purpose of image enhancement (human/automation)
- improve clarity of information in images for human viewers,
- provide `better’ input for other automated image processing techniques
what is image enhancement
process of adjusting digital images for results better suited for display/ further image analysis.
Eg. remove noise, sharpen, or brighten an image, making it easier to identify key features.
image enhancement (method) categories (2)
- Spatial domain methods
- Frequency domain methods
what is SPATIAL domain method in image enhancement n whats it based on
Manipulate the image pixel value in spatial domain;
based on distribution statistics of the entire image/ local region.
Technique are based on direct manipulation of pixels in an image. eg. Mean Filtering, Medial Filtering
what is FREQUENCY domain method in image enhancement n whats it based on
Manipulate information in the frequency domain;
based on modifying the Fourier Transform of an image
s = T(r) [value of input n output pixels]
T is the transformation that maps a pixel value r into pixel value s
types of transformation in Gray Level Transformation (3)
- Linear (identity transform & inverse transform)
- Logarithmic
- Power-law
Linear Identity Transformation in Greyscale T. (2)
- Output intensities identical to input intensities
- Does not have any effect on image. But added to the previous graph for completeness
Linear Inverse Transformation in Greyscale T. (3)
- opposite of identity transformation.
- Each value of input image subtracted from L-1 and then mapped onto the output image.
- light areas appear dark & dark areas appear light
- Suitable for enhancing white or gray details embedded in dark regions esp when dark areas are dominant
S = T(r)
= (L-1)-r
linear inv. transformation
L number of gray levels
Logarithmic Transformation in Greyscale T. [exp]
expands values of dark pixels n compresses lighter pixels in img
Logarithmic Transformation in Greyscale T. effects and uses (1,2) [maps]
Effects:
Maps a narrow range of low pixel levels -> wider range of output values.
opposite is true for higher input values
Uses:
- very useful when input pixel values have extremely large range.
- Suitable for compressing dynamic display on devices.
s = T(r)
= c log(1+r)
logarithm transformation
c is a constant
r >= 0
Inverse Logarithmic Transformation in Greyscale T. (2)
- opposite of log transform
- expands high pixel values n compresses darker level values
Power (Gamma) Transformation in Greyscale T. effect (2) [maps]
- Maps a narrow range of dark input values -> wider range of output values and vice versa.
- More precise than log
s = T(r)
= c r
power trans
c & (gamma) are +ve constant
