Bayesian Correction of Image Intensity with Spatial Consideration

Jiaya Jia    

Jian Sun   

Chi-Keung Tang   

Heung-Yeung Shum   

Abstract ! Abstract. Under dimly lit condition, it is difficult to take a satisfactory image in long exposure time with a hand-held camera. Despite the use of a tripod, moving objects in the scene still generate ghosting and blurring effect. In this paper, we propose a novel approach to recover a high-quality image by exploiting the tradeoff between exposure time and motion blur, which considers color statistics and spatial constraints simultaneously, by using only two defective input images. A Bayesian framework is adopted to incorporate the factors to generate an optimal color mapping function. No estimation of PSF is performed. Our new approach can be readily extended to handle high contrast scenes to reveal fine details in saturated or highlight regions. An image acquisition system deploying off-the-shelf digital cameras and camera control softwares was built. We present our results on a variety of defective images: global and local motion blur due to camera shake or object movement, and saturation due to high contrast scenes.

Publications:

BibTex:

@inproceedings{Jia2004correction,
    author = {Jiaya Jia and Jian Sun and Chi-Keung Tang and Heung-Yeung Shum},
    title = {
Bayesian Correction of Image Intensity with Spatial Consideration},
    booktitle = {ECCV},
    year = {2004},
    pages = {III:
342-354}
}

 


Motivations: Under dimly lit condition, it is difficult to take a satisfactory image in long exposure time with a hand-held camera.

In this project, we study the above imaging property, and propose to combine two imperfect images (one is blurred, but has sufficient exposure; one is clear, but is too dark) to re-produce a visually satisfying image f*. Our method is a Bayesian approach as shown in the following:


Static object results:

 

Blurred images   Under-exposure images   Corrected images

 

+

 

=

 

 

+

 

=

 

 


Moving object results:

Our method can also be applied to correct images containing high-speed-motion objects, or partially blurred images given the same configuration.

Blurred images   Under-exposure images   Corrected images
+ =

 

 

Handling high-dynamic-range scenes:

In high dynamic range scenes, after color correction using two defect images, the intensity of some pixels may be large than 255. Our method can directly produce high-range images (grayscale level > 256). After tone mapping operations, our result images can show all details in a commonly-used monitor.

 

  Blurred images with over-exposure   Under-exposure images   Corrected images
Images + =
Local view