Cannot understant theroy of HOG

[MeineLiebeAxt]

Hi,
I am learning unit3 in course “OpenCV for Robotics”. In theory of HOG, I found this point a little confused:

The distance to 40 and 60 is 5 and 15, respectively; 
therefore, the ratios will be 5/20 and 15/20, 
so the magnitude assigned will be Gm*1/4 for 40 and Gm*3/4 for 60.

I am confused about: why is it assigned more magnitude to a further direction? It is obviously that 45 is closer to 40. If it is like this, then 41 is only with difference of 1 to 40, they are almost the same direction, but 40 can only be assigned with G_m*1/20 ?
2.
What is actually assigned? The exact direction of 40 degree or 60 degree, or the interval between 40 and 60? According to the rule above, it is the exact direction of 40 or 60 degree that are assigned, not the interval. Then what is the meaning of this histogram? In histogram it is the interval that really counts.
3.
And the contxt sayd 180 is split into 9 bins, then in a complete circle, we should have 18 bins for 360 degree, right? Then why are the arrow-icons only have 10 direction?
Please help with theoredical explanation…

[girishkumar.kannan]

Hi @MeineLiebeAxt ,

Histogram of Gradients [HoG] is a concept of Computer Vision that is commonly used as a feature descriptor for images for learning about the image.

When I learned Computer Vision course in my University I was taught from the same content as in the link:

Histogram of Oriented Gradients explained using OpenCV

This explains the use of HoG at best. HoG is something similar to Fourier Transform (FFT) that we would do in sound analysis - but this is for images.

Also refer this link for a quick intro about the theory: HOG (Histogram of Oriented Gradients): An Overview | by Mrinal Tyagi | Towards Data Science

It is actually better to understand HoG concept by doing it from scratch - without using OpenCV to understand it better. That is how I learned that concept.

Regards,
Girish

[MeineLiebeAxt]

Thank you! @girishkumar.kannan