Development of a machine vision-based colorimetric system for bell peppers (Capsicum annuum L.) using intelligent modelling techniques
The appearance color of bell peppers is significantly related to their quality which affects consumer’s acceptance to buy. Also, homogeneity of color are among the most essential export standards of bell peppers. Commercially standard colorimetric devices are very expensive and are often available in scientific research centers. The aim of the present study was to develop and calibrate a simple, cheap, and portable machine vision (MV)-based system to accurately measure the chromatic parameters of bell peppers. For this purpose, a MV system possessing a digital CCD camera, and an artificial lighting system was developed. To calibrate the color of the utilized camera, the standard color cards were used. An appropriate algorithm based on image processing techniques was developed to calculate the chromatic parameters of the crop in the CIELAB color space. The development system was calibrated and compared with a standard colorimetric device using various developed models including linear regression (LR), multivariate linear regression (MLR), and artificial neural networks (ANNs). All the models were employed in diagnose of the chromatic properties of bell peppers using the developed MV system. The results revealed that the LR model outperforms MLR and ANN models. The overall accuracy of the proposed MV system was 91.7% in comparison with the standard device. The results showed that the proposed system can be considered as a more reliable device compared to traditional commercial devices and could be a suitable alternative in the absence of a specialized color measurement device.