Research On The Acquisition Method Of Background Interference-free NDVI For Ground Crop Canopy Based On Narrowbanded Camera

This study developed an image processing algorithm for images captured by narrow banded camera developed by adding narrowband filters to a standard industrial RGB camera to determine NDVI of crop canopies at the ground level.The photos of green pepper(Capsicum annuum)were captured under sunlight conditions at two bandwidths the red bandwidth(660 nm)and the NIR bandwidth(770 nm).The acquisition of the images using the single camera directly perpendicular to the plant,with the plant remaining at the same position,satisfied the need for alignment of the pixels for the bands to make image processing more efficient.This approach uses a single camera and captures the images in the same exact position and distance,which leads to precise corresponding pixels for both images taken in 660 nm and 770 nm bands to ensure stability in the results obtained.It simplifies the technicalities that arise when two cameras are used simultaneously.In the study,the background removing model was developed that processed the acquired narrowband images,discriminating the unwanted noise that comes in the background,information that is unnecessary,obtaining the Region of Interest(ROI)and computed the NDVI of the plants.The algorithm obtains the constant Cw which is the ratio of the narrowband wavelength 770 nm to 660 nm of the whiteboard.This coefficient is independent on the varying heights and is used to compute NDVI.During the experiments,the images were obtained at different heights from 30cm to 70cm of the camera to the ground with different lighting conditions.The method of separating crop canopy from soil background is achieved efficiently based on the characteristics of strong contrast between the canopy and background in narrowband images.A set of experimental equipment was set up to test the canopy of the three-week pepper seedling to compute NDVI.The experimental results showed that the stable NDVI value could be obtained when the measurement height changed under the outdoor natural light.The proposed algorithm proved to obtain the value of NDVI without the background interference.It provides a new idea for practical application in field to get the NDVI of crop canopy.Comparable processing of data acquired was done to prove the stability of NDVI and shows the relevance and reliance of the model developed and its practicality.From the results,it can be seen that with a mean standard deviation of the relative error is 0.1979%,the value of Cw shows stability and minimal deviation.This proves that the Cw is a constant and the model is reliable and stable.The mean standard deviation of the relative error for NDVI result is 2.339%which makes this model very reliable.For calibration and standardization,it shows the reliability and stability of the proposed model.The ideal height for the proposed model ranges from height 30-60 cm.At height 70,the camera is very high and it’s field of view covers more whiteboard than the plant,makes it have the highest deviation.This is because the image has more interference than any image at the other lengths.