Image processing and analysis techniques for estimating weight of Chokanan mangoes

C.C. Teoh and A.R. Mohd. Syaifudin

Abstract

Size is one of the major parameters that the consumer identifies to be related to the quality of mango. According to Federal Agricultural Marketing Authority (FAMA) Malaysia, size of mango is determined by weight. This paper presents a study on measuring the weight of Chokanan mangoes using image processing and analysis techniques. The computing PCI software was used to process and analyse mango image obtained from image acquisition system. Number of pixels of mango region in the captured image was counted by the software after image processing and analysing processes. A relationship between mango pixels and mango weights was analysed using statistical method of regression. The statistical analyses results showed that the mango pixels counted has high relationship with the mango weights with correlation coefficient value of 0.9769. The linear model developed from the linear regression analysis was evaluated by a set of 50 mango samples in terms of mango weight estimation accuracy. The evaluation results showed that, the mean absolute percentage error of estimating a mango weight using the linear model was 3.76%. As a result, the image processing and analysis techniques are practical, feasible and effective in estimating weight of Chokanan mangoes.

Full Text (706 KB)