top of page
Search

Week 10

brenden728

Visible/Near Infrared Fusion

  • With the algorithm finally running well on the H7 we took a step back and spent this week working on the MATLAB code since it can actually show results

  • we played with the model parameters for quite some time to see how much detail we could extract


Normal RGB Image


Visible and NIR Fused Image


You can see extra rings in the fence and extra ridges in the darker portion of the tree










To us, these are pretty promising results, however, we are also noticing the issues in this algorithm that were explained in other papers as well as the paper we consulted: Glowing/Artifacting. In a way, this algorithm brightens a lot of dark edges and generally hard to see parts of an image. But we cannot seem to find parameters that are a good balance of added quality and less glowing/artifacting.


The final issue is finding good RGB/NIR image pairs to use. Having failed to find a good piece of hardware that could accommodate our needs, we are stuck with medium resolution images courtesy of a specific website in which we will credit.


Conclusion:


In the end, the algorithm does exactly what it is intended, it adds information to the image in its darker location by making use of information saved in the NIR image. The improvement in quality however, may be subjective...

 
 
 

Recent Posts

See All

Week 9

Visible/NIR Fusion The H7 algorithm is running!!! Bochi and I finally found out that the order in which we ran each modular parts of this...

Week 8

Visible/NIR Fusion Made a lot of optimizations to the H7 code, and added the functionality to load both the Visible Light and NIR images...

Week 7

Visible-NIR Fusion Wrote the High Pass Filter code on the H7 Code uses a convolution (really correlation) sliding filter just like the ML...

Comments


bottom of page