Below are the results of a set of ground tests conducted to compare 3 Canon Point and Shoot Cameras and 10 filters for single camera NDVI generation. Details on the effort are explained here: http://flightriot.com/vegetation-mapping-ndvi/
The Canon Cameras used in this test are as follows:
- Canon PowerShot A490 (CCD Sensor)
- Canon PowerShot A810 (CCD Sensor)
- Canon PowerShot SX230 (CMOS Sensor)
All filters were placed on the outside of the lens using tape.
All images were captured mid-day with White Balance set manually prior to each shot using a stack of white copy paper. The paper was included in each shot for reference. I made the mistake of not filling the frame with the “white card” so in some of the pictures the white is a bit off. I re-captured one such image taken with the 2007 filter so that I could compare NDVI outputs with each WB (full frame vs not full frame) and the results in my opinion do not show enough deviation to nullify the value of test results.
Camera color settings were set to “Off” except for the final set with the sx230 where I implemented “Custom Color” settings. The images with Custom Color settings are represented by the files that start with sx230_C_… Here are the Custom Color settings used:
SCALE —| | | | |—
R, G, & B X
I used these settings to get better NDVI value range but I don’t know that this is a good idea. Ultimately any camera selected will need to be calibrated to some ground truth value.
The following settings were used in the Fiji/ImageJ Photo Monitoring Plugin:
The picture below is an RGB reference image
Hover your mouse over the pictures below to display camera and filter. The label is camera first, underscore, and filter name. An example would be: A490_2007 = Canon A490 with Rosco 2007 filter.
Lee HT#363 (HT =polycarbonate version = not recommended)