In a recent study with Emgucv image registration, from the extraction of feature points to measure of feature matching to transform model, all is not very good, halfway out, hope to have a great god give me a reference example code,

CodePudding user response:

https://wenku.baidu.com/view/3b1c1bf76bd97f192379e907.html

CodePudding user response:

Code function at the end of the day is not others to help you see or interpretation or comment; But by myself calm down and take a long enough time and energy to do it yourself step or set a breakpoint or step to perform to a certain intermediate result shows or written to the log file analysis step by step,
Remind: cow x teacher cannot replace the student understand and use the toilet!
Single step debugging and set breakpoint debugging (VS IDE compilation connection through later, press F10 or F11 key step, press Shift + F11 exit the current function; In a press F9 set breakpoints after press F5 execution stops at the breakpoint,) is one of the programmers must master the skills,

CodePudding user response:

I recently also studied the image registration, roughly divided into two kinds of thinking, and global search based on feature points, mainly is the sift algorithm based on feature points, probably will feature points with 128 d feature vectors, and then matching opencv has written function, want to read the code, need to understand the principle of first, I study in a couple of days, but not mathematical skills, or a little knowledge, the use of words, direct call opencv library function, based on global search more simple, is to use optimization algorithm to optimize the x and y Angle to find the optimal solution, the two methods have limitations,