales based on yield variation, determined optimized bale stack locations of methods, and evaluated distances of aggregation to the stack, transport from the stack to the outlet, and total logistics. The origin method used 1-bale handling tractor for direct aggregation to the outlet, while others formed the bale stacks and transported bales to the outlet using 6-bales/trip equipment. Results indicated for aggregation that geometric median was the best, followed by field middle or centroid, middle data range, medoid, and finally origin. Methods aggregation were about 76% and transport about 24% of the total (for >2 ha); and total distance were about 65% of the origin. ANOVA, excluding origin, indicated that all methods were not significantly different (p < .05) for the areas studied. The 'field middle' was recommended as an easy and practical method for locating field stacks. Fitted power models described well (R 2 > 0.99) all the logistics distances.

No abstract