The multispectral orthomosaics of the same flight were obtained with Pix4D and Agisoft PhotoScan. On the pix4d index map NDVI the predominant value is about 0.7, while in PS it is about 0.3. Why do NDVI values calculated with both sofwares differ so much?
From what we know, Photoscan does not apply radiometric corrections. This could very explain discrepancies in values between the two software. Applying radiometric corrections for the images allows to normalize the images and compare different images on the same scale (e.g. it is normal for the pixel values of an image with twice the exposure time to be higher. This should be accounted for, which is what radiometric corrections do).
Thanks for the reply.
The processed images were taken with the Sequoia camera and the active illumination sensor and an NDVI map was obtained with the radiometric corrections and the other without radiometric corrections. In the map obtained without radiometric corrections the values of NDVI are more similar to those obtained with Agisoft PS, but in my opinion they are very low values, for example for leafy trees present in the map corresponds values of NDVI without radiometric correction of approximately 0.5 . Which of the two maps is closer to reality?
We cannot speak to the accuracy of Photoscan (especially if they do not apply radiometric corrections). We recommend applying the radiometric corrections in Pix4D: this will give you results closer to the actual values for the NDVI (and if you use a radiometric calibration target, to the actual reflectance values for the reflectance maps).