In Step 8 of the Target Detection Wizard, you have the opportunity to use different filtering options to clean up mis-detected pixels and false positives. Sometimes a single target object can be separated into multiple objects by only several mis-detected pixels, and sometimes false positives are detected as a blob with a single or a few pixels.

  1. Select the target to filter from the Select Target drop-down list.
  2. Choose from the following Filter Options:
    • Clumping: Groups the separated pixels into one object. The target pixels are clumped together by first performing a morphological dilate operation then by an erode operation using a kernel of the size specified in the clumping parameters.
    • Sieving: Removes small isolated objects. It looks at the neighboring four or eight pixels to determine if a target pixel is grouped with other target pixels. If the number of target pixels that are grouped is less than the Group Min Threshold value, ENVI removes those pixels from the output image.
  3. If you are using the Clumping filter, set the morphological Operator Size for Rows and Columns.
  4. If you are using the Sieving filter, set the minimum number of pixels contained in a class group in the Group Min Threshold field. ENVI removes any groups of pixels smaller than this value from the class.
  5. For Sieving, toggle the Number of Neighbors button to select 4 or 8 as the neighboring pixels to look at when determining the number of pixels in a class group. The four-neighbor region around a pixel consists of the two adjacent horizontal and two adjacent vertical neighbors. The eight-neighbor region around a pixel consists of all the immediately adjacent pixels.
  6. Click Next to filter the display. The Export Results panel appears.