Intro :

 we present an improved method for low light exposure image enhancement using exposure
fusion technique which is very useful for directly fusing multi-exposure images into a high
quality image, without the physically based high dynamic range device.
 The method for multi-exposure images fusion is by using multi-resolution Laplacian
pyramid weighted blending.
 We propose a Weight Modification Factor to modify the original weight map for each image
guided by contrast, saturation and well-exposedness of each pixel in the image, and to
enhance the weight of the informative pixels in ultra-bright or ultra-dark areas
 our approach preserves more useful details for scenes with very large dynamic range

 These weight maps are then refined using a guided filter, and the fusion is performed
using a pyramidal decomposition.

Proposed wrk:
 When performing Component Analysis on images, one must create a vector of features,
where each pixel's intensity is a feature and each image is represented as a vector.
 Component Analysis assists us in identifying patterns based on their correlation, or it is
simply a feature extraction technique that allows us to drop the least important information
while retaining the important information in the dataset.
 Laplacian pyramid decomposition is used to compute weight maps using local efficient
energy and pyramid correlation
 Weight Modification Factor to modify the weight of Laplacian pyramid coefficients, making
the weights of informative pixels dominant even if on the high level pyramid.
 Using Mertens we define several pixel quality metrics , which determines the weight map of
each image according to contrast, saturation and well-exposedness.
 We discover information-rich regions in images using gradient magnitudes or a bilateral
filtering process
 The weights aims to maintain pixel intensities that are not close to 0 or 1.
 As a result, the favour pixel intensity in well-exposed pixels for weight extraction will be
close to 0.5
 but this features is not always sufficient for preserving bright-exposure images.
  To address this issue, we proposed well-exposure based on the mean of pixel
intensities in the exposure image.
 So, the adaptive well exposedness that we proposed is used to allocate large weights to dark
regions when the image is long exposed and small weights to bright regions when the image
is short exposed.
 To display the synthesized HDR image on an LDR display device, we need a tone-mapping
process to compress the HDR into the LDR [3,4].

Conclusion:
 we obtain a detail-preserving and visually natural image in a multi-resolution fashion by
fusing a sequence of multi-exposure images directly.
 We propose Weight Modification Factor to modify weight maps, increasing the weight of
informative pixels in ultra-bright and ultra-dark areas and preserving detail information.

 Our approach has a wider adaptability to scenes with very large dynamic range. The quality
of our image is superior from the initial product from a LDR device