enhancement of backlight scaled images
TRANSCRIPT
ENHANCEMENT OF BACKLIGHT SCALED IMAGES
P. VIBHA BHANDARY4SF10EC101
INTRODUCTION
Goal- to reduce undesired effects of dim backlight on image quality with backlight 10% or less.
Human Visual System( HVS)- a perception based approach.
Luminance < threshold => image detail invisibility
INTRODUCTION
BACKGROUND
1) Just Noticeable Difference (JND):
1. Linear curve-behavior
2. Smallest difference in the sensory input discernible by human being.
3. ΔL= J(L)=0.0594(1.219+L 0.4 )2.5
L= Background luminance
2) Human Visual Response Model:
1. Nonlinear curve-behavior
2. Li= Li-1 +J( Li-1) , L>0
3. Li reaches upper bound of luminance range
3) Effects of dim backlight on images:
V/Vm = normalized response
I=Perceived light intensity
σ= half saturation parameter
APPROACHES
PROPOSED ALGORITHM
1) Prediction of the Detail Loss Effect
PL= PLF (1-PL
D )
2) Enhancement of invisible pixels
JND Decomposition:
E= 2D tan (2.5 π/180)
Luminance Boosting and Compression:
Color Restoration:
Compensation for the Halo Effect
PERFORMANCE EVOLUTION
ABIE: Adaptive Backlight Image Enhancement
CBCS: Concurrent Brightness Contrast Scaling
TABS: Temporally aware backlight scaling
GD: Gradient Domain
Subjective Evaluation
Objective Evaluation
Visualization
CONCLUSION
Algorithm effectively
1. enhances the visibility of image dark region.
2. applies counter shading to eliminate halo effect.
3. enhances perceptual contrast of bright regions.
“A goal is not always meant to be reached. But often something to aim at”
-Bruce Lee