In remote sensing, using panchromatic band with high spatial resolution to fuse spectral bands is a way to enhance images. That makes images rich in spectral and spatial resolution simultaneously.
Without having a reference image, how can we evaluate the result?
OK this is what I got:
some indexes are available when we haven't a reference image. They compare fused image with high spatial or high spectral images to evaluate and measure amount of details that is implanted from them.
But I looking for some strong evaluations so:
What is the way to access the reference image?
Yes there is good review for assessment of performance of imsge fusion algorithms.
I am just trying whether entropy. :. -p log p, could be one of the useful index. Some literature is available. But can some one has any tested resullts on using Entropy as imformation measurement of fused images by different alhorithms or spectral combinations. Thanks in advance
Dear Mr.Reza
There are many studies discussing A Novel Metrics for Performance Evaluation of Image Fusion Algorithms such as the attached papers...
You can see the paper for FMI metric:
http://dx.doi.org/10.1016/j.compeleceng.2011.07.012
The MATLAB code for this metric is available at:
http://www.mathworks.com/matlabcentral/fileexchange/45926
I saw a method that is used frequently in IEEE papers. In this method the original image is assumed as reference image. Then this image is downgraded to lower resolution. If image is for Quickbird, level of downgrade will be 4. Also the Pan image is downgraded to fit the process.
At last we have original image as reference image, downgraded image as low spatial image and Panchromatic image as the source of detail.
Now we are able to compare methods!
- Badges
- Science topic
- Similar topics
- Engineering