Modeling of Orange Mass Based on Dimensions

There are instances in which it is desirable to determine relationships among fruit physical attributes. For example, fruits are often graded on the basis of size and projected area, but it may be more suitable and/or economical to develop a machine which grades by mass. Therefore, a relationship between mass and dimensions or projected areas and/ or volume of fruits is needed. Various grading systems, size fruits on the basis of specific parameters. Sizing parameter depends on fruit and machine characteristics.Models for predicting mass of orange from its dimensions and projected areas were identified. Models were divided into three classifications: 1- Single and multiple variable regression of orange dimensions (1 st classification). 2- Single and multiple variable regression of projected areas (2 nd classification). 3- Estimation of orange shape; ellipsoid or spheroid based on volume (3 rd classification). Ten Iranian varieties of oranges were selected for the study. 3 rd classification models had the highest performance followed by 2 nd and 1 st classifications respectively, with R 2 close to unity. The 2 nd classification models need electronic systems with cameras for projection whereas, 1 st classification models are used in the simple mechanical systems, except multiple variable ones, of and 3 rd classification models need more complex mechanical systems. Among the systems that sorted oranges based on one dimension (Model 2), system that applies intermediate diameter suited better with nonlinear relationship as: M = 0.07b 2 – 2.95 b + 39.15 with R 2 = 0.97.