I want to exclude from the stitched DEM certain small features of my subject that protrude higher than the rest of the surrounding subject matter. For example remove a small narrow blue rod sticking up in the middle of a much larger predominantly red scene. I could do this by mapping pixels between the stitched orthophoto and the dem then generating a new dem based on colors of the features I want to mask out. However the blue rod is a finer feature than the point cloud density so the point cloud averaged the elevations of other pixels I want to keep that are close to the pixels of the blue rod targeted for removal. So this is ineffective.
The other solution is mask the pixels of the blue rod from each of the overlapping original JPG images prior to stitching. My question is how to account for the removed pixels in the JPG so that Agisoft will not be confused and ignore them and stitch with only the remaining pixels, and obviously produce an accurate DEM?
Has anyone tried to do this, mask out details this fine? Thanks!