Li Minzan, Sun Hong, Zheng Lihua, Yang Wei, Li Xiuhua,
Zhao Ruijiao, Zhang Feng, Wang Wen
Key Laboratory of Modern Precision Agriculture Integration Research, MOE,
China Agricultural University

Nitrogen content of crop plant is an important indication for evaluating crop health and predicting yield. Using spectral analysis and ground-based remote sensing, some vegetation indexes (VI), such as NDVI, RVI, and DVI, can be calculated to evaluate growth status of crops. Thus we developed two kinds of hand-held spectrometer, an optical fiber spectrometer and an optical sensing spectrometer, and a vehicle-mounted detector were developed for the crop detection.
The wavelengths of 765nm and 530nm were selected to calculate VI in the optical fiber spectrometer. The optical fibers and the interference filters were applied for obtaining the light at the required wavelengths. The instrument consisted of a photocell as the transducer, a connection adapter for eliminating external interference and a LCD screen for the data display. It could communicate with a PC and send the data. The expert knowledge was stored in the memory of the instrument, so that the instrument could give the diagnostic result of the crop growth status automatically after measurement.
The optical sensing spectrometer was developed for the NDVI measurement under in-field natural light conditions. The optical sensing unit consisted of four transducers to acquire light intensity data: two for measuring the sunlight intensity and the other two for measuring the reflected light of crop canopy at bands of 610 and 1220 nm. The NDVI value was real-time calculated based on measured spectral reflectance. An electronic control unit was designed to control the operation and data recording, and perform real-time NDVI value calculations. The field results indicated that the detected data had a close correlation with chlorophyll contents measured in laboratory.
The vehicle-mounted detector was developed to work as a wireless sensor network with a measuring unit and a control unit. The measuring unit consisted of several optical sensor nodes. Here, each sensor node was an optical sensor and contained four optical channels, which allowed the sensor work at the wavebands of 550, 650, 766 and 850 nm. All the sensors were installed on an on-board mechanical structure so that the measuring unit could collect, process and transmit the spectroscopic data in mobile condition. The controller was a PDA embedded a ZigBee wireless communication module. The controller was the coordinator of the whole wireless network. It was used to receive, display and store all the data sent from different sensor nodes. The field test indicated a good stability of the wireless network and the detector.
In addition, two different image acquisition systems were developed for the plant detection. One was a 2-CCD near-ground remote sensing image acquisition system. The other one was a two-channel image collection system based on DSP. The 2-CCD near-ground remote sensing image acquisition system integrated a 2-CCD multi-spectral camera, two collecting boxes and a field computer. The visible light between 400-700 nm and the near infrared light between 700-1000 nm were separated by an optical prism structure. It was contained a single lens input and multiplexed output for image information. The visible image and NIR image of one region could be captured synchronously and four wavebands (R, G, B, and NIR) of the multispectral images could be acquired. The collecting boxes were connected with the camera and used to process and transfer digital image data. Acquired image data were stored and analyzed in the field computer. It provided a new instrument for the crop detection.
The two-channel image collection system integrated a binocular camera device and an embedded system based on DM642. Binocular camera device was composed of a visible light camera, a near-infrared camera and a bracket. The embedded system was mainly used to complete the calibration of the binocular camera and the image processing. It was shown that the system provided a feasible and low-cost method for the crop detection.
These developed devices mentioned above had been used in the crop detection such as cucumber, tomato, wheat, corn and so on. All of them showed great performance in green house and in the field. The application of spectral sensors in ground-based remote sensing conducted a great support for the precision agriculture.
Key Words: ground-based remote sensing, spectroscopy, sensor, precision agriculture