Image Rectification and Restoration

1
Chapter

• Image Rectification and Restoration
• Analysis and applications of remote sensing
  imagery
• Instructor Dr. Cheng-Chien Liu
• Department of Earth Sciences
• National Cheng Kung University
• Last updated 3 March 2015

2
Introduction

• Rectification ?? ? distortion ??
• Restoration ??? degradation
• Source
• Digital image acquisition type
• Platform
• TFOV

3
Geometric correction

• Geometric distortion ????
• Altitude, attitude, velocity of sensor platform
• Panoramic distortion, earth curvature,
  atmospheric refraction, relief displacement,
  nonlinearities in the sweep of a sensors IFOV

4
Geometric correction (cont.)

• Two-step procedure
• Systematic (predictable)
• e.g. eastward rotation of the earth ? skew
  distortion
• Deskewing ? offest each successive scan line
  slightly to the west ? parallelogram image
• Random (unpredictable)
• e.g. random distortions and residual unknown
  systematic distortions
• Ground control points (GCPs)
• Highway intersections, distinct shoreline
  features,
• Two coordinate transformation equations
• Distorted-image coordinate ? Geometrically
  correct coordinate

5
Two coordinate transformation equations

• Affine coordinate transform
• Six factors
• Transformation equation
• x a0 a1X a2Y
• y b0 b1X b2Y
• (x, y) image coordinate
• (X, Y) ground coordinate
• Six parameters ? six conditions ? 3 GCPs
• If GCPs gt 3 ? redundancy ? LS solutions

6
Resampling

• Resampling
• Fig 7.1 Resampling process
• Transform coordinate
• Adjust DN value ? perform after classification
• Methods
• Nearest neighbor
• Bilinear interpolation
• Bicubic convolution

7
Resampling (cont.)

• Nearest neighbor
• Fig 7.1 a ? a? (shaded pixel)
• Fig C.1 implement
• Rounding the computed coordinates to the nearest
  whole row and column number
• Advantage
• Computational simplicity
• Disadvantage
• Disjointed appearance feature offset spatially
  up to ½ pixel (Fig 7.2b)

8
Resampling (cont.)

• Bilinear interpolation
• Fig 7.1 a, b, b, b ? a? (shaded pixel)
• Takes a distance-weighted average of the DNs of
  the four nearest pixels
• Fig C.2a implement
• Eq. C.2
• Eq. C.3
• Advantage
• Smoother appearing (Fig 7.2c)
• Disadvantage
• Alter DN values
• Performed after image classification procedures

9
Resampling (cont.)

• Bicubic (cubic) interpolation
• Fig 7.1 a, b, b, b, c, ? a? (shaded pixel)
• Takes a distance-weighted average of the DNs of
  the four nearest pixels
• Fig C.2b implement
• Eq. C.5
• Eq. C.6
• Eq. C.7
• Advantage (Fig 7.2d)
• Smoother appearing
• Provide a slightly sharper image than the
  bilinear interpolation image
• Disadvantage
• Alter DN values
• Performed after image classification procedures

10
Radiometric correction

• Radiometric correction ????
• Varies with sensors
• Mosaics of images taken at different times ?
  require radiometric correction
• Influence factors
• Scene illumination
• Atmospheric correction
• Viewing geometry
• Instrument response characterstics

11
Radiometric correction (cont.)

• Sun elevation correction
• Fig 7.3 seasonal variation
• Normalize by calculating pixel brightness values
  assuming the sun was at the zenith on each date
  of sensing
• Multiply by cosq0
• Earth-Sun distance correction
• Decrease as the square of the Earth-Sun distance
• Divided by d2
• Combined influence

12
Radiometric correction (cont.)

• Atmospheric correction
• Atmospheric effects
• Attenuate (reduce) the illuminating energy
• Scatter and add path radiance
• Combination
• Haze compensation ? minimize Lp
• Band of zero Lp (e.q.) NIR for clear water
• Path length compensation
• Off-nadir pixel values are normalized to their
  nadir equivalents

13
Radiometric correction (cont.)

• Conversion of DNs to radiance values
• Measure over time using different sensors
• Different range of reflectance
• e.g. land ? water
• Fig 7.4 radiometric response function
• Linear
• Wavelength-dependent
• Characteristics are monitored using onboard
  calibration lamp
• DN GL B
• G channel gain (slope)
• B channel offset (intercept)
• Fig 7.5 inverse of radiometric response function
• Equation
• LMAX saturated radiance
• LMAX - LMIN dynamic range for the channel

