Study on Quality Parameters of Yellow Matoa Fruit (Pometia pinnata) Using Digital Image Processing

1st International Conference on Mechanical Electronic and Biosystem Engineering

Furay, A.^1 ; Ahmad, U.^2 ; Widodo, S.^2

Postharvest Technology Master Program, Graduate School, Bogor Agricultural University, 16680, Indonesia^1

Department of Mechanical Engineering and Biosisitem, Faculty of Agricultural Engineering and Technology, Bogor Agricultural University, 16680, Indonesia^2

关键词: Horticultural products;    Image processing algorithm;    Manual measurements;    Overall accuracies;    Quality evaluation;    Quality evaluation method;    Quality parameters;    Three categories;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/557/1/012025/pdf
DOI  :  10.1088/1757-899X/557/1/012025

来源: IOP

【 摘 要 】

At this time matoa fruits are sold without grading them first so high and low quality fruits are mixed. In addition to not having SNI, there is also no standard method for grading according to standardization of fresh horticultural products. Related to this issue, digital image processing can be used as an alternative method for grading. The purpose of this study were to study quality parameters and to develop quality evaluation method for yellow type matoa using digital image processing. Manual measurement using weight as quality parameter of 203 yellow matoa produced three categories of quality classes namely A, B and C. Image processing algorithm was then developed to replace the manual measurement by estimating the weight using projected area of fruits image and to measure skin color of the fruits. The overall accuracy of quality evaluation based on weight using developed algorithm was 73.89%. For quality evaluation based on visual parameter or skin color represented by red and blue color ratio (R/B), and hue and value (H/V) color ratio, it was found possible to classify the fruits into three quality classes namely class 1 (brown to yellow-brown), class 2 (yellow), and class 3 (yellow-green to green) with overall accuracy of 74.38%. Finally, using combination of those two parameters (i.e. area and skin color) nine new quality classes, namely A1, A2, A3, B1, B2, B3, C1, C2, and C3 classes were obtained with overall accuracy of 52.71%.