[attachment=6560]

HIERARCHICAL COLOR CORRECTION FOR CELL PHONE CAMERA IMAGES


Presented by
CH.RAMESH
(06C61A0414)

---

ABSTRACT
For enhancing the color of digital images obtained from low-quality digital image capture devices ,we propose color correction algorithm.
The hierarchical color correction is performed in three stages.
We compare the performance of the proposed method to other commercial color correction algorithms on cell phone camera images.

• INTRODUCTION
• Digital color enhancement is the process of adjusting the color values in an image.
• The processed image i.e output image looks pleasant to human viewer.
• It involves various processing steps like contrast enhancement ,dynamic range compression and improving color rendition.

DISTORTIONS
Typical distortions in cell phone camera photos include ::
Poor contrast
Incorrect exposure
Color fringing
Color infidelity
MSRCR is a non-linear color correction algorithm whose goal is to improve the overall image quality.

Hierarchical color correction
It achieves color enhancement by recovering the minimum mean squared error(MMSE).
The proposed algorithm is based on a hierarchical stochastic framework.
This algorithm is used to achieve multiple color enhancement objectives.
In this algorithm we will be concerned with correcting the color balance and color infidelity issues.
The color defects we are trying to fix can arise either due to imaging hardware & image processing in cell phone camera.


MULTI LAYER HIERARCHICAL FRAME WORK
The proposed color correction includes two layer hierarchical classifier in order to identify the color distortion.
Global and local image classifications are the two layers in frame work.
The global image alg classifies the image into different groups ,where each group specifies images showing a similar defect behavior.
The cell images includes either too reddish ,too blue or too greenish/yellowish compared with reference images.

EFFICIENT COLOR CORRECTION
Two critical parameters that affect the performance of the color correction algorithm are the number of global classes M and the number of subclasses in each of the MRSCC predictors.
The efficient color correction algorithm is developed in order to reduce computation while preserving the output image quality.

Conclusion
The proposed color correction algorithm, based on global and local classification of image color attributes, provides a robust color correction.
Both subjective evaluation and a visually weighted image quality metric shows the proposed algorithm performs better on camera cell phone images