14
Noise removal

• Noise
• Definition
• Sources
• Periodic drift, malfunction of a detector,
  electronic interference, intermittent hiccups in
  the data transmission and recording sequence
• Influence
• Degrade or mask the information content

15
Noise removal (cont.)

• Systematic noise
• Striping or banding
• e.g. Landsat MSS six detectors drift
• Destriping (Fig 7.6)
• Compile a set of histograms
• Compare their mean and median values ? identify
  the problematic detectors
• Gray-scale adjustment factors
• Line drop
• Line drop correction (Fig 7.7)
• Replace with values averaged from the above and
  below

16
Noise removal (cont.)

• Random noise
• Bit error ? spikey ? salt and pepper or snowy
  appearance
• Moving windows
• Fig 7.8 moving window
• Fig 7.9 an example of noise suppression
  algorithm
• Fig 7.10 application to a real imagey

17
Tutorial image georeferencing and registration

• Georeferenced Data and Image-Map
• Image to Image Registration
• Image to Map Registration
• HSV Merge of Different Resolution Georeferenced
  Data Sets

18
Georeferenced Data and Image-Map

• Georeferenced Data and Image-Map
• Construct an image-map complete with map grids
  and annotation, and produce an output image
• Start ENVI
• Open and Display SPOT Data
• bldr_reg subdirectory bldr_sp.img
• Edit Map Info in ENVI Header
• Edit Map Information
• The basic map information used by ENVI in
  georeferencing.
• Click on the arrow next to the Projection/Datum
  field
• Click on the active DMS or DDEG button
• Cursor Location/Value

19
Georeferenced Data and Image-Map (cont.)

• Overlay Map Grids
• Overlay ??Grid Lines.
• File ??Restore Setup
• file bldr_sp.grd
• Options ??Edit Map Grid Attributes
• Options ??Edit Geographic Grid Attributes
• Apply
• Overlay Map Annotation
• Overlay ??Annotation
• File ??Restore Annotation
• file bldr_sp.ann
• Object
• Output to Image or Postscript
• Direct Printing

20
Image to Image Registration

• Image to Image Registration
• The georeferenced SPOT image will be used as the
  Base image, and a pixel-based Landsat TM image
  will be warped to match the SPOT.
• Open and Display Landsat TM Image File
• bldr_reg directory file bldr_tm.img
• Band 3
• Display the Cursor Location/Value
• Start Image Registration and Load GCPs
• Map ? Registration ? Select GCPs
• Base Image Display 1 (the SPOT image)
• Warp Image Display 2 (the TM image).
• SPOT image to 753, 826
• TM image to 331, 433
• Add Point
• Show List
• Try this for a few points to get the feel of
  selecting GCPs. Once you have at least 4 points,
  the RMS error is reported.
• Options ? Clear All Points to clear all of your
  points.

21
Image to Image Registration (cont.)

• File ? Restore GCPs from ASCII.
• file name bldr_tm.pts
• Working with GCPs
• On/Off
• Delete
• Update
• Predict
• Warp Images
• Options ? Warp
• Displayed Band.
• Warp Method
• RST
• Resampling
• Nearest Neighbor
• filename bldr_tm1.wrp
• repeat steps 1 and 2 still using RST warping but
  with both Bilinear, and Cubic Convolution
  resampling methods.
• Output the results to bldr_tm2.wrp and
  bldr_tm3.wrp, respectively.
• Repeat steps 1 and 2 twice more, this time
  performing a 1st degree Polynomial warp using
  Cubic Convolution resampling, and again using a
  Delaunay Triangulation warp with Cubic
  Convolution resampling.
• Output the results to bldr_tm4.wrp and
  bldr_tm5.wrp, respectively.

