I am trying to classify an aerial image of a dike into grass and weeds. I used an aerial RGB image of 10 cm resolution being a TIF file. This aerial image is cleaned and contains no gaps. Only outside the aerial image raster NoData is present.
I want to perform a object based supervised classification on this image. Before classifying the image, the image was segmented using the Segment Mean Shift tool with the values 19 for spectral detail, 10 for spatial detail and 10 for minimum segment size.
After the segmentation it turned out that the segmented raster has holes in it with NoData cells for the red band at some locations.. These locations are visualised with light blue in the attached figure. The remaining colours are from the segments. The second figure shows approximately the same part of the study area on the RGB image.
Do you have any tips or suggestions why these NoData gaps, shown in light blue, appear in my segmented dataset?
