Spectral Sensor Calibration for Maize Detection under Different Conditions

Abstract In order to satisfy flexible spectral measurement and chlorophyll in the field, a crop monitoring sensor with ZigBee protocol was developed. It could measure sunlight and crop reflected light at the wavebands of 550, 650, 766 and 850nm in real-time. The NDVI was calculated based on the digital signal. After the sensor calibration, the field experiment was conducted to detect the chlorophyll content of maize. The calibration experiments involved optical performance, wireless network performance and consistence Performance, in which the optical calibration was conducted to confirm the constant setting and the optical stability. The constant k was calculated by the white panel and the average value was 1.042. The average correlation coefficient between illuminometer and every channel was 0.9855. The wireless performance was evaluated by LQI (link quality indicator) and packet loss rate. There was no packet loss and the LQI was in [91.08, 240.12] when the distance less than 72m. The recommended distance was between 72m and 112m with packet loss rate [0, 0.02] and LQI [76.67, 83.42]. To test the consistence performance, 20 sensors were involved and the full magnification of each sensor was MAX. they were test under different magnifications including D1, D2, D3 and D4 (corresponding to the range 1/5MAX, 2/5MAX, 3/5MAX, and 4/5MAX). The results showed that D1 and D2 were recommended with stable measurement DN value range around 500 and 1000 respectively. The field experiment showed that the best resolution for the chlorophyll content detection was 0.5mg/L using the developed sensor. The MLR model was built by NDVI (850,550) and NDVI (766,550). The determination coefficient was R c 2 =0.72 and R v 2 =0.68 respectively.