the ability of multispectral images and chemometrics for ripeness prediction of kiwi fruit

Spectral imaging combine the spectroscopic attributes of chemical measurements with those of imaging for chemical analysis. it has many applications in chemistry, biology, medicine, food science, and agriculture. in spectral imaging techniques, a spectrum is measured per pixel and the sample can be scanned without preparation [1]. hyperspectral images (hsis) are often composed of a large number of pixels a large number of variables to get sufficient spectral resolution and selectivity. however, in the multispectral image spectroscopic features are recorded within specific wavelength ranges. the wavelengths can be separated by filters or detected via the use of instruments that are sensitive to particular wavelengths. the size of the multispectral images can be very small but may carry more information than an rgb image. kiwis are exceptionally high in vitamin c, a nutrient that helps protect our cells from oxidative damage and plays many other important roles in the body [3]. however, a visual inspection cannot detect its ripness easily, and squeezing is needed. on the other hand, the kiwis’ripeness can be changed over time until they over-ripeness in this study, we utilized a simple set-up including a laptop and a smartphone to monitor the kiwi’s change over time. this set-up can collect multispectral data by changing the color of the laptop screen (9 colors). in each color of the screen, one rgb image is recorded by a smartphone. each image is taken in a separate spectral band in the visible spectral region. by augmenting all of the unfolded rgb images, the multispectral image was generated. the multispectral images of kiwi fruit in 12 hours were used to predict time using partial least squares regression (plsr). according to the results, the root mean square error values (rmse) for the calibration (0.26) and cross-validation (1.8914) were the most optimal by selecting three factors. therefore, three factors were used as the optimal number of factors to evaluate the plsr model. the precision and accuracy of the model are adequate.