Please over look my lack of language skills, I just want to share ideas where many others over the years already thought of & exercise.
Objective: To remove the problem parts in 'red'.
Conventional working knowledge (outside of knife sharpening) tells us to directly remove those 'red' parts and leave 'green' good parts alone. If these figures represent edges of aluminum sheet where the 'red' part is 3mm thick, simply we would just file the 'red' part off. For a knife, the 'red' part usually less than 40um (all the way to submicron) thick but somehow the removal technique deviated from common/file method. Why? Let's see
a) Steeling - Abrading component is more or less filing the 'red' part off. The burnishing component also affect the good 'green' part.
b) Strop - Fine for fig1 but for fig2&3, one must be careful not to scrape strop or stone. The good 'green' part directly affected.
c) Sharpen - For fig2&3 problem. deburr for fig1. The good 'green' part directly affected. Improper sharpening may end up with fig1 problem.
a) b) c) approaches are affecting the good 'green' parts in a lack of total control manner. The end result edge 'green' is altered into new shape or same shape but minutely shorter blade height.
Using conventional/file method.
1) For fig1, use a flat piece of wood bent(file/scrap perpendicular to edge) the wire over, changed problem into fig2. Treat fig2 & fig3 problems as fig3.
2) Hold the knife with edge facing you. Use flat stone(or round stone for recurve) and file away the 'red' part at the target angle.
* Abrasive selection is important, use abrasive 1/3 to 1/10 the thickness of the 'red' part worked well for me.
3) Use a flat pieace of wood (mdf and or balsa) to file/scrap the remnant of wire to bent over the opposite side, repeat 2) & 3) until done. Done when edge is free of wire/rolled/burr.
This approach directly attacks the problem 'red' area and with more control, while leaving the good 'green' area mostly intact.
For filing 90% of the time, I use DMT diasharp EF/EEF. For the rest 10% - dealing with super fine edges, I use home-made files with grit range from 12microns down to 0.1micron.
Whew, don't you glad I'm not rambling about cosmo physics