Remote Sensing Image Enhancement
Image Enhancement ► Increases distinction between features in a scene ► Single image manipulation ► Multi-image manipulation
Single Image ► Contrast manipulation ► Spatial feature manipulation
1. Contrast Manipulation ► Gray-level threshold ► Level slicing ► Contrast stretching ► Histogram-equalized stretching
Contrast Manipulation.. ► Gray-level threshold segmenting an image into two classes - binary mask ► Level slicing dividing the histogram of DNs into several slices
Color-coded temperature maps derived from NIMBUS
Contrast Manipulation.. ► Contrast stretching Expanding a narrow range of DNs to a full range DN - Min Expanding a narrow range of DNs to a full range DN - Min Linear stretch: DN = ( ) *255 Linear stretch: DN = ( ) *255 Max - Min Max - Min ► Advantage: simple computation Disadvantage: rare and frequent values have the same amount of levels
Stretching
Contrast Manipulation.. ► Histogram-equalized stretching ► Stretch based on frequency of occurrence ► Frequently occurred DNs have more display levels ► Special stretch
2. Spatial Feature Manipulation ► Spatial filtering ► Edge enhancement ► Convolution ► Directional first differencing
Spatial Filtering ► Low pass filters emphasize low frequency features ► Compute the average values of moving windows
Low Pass Filter MeanMoving windows
Spatial Filtering.. ► High pass filters emphasize local details ► It subtracts the low-pass filter from the original image
Edge Enhancement ► Add back the high frequency image component to the original image ► Preserve both the original and the high frequency features
Convolution ► A moving kernel with a weighting factor for each pixel
Convolution
Directional Differencing ► Displaying the differences in gray levels of adjacent pixels ► The direction can be horizontal, vertical, or diagonal ► It is necessary to add a constant to the difference for display purposes ► Add back the directional difference to the original image ► Contrast stretching is needed for all feature manipulations
Convolution
3. Multi-image Manipulation ► Spectral ratioing ► Principle component transformation ► Kauth-Thomas tasseled cap ► Intensity-Hue-Saturation transformation (IHS)
3.1 Spectral Ratioing ► A ratio of two bands (with great difference in reflectance) ► Useful to eliminate effects of illumination differences ► Select bands with distinct spectral responses ► Necessary to stretch the resultant values to a full range of DN values after ratioing
Band Ratioing..
► Based on the observation that the DNs for a same feature are lower in the shadow, and the DNs are reduced in a similar proportion between features =÷ Band ABand BRatio Band÷=
Hybrid Color Ratio Composite ► Problem: different features but of similar ratio may appear identical ► Solution: when display, combine two ratio bands + one original band to restore the absolute DN values
3.2 Principle Component Transformation ► To reduce redundancy in multi-spectral data ► The transform DN I = a 11 DN A + a 12 DN B + a 13 DN C + a 14 DN D DN II = a 21 DN A + a 22 DN B + a 23 DN C + a 24 DN D DN III = a 31 DN A + a 32 DN B + a 33 DN C + a 34 DN D DN IV = a 41 DN A + a 42 DN B + a 43 DN C + a 44 DN D DN I, - DN IV, - DNs in new component images DN A, -DN D - DNs in the original images a 11, a 12,,,, a 44 - coefficients for the transformation
PCA
PC Transformation.. ► After the axes rotation, the original n bands images are converted into n principle components images ► The first component (PC1) image contains the largest percentage of the total scene variance (90%+) ► The second component (PC2) contains the largest of the remaining variance
PC Transformation.. ► Percentage of variance explained by each component ► %: ► Cul:
PC Transformation.. ► Loading: the correlation between each band and each PC for output interpretation purposes Components Band … …… 567
PCA
PC Transformation.. ► Successive components are orthogonal, and they are not correlated to each other ► PCs can be used as new bands for image classification ► PCA is scene specific
3.3 Kauth-Thomas Tasseled Cap ► An orthogonal transformation ► The 4 MSS bands can be converted into 4 new bands: brightness greenness yellow stuff yellow stuffnon-such
K-T Tasseled Cap ► SBI = 0.332MSS MSS MSS MSS7 ► GVI = MSS MSS MSS MSS7 ► YVI = MSS MSS MSS MSS7 ► NSI = MSS MSS MSS MSS7
Kauth-Thomas Tasseled Cap ► The first two indices contain the most info (90%+) ► Brightness is related to bare soils ► Greenness is related to the amount of green vegetation
Kauth-Thomas Tesseled Cap
Kauth-Thomas Tasseled Cap ► The 6 TM bands can be converted into a 3D space: plane of soil plane of vegetation and a transition zone ► A third feature, wetness ► The K-T transformation is transferable between scenes
K-T for TM ► Brightness = 0.33TM TM TM TM TM TM7 ► Greenness = -0.25TM TM TM TM TM TM7 ► Third = 0.14TM TM TM TM TM TM7 ► Fourth = 0.85TM TM TM TM TM TM7
3.4 IHS ► Intensity-Hue-Saturation transformation (IHS) ► Transform the RGB space into the IHS space to represent the information ► Intensity: brightness ► Hue: color ► Saturation: purity
IHS ► The hexcone model projects the RGB cube to a plane, resulting in a hexagon ► The plane is perpendicular to the gray line and tangent to the cube at the "white" corner
IHS ► Intensity = distance along the gray line from the black point to any given hexagonal projection ► Hue = angle around the hexagon ► Saturation = distance from the gray point at the center of the hexagon
IHS ► I,H,S = f(R,G,B) I' = f(I+Ipan) I' = f(I+Ipan) H' = f(H+Hpan) H' = f(H+Hpan) S' = f(S+Span) S' = f(S+Span) R',G',B' = f(I',H',S') R',G',B' = f(I',H',S')
Readings ► Chapter 7
PCA..