Visible/IR Fusion
This week involved implementing a specific high pass filter. The high pass filter in the picture below is a gaussian filter subtracted from a delta function h = d - g.
Though it has more resolution below, when implemented to extract high frequency details from the Infrared image, it will have a kernel size k = 19, and a radial cutoff frequency of 0.05 cycles/picture height to start
According to the IEEE paper regarding this filter in the use case of Visible/IR Fusion, it will help extract Spatial Details from the IR image that will be weighted and combined with the visible light image
Now, since my partner Bochi has finished the first try of the luminance channel (LC) extraction, we will try to combine the filter and the LC to make the fusion map which gives the exact weighting scheme for application of the IR image to be applied to the visible light image. We should be able to get a detailed visible image by next week.
The rest of my week involved trying to understand image resolution descriptions ( lines/pixel, cycles/pixel, cycles/picture height) and I am still having a lot of trouble trying to understand them
Comments