Dynamic image processing method for vegetation indexes in precision agriculture

Background: Currently, aerial images, tissue analyses, soil sampling analyses, and soil plant analysis development (SPAD) readings are used to assess the crop nutrition status. The remote sensing techniques are very popular and used in many areas including PA in these days. But in the case of remote sensing which uses passive light system, the image data should be processed for vegetation index in PA. This paper presents the dynamic calibration image processing method of a CMOS image sensor, which uses three channels (green, red, blue) of crop images to determine crop reflectance for vegetation index.

Methods: The real-time crop image was acquired using CMOS image sensor. The crop images were acquired during a different light condition from sunny to cloudy. And dynamic calibrations were investigated for a true reflectance calculation. The background elimination algorithm and the crop canopy reflectance analysis algorithm were also used for this research.

Results: To eliminate the effect of ambient illumination variation on gray levels of crop image caused by either the clouds or the solar radiation angle, the dynamic calibration model calibrates the measured crop reflectance. The core of this investigation is the calibration methods between the CMOS image and the reflectance in crops. Some noticeable relationships between the CMOS image reflectance and light condition were found from this study.

Discussions: Development of a system would identify where vegetation index is low and would apply fertilizer only to these identified areas.

Conclusion: The developed dynamic calibration model can be used to compensate for the variation of ambient light caused either by the weather condition or the solar zenith angle effectively. The CMOS image sensor is capable of detecting crop reflectance reliably in real-time.