22
Image to Image Registration (cont.)

• Compare Warp Results
• Tools ? Link ? Link Displays
• Load bldr_tm2.wrp and bldr_tm3.wrp into new
  displays and use the image linking and dynamic
  overlays to compare the effect of the three
  different resampling methods nearest neighbor,
  bilinear interpolation, and cubic convolution.
• Note how jagged the pixels appear in the nearest
  neighbor resampled image. The bilinear
  interpolation image looks much smoother, but the
  cubic convolution image is the best result,
  smoother, but retaining fine detail.
• Examine Map Coordinates
• Tools ? Cursor Location/Value
• Close All Files

23
Image to Map Registration

• Image to Map Registration
• The map coordinates picked from the georeferenced
  SPOT image and a vector Digital Line Graph (DLG)
  will be used as the Base, and the pixel-based
  Landsat TM image will be warped to match the map
  data.
• Open and Display Landsat TM Image File
• File ? Open Image File.
• bldr_reg directory file bldr_tm.img
• Gray Scale
• Band 3

24
Image to Map Registration (cont.)

• Select Image-to-Map Registration and Restore GCPs
• Map ? Registration ? Select GCPs
• Image to Map
• UTM
• enter 13 in the Zone text field.
• Leave the pixel size at 30 m and click OK to
  start the registration.
• Add Individual GCPs by moving the cursor position
  in the warp image to a ground location for which
  you know the map coordinate (either read from a
  map or ENVI vector file see the next section).
• Enter the known map coordinates manually into the
  E (Easting) and N (Northing) text boxes and click
  Add Point to add the new GCP.
• File ? Restore GCPs from ASCII
• file bldrtm_m.pts.
• Show List

25
Image to Map Registration (cont.)

• Select Image-to-Map Registration and Restore GCPs
• Add Map GCPs Using Vector Display of DLGs
• File ? Open Vector File ? USGS DLG.
• bldr_rd.dlg
• Memory
• ROADS AND TRAILS
• BOULDER, CO file in the Available Vectors Layers
• Load Selected
• New Vector Window
• Click and drag the left mouse button in the
  Vector Window 1 to activate a crosshair cursor.
• Tools ? Pixel Locator
• 402, 418
• Apply.
• In the Vector Window 477593.74, 4433240.0
• Select Export Map Location. The new map
  coordinates will appear in the Ground Control
  Points Selection dialog.
• Add Point
• observe the change in RMS error

26
Image to Map Registration (cont.)

• RST and Cubic Convolution Warp
• Options ?Warp File
• file name bldr_tm.img
• select all 6 TM bands for warping.
• Warp Method RST
• Resampling Cubic Convolution
• background value 255
• output file name bldrtm_m.img
• Display Result and Evaluate
• Close Selected Files

27
HSV Merge of Different Resolution Georeferenced
Data Sets

• HSV Merge of Different Resolution Georeferenced
  Data Sets
• We will use the TM color-composite image
  registered above as the low-resolution color
  image and the georeferenced SPOT image as the
  high resolution image. The result is a color
  composite image with enhanced spatial resolution.
• Display 30 m TM Color Composite
• file bldrtm_m.img.
• RGB load bands 4, 3, and 2 (R, G, and B) into a
  new display.
• Display 10 m SPOT Data
• file bldr_sp.img.
• Gray Scale
• New Display

28
HSV Merge of Different Resolution Georeferenced
Data Sets (cont.)

• Perform HSV Sharpening
• Transform ? Image Sharpening ? HSV
• Select Input Band SPOT image
• HSV Sharpening Parameters dialog, enter the
  output file name bldrtmsp.img
• Display 10 m Color Image
• Transforms ? Image Sharpening ? Color Normalized
  (Brovey),
• Overlay Map Grid
• Overlay ? Grid Lines.
• File ? Restore Setup
• bldrtmsp.grd
• Overlay Annotation
• Overlay ? Annotation.
• File ? Restore Annotation
• file bldrtmsp.ann
• Output Image Map

29
Orthorectification

• Orthorectification
• Definition
• The geometry of an image is made planimetric
  (map-accurate) by modeling the nature and
  magnitude of geometric distortions in the imagery
• Steps
• Build the interior orientation (aerial photograph
  only)
• Build the exterior orientation
• Orthorectify using a Digital Elevation Model (DEM)

30
Georeferencing Images Using Input Geometry

• Georeferencing Images Using Input Geometry
• Modern sensors ? detailed acquisition (platform
  geometry) information ? model-based geometric
  rectification and map registration
• Users must have the IGM or GLT file as a minimum
  to conduct this form of geocorrection
• The Input Geometry (IGM) file the X and Y map
  coordinates for a specified map projection for
  each pixel in the uncorrected input image.
• The Geometry Lookup (GLT) file the sample and
  line that each pixel in the output image came
  from in the input image.
• If the GLT value is positive, there was an exact
  pixel match. If the GLT value is negative, there
  was no exact match and the nearest neighboring
  pixel is used

