Claims:
1. Using a combination of light sources and detectors which illuminate fruit samples with ultraviolet, visible and near infrared light such that this light penetrates the skin of the fruit(s)
2. Measuring absorbance of the fruit through reflectance at particular wavelengths to reveal parameters of total soluble solids, dry matter content, moisture, titratable acidity, chlorophyll content
3. Finding index of freshness, ripeness and sweetness of the fruits based results of parameters obtained by the method described
4. Inclusion of Geo-Location analytics for data logging, pricing information, producer connects, quality metrics over time.
5. Using a wavelength based semiconductor CMOS to detect Near Infrared frequencies for the purposes of detection of light in the 600-1100 nm wavelength range with a resolution of 20-25 nm Full width half maximum

Description:
The invention relates to a method of measuring total soluble solids, titratable acidity, moisture content, chlorophyll content, dry matter content and appearance parameters as an index of freshness, ripeness and sweetness of fruits non-invasively by introducing ultraviolet, visible and near-infrared radiation which penetrates the skin of the fruit.
The spectrometer used is AS7265x and Hamamatsu C14834MA-01 with wavelength ranges 400-1100 nm with a typical spectral resolution of 20 nm and maximum 25 nm (Full Width Half Maximum at 840 nm)
Research was conducted on multiple samples of fruits on parameters like brix rating using a refractometer, moisture content using oven method and titratable acidity by using wet chemistry methods. The sample set selected consisted of fruits using various colour types
and different skin thickness. The sample set was procured from diverse set of locations for creating a robust model. A few of the fruits selected are grapes, apples, mangoes, kiwi and papaya. The methodology can also be used for fruits with higher skin thickness like
watermelons, jack fruit, pineapple by using different points of scanning on the fruit and brighter light sources. Light released form the sensor’s light emitting diodes illuminates the sample fruit and thereby excites the molecules in the fruit. The initial light intensity is measured by placing a
reference material that is 99% reflective in nature (such as a mirror or Spectralon). Then the fruit of interest is kept near the optical window of the sensor where the light intensity thatis reflected by the fruit is measured. It is known that near infrared radiation can penetrate
fruits and reveal underlying molecular information in the form of Near infrared overtones which occur for certain molecules.
The most common molecules which show near infrared absorbtion are C-H, C-H2, C-H3, O-H, H2O, N-H2, R-N-H2, Ar O-H, Ar C-H, R-O-H, C-O-N-H2, R-N-H-R. These molecules show their fingerprint by being absorbed at different wavelengths of the electromagnetic spectrum
that is between 400 – 1100 nm. It is known that near infrared radiation also indicates physical properties such as temperature, humidity, hardness, particle size and light scattering of the samples scanned. These physical properties also affect the spectrum, so a mathematical correction curve and a chassis with temperature control was developed to overcome these drawbacks so the device can be operational at different locations with diverse environmental conditions.
The ratio of this reflected intensity by the fruit and the reference scan taken earlier is known as reflectance. The negative of mathematical logarithm (base 10) is taken for each one of these reflectance values at each wavelength which computes to give the absorbance of that
fruit at that wavelength.
Mathematically:
Reflected Intensity of fruit / Reference reading of intensity = Reflectance
Absorbance = - log10 (Reflectance)
Light sources used are 405nm ultraviolet LED for shelf life, a Xenon lamp for visible range and a tungsten
halogen lamp for near infrared for brix, moisture and acidity. Light absorbance follows principle of Beer Lambert law which states that light absorbed at a wavelength of light is proportional to the concentration of molecules in the sample of interest, in this case fruits.
During illumination through which absorbance is computed, molecules get excited because the bond between them absorbs some amount of energy from regions of the wavelength thereby causing a change in the light intensity.
The absorbance is then measured and plotted for the data set created for individual fruits with X axis as wavelength and Y axis as the Absorbance values.
To extract qualitative and quantitative information from the device described, a software and app was developed using the principles and mathematics involved with calculating absorbance, reflectance to collect and store data of different fruits and develop a database of fruit absorbances. Pre-processing techniques like smoothening, Savitsky Golay 2nd order derivative, baseline correction and standard normal variate are used to remove noise andamplify features in the data sets. Algorithms based on Partial Least Squares Regression, Principle Component Regression and Neural Networks were implemented on the datasets to extract information from the near infrared spectrum and reveal underlying parameters of interest.
During this process, 150 intact grapes were measured by the method indicated and then were subjected to brix tests using a refractometer (range: 0-64% brix) which is an accepted and approved method of analysis of total soluble solids in fruits by Food Safety and
Standards Authority of India (FSSAI) and Association of Official Analytical Chemists (AOAC 932.12). The 150 grapes were crushed, and the juice extracted was measured with the help of refractometers as the reference method. A machine learning model was developed using
partial least squares regression for measuring total soluble solids in grapes non-destructively using brix refractometers as the reference and validation method.
A correlation coefficient of R2 = 0.92 was obtained with an SEP (Standard error of prediction) of
0.231. This indicates that the model developed is usable and practically accurate and shows
accuracy of up to plus-minus 0.5% of the actual brix value as determined by the refractometer.

The product includes a Geo-positioning system which records data collected at ground levels and uses it for forming various analytics which are provided to end user like quality metrics over a period of time, detailed data logs, pricing information and farmer or producer connects.