HI- I am trying to calculate second-order texture measures using the co-occurrence filter in ENVI on images that are not squares. ENVI automatically assigns a value of zero to the area outside my area of interest. Does anybody knows how to deal with the zero in the co-occurrence filter? I tried subsetting my image using a ROI, but it still seems like the values outside the subset are set to zero, and they induce some bias in the calculation of the texture near the edge of the image. Thank you, Veronique