NIR images in itself does not contain all the parameters required for further processing. We need to install an additional filter to alienate unnecessary parameters and allow the entry of only necessary spectrum. The camera sensor is designed in a way that it distributes all the spectrum received into three bands of Red, Green and Blue respectively. The NIR spectrum overlaps with the RGB spectrum and hence the image formed has a little bit of NIR spectrum in every band, which ultimately spoils the information required for further processing. The use of multi-spectral cameras enable us to take images using multiple filters and hence have an option to utilize any spectrum as required. But we don’t have that luxury while using a single camera. Hence, we need to choose the best filter option which after installation can alienate inappropriate spectrum from the necessary ones.
Today, I will show you about the Red filter and its efficiency in finding the plant health. The camera used here is a Canon SX280 HS. The IR filter has been removed to enable the camera shoot NIR images. As told before the NIR images in itself have a number of unnecessary spectrums that will hinder the perfect crop health analysis. Hence, to counter that, we are using a red filter over the camera. The red filter is used to block the unnecessary blue spectrum and allow use that band to store the NIR spectrum. The first band is Red, second is Green and third is NIR. The plant health is calculated by NDVI formula. NDVI stands for Normalized Vegetation Differential Index. The formula is (NIR – Red)/(NIR + Red).
This is the camera that I have used for this experiment. As you can see, there is a red filter on top of it. This red filter is used to block blue spectrum and use that band for NIR spectrum.
This image is a scene captured outside my office. It contains gravels, dry bamboos, plants, decorative cactus, building etc. The objective here is to find the health of the plants in the image. The plants can be easily distinguished in the image. I am going to apply the NDVI formula to every pixel in the image and then formulate a reflectance map. The software used in this process is QGIS 2.16. In the later posts I will show you step-by-step process on how to use QGIS to formulate this reflectance map and hence find NDVI of the plant.
The image shown here, is the reflectance map of the image shown above. The NDVI formula has been applied to the image to formulate this map. The reflectance basically shows the amount of IR reflected from the object, which in this case we can observe that the plant area has more reflectance in comparison to the dry bamboos or the gravels. NDVI calculation is solely based on the reflectance of NIR rays from plants. After finding the reflectance, we render the values from red to green, green being the healthiest. This process is done to give it an aesthetic look and present it to any layman.
The color render shows the health of the plants according to the calculated reflectance map. You can notice that plant area in the right corner is also shown as red color. The first conclusion that can be derived is that the plant is unhealthy. As it is the conclusion in most cases, it is not the reason here. If you refer to the first image, there is a red demarcation referring to a different variety of plant. This plant belongs to the family of succulent. These survive on the water retained in their stems and does not have the mesophyll layer which ultimately helps in the reflectance of NIR rays. Hence, when calculated, these plants show a low reflectance thereby rendering red color. But it doesn’t mean that they are unhealthy. It just means that, we cannot find the health of succulent plants via NIR reflectance. Other plants in the image as shown have a high reflectance due to thick mesophyll layer which is the reason for food generation.