How adequate is the implementation of the FAST corner detector algorithm?

Question

Algorithms

Hello, as part of the coursework, I had to deal with the FAST angle detector (namely FAST, not FASTer). I decided to look at the source code of the author of the algorithm (the source code can be viewed here ). After scrolling through two hundred lines, if-else-goto came to some bewilderment. Decided to read their post and make my own. My code contains less than a hundred lines, while it works similarly to the original one (there is, however, some error during operation: some points are not detected by me, but the number of such points is not more than 10 per test case).
Actually the question is: who can explain what is the reason for using such a "footcloth" by the authors of the algorithm? In theory, such a large number of if-s leads to a decrease in performance. Or am I wrong?

Answer 1

pastebin.com/E5bzULQq - we have a reference example of Hindu code. No matter how much the author calls his smeared chocolate academic work, it does not change the essence. If any junior showed such a code to his team leader, it would be his last day of work. It's strange, but in academies, such code rolls, as you can see.
Do not try to correct anything in this code (you touch it, then you will not wash off), try to understand what the author meant by all this and rewrite from scratch.
If some 10 points are detected by his code but not by yours, this does not mean that his code is correct and yours is not.
But still, it is worth double-checking your code and comparing its behavior at these points in the debugger with the original.

Answer 2

some points are not detected by me, but the number of such points is not more than 10 per test case

only called this - does not work :

Mum! Mum! I have already learned how to use the potty ... but not always well.