Image Enhancement using Alpha

Rooting based Hybrid Technique

Image Enhancement

Image Enhancement in Spatial domain

Gray Level Transformation Functions

Gray Level Transformation Functions

Negative Image Transformation

s = 1 - r

s = 1 – r

where, s = output image gray level

r = input image gray level

Log transformation

s = c log ( 1 + r ) where, c = scaling factor

Power law transformation function

Power law transformation function

Power law transformation function

Power law transformation function

By Alpha Rooting

Image Enhancement in Frequency domain

Orthogonal transformation

Image

|X(p,s)|

Magnitude(contains frequency

content of image)

θ(p,s)

Phase(angle of the

magnitude content)

Orthogonal transform

X(p,s) = |X(p,s)| e jθ(p,s)

Orthogonal transformation

Inverse Orthogonal transform

Enhanced Image

Operation

Separation

Combination

Alpha Rooting

X(p,s) = |X(p,s)| e jθ(p,s)α^

• Operation in orthogonal transformation,

where, 0 < α < 1

• here, X(p,s) is alpha rooted image^

(a) (b)

Alpha Rooting Based Enhancement Procedure

Conventional Alpha Rooting results in• Subtle edge information in images• Increases sharpness• Makes image crisper

Enhancement in sharpness is mostly subdued by overall darkeningof image which is artifact of alpha rooting

Requirements• Eliminate Tonal change• Achieve good contrast and brightness

Remedy• Apply log transform• Apply power law transform

Proposed Method

Input Image

Apply Orthogonal Transformation

Separate Magnitude and phase coefficients

X(p,s) = |X(p,s)| e jθ(p,s)

Apply Alpha Rooting on magnitude|X(p,s)|α

Combine Magnitude and

phase components

Inverse Orthogonal Transform

Apply Log Transformation

s = c log(1+r)

Apply Power Law Transformation

s = b rγ

Enhanced Image

Experimental Results

(a)

(b) (c) (d)

Fig (a) : Original ImageFig (b) : Alpha Rooted using α = 0.72Fig (c) : Hybrid Alpha Rooted using α = 0.78, b = 1.2, γ = 0.8Fig (d) : Hybrid Alpha Rooted using α = 0.7, b = 1.2, γ = 0.8

Experimental Results

Fig (a) : Original ImageFig (b) : Alpha Rooted using α = 0.78Fig (c) : Hybrid Alpha Rooted using α = 0.75, b = 1.1, γ = 2

(a)

(b) (c)

CONCLUSION

In this paper we elucidated an Alpha Rooting Based HybridProcedure for image enhancement which is free of the conventionallimitations associated with the transform domain imageenhancement techniques.

Thank You !!!