Smartphone near infrared monitoring of plant stress

https://doi.org/10.1016/j.compag.2018.08.046

Abstract

Graphica Image with AbstractThe most widely used method for monitoring plant stress is the use of near infrared (NIR) spectrophotometry to calculate normalized difference vegetation index (NDVI), as defined by [NIR reflectance???red reflectance]/[NIR reflectance?+?red reflectance].

NDVI measures the chlorophyll absorption in the red spectrum relative to the scattering by cellular structure in NIR, and has been used to monitor vegetation health and subsequently its stress from aerial or satellite images. Rather than using an NIR spectrophotometer or an NIR camera that is rather expensive, we attempted to use a commercial smartphone, utilizing its (potentially unintended) ability in recognizing near infrared (NIR) color.

Some of the most recent versions of smartphones have eliminated the NIR block filters on their cameras, and able to recognize NIR in their red pixels of CMOS array.

Through attaching an inexpensive high pass filter at 800?nm to a smartphone camera, we were able to collect the NIR reflectance (with a high pass optical filter) and the red reflectance (without a filter), enabling NDVI assessments.

This method was verified by measuring the NDVI values from a series of chlorophyll solutions, and showed a strong linear correlation with R2?=?0.948, corroborating the smartphone’s ability in evaluating NDVI.

Using the leaves from three different plant species, the NDVI values were evaluated using the smartphone and compared with the plants’ chlorophyll contents using acetone extraction and subsequent spectrophotometry.

A good linear relationship was found with R2?=?0.88–0.92.

We further evaluated the NDVI values against the plants’ water contents (measured by oven-drying), showing the non-linear relationship with the NDVI saturation above 50% water content.

The assay time was almost instantaneous, requiring only a smartphone and a high pass filter, thus allowing inexpensive, easy-to-use, rapid, and early prediction of plant stress that can be used for field and household applications.

Source: https://www.sciencedirect.com/science/article/pii/S0168169918305519