31
Georeferencing Images Using Input Geometry (cont.)

• Uncorrected HyMap Hyperspectral Data
• HyMap
• Aircraft-mounted commercial hyperspectral sensor
• 126 spectral channels covering the 0.44 - 2.5 mm
  range with approximately 15nm spectral 162
  resolution and 10001 SNR over a 512-pixel swath.
  Spatial resolution is 3-10 m
• Gyro-stabilized platform
• Open HyMap data
• envidata/cup99hym directory
• File cup99hy_true.img
• Examine Uncorrected Data
• Cursor Location/Value
• Examine IGM files
• envidata/cup99hym directory
• File cup99hy_geo_igm
• Available Bands List dialog
• Gray Scale
• IGM Input X Map
• New Display
• IGM Input Y Map
• New Display

32
Georeferencing Images Using Input Geometry (cont.)

• Uncorrected HyMap Hyperspectral Data (cont.)
• Geocorrect Image Using IGM File
• Map ??Georeference from Input Geometry
  ??Georeference from IGM
• File cup99hy.eff
• Input Data File
• File cup99hy.eff
• Spectral Subset
• File Spectral Subset band 109
• Input Data File
• Input X Geometry Band IGM Input X Map
• Input Y Geometry Band IGM Input Y Map
• Geometry Projection Information
• UTM, Zone 13, datum North America 1927
• the same map projection as the input geometry.
• Build Geometry Lookup File Parameters
• background value of -9999, output filename
• Display and Evaluate Correction Results
• Available Bands List
• Georef band

33
Georeferencing Images Using Input Geometry (cont.)

• Geocorrect Image using GLT File
• Map ??Georeference from Input Geometry
  ??Georeference from GLT
• Input Geometry Lookup File cup99hy_geo_glt
• Input Data File cup99hy.eff
• Spectral Subset
• File Spectral Subset band 109
• Input Data File
• Georeference from GLT Parameters -9999
• output filename
• Display and Evaluate Correction Results
• Available Bands List
• Georef band.
• Cursor Location/Value

34
Georeferencing Images Using Input Geometry (cont.)

• Using Build GLT with Map Projection
• File ??Open Image
• File cup99hy_geo_igm
• Input X Geometry Band
• IGM Input X Map
• Input Y Geometry Band
• IGM Input Y Map
• Geometry Projection Information
• State Plane (NAD 27)
• Set Zone
• Nevada West (2703)
• Build Geometry Lookup File Parameters
• Overlay Map Grids

35
IKONOS and QuickBird Orthorectification

• Orthorectification
• Use the Rational Polynomial Coefficients (RPCs)
  provided by the data vendors with some products
• Orthorectification ????
• Open files
• File ? Open Image File
• ortho subdirectory
• File po_101515_pan_0000000.tif
• File ? Open External File ? Digital Elevation ?
  USGS DEM
• File CONUS_USGS.dem
• USGS DEM Input Parameters dialog
• output filename ortho_dem.dat
• New Display
• Load Band

36
IKONOS and QuickBird Orthorectification (cont.)

• Run the Orthorectification
• Map ? Orthorectification ? Orthorectify IKONOS.
• File po_101515_pan_0000000.tif
• Enter Orthorectification Parameters dialog
• Image Resampling Bilinear
• Background 0
• Input Height
• specifies whether a fixed elevation or a DEM
  (more accurate) value will be used for the entire
  image
• ortho_dem.dat
• DEM Resampling
• Bilinear
• Geoid Offset
• The height of the geoid above mean sea level in
  the location of the image.
• -35 means that the ellipsoid is about 35 meters
  above mean sea level in this area
• Many institutions doing photogrammetric
  processing have their own software for geoid
  height determination, or you can obtain software
  from NOAA, NIMA, USGS, or other sources. A geoid
  height calculation can currently be found at the
  following URL http//www.ngs.noaa.gov/cgi-bin/GEO
  ID_STUFF/geoid99_prompt1.prl
• Save Computed DEM
• Orthorectified Image
• File ikonos_ortho.dat

37
IKONOS and QuickBird Orthorectification (cont.)

• Examine the Orthorectification Results
• Tools ? Link Displays ? Link
• Notice the difference in geometry, especially in
  the upper right corner of the two images. That is
  the result of the orthorectification process

38
(No Transcript)