I am having set of pair of line segments (2D). Though each pair should be coincided on top of each other they are not so. I have been the reference data and then I extracted other line segments via an image based approach. for e.g. CD is an extracted line segment and AB is its reference segment. Now I know coordinates of end vertices i.e. vector data. AS YOU CAN SEE, LENGTH OF CORRESPONDING PAIRS ARE NOT EQUIDISTANT. SOME TIME PARALLEL AND SOMETIME OBLIQUE AS SHOWN IN FIGURE.
So, Now I want to find is there any systematic patterns within my extraction. Also, I want to give an overall figure for the accuracy of image based method with respect to reference line segments.
For that, I am not quite sure how would be my measure.. I mean should I use rmse or some other measures for this. (Actually, I can not thing how rmse should be measures as my lines are not parallel and equidistant)
SO, I want your idea to estimate
- any systematic errors of my extractions
- to say something like mean error for the whole extraction.
any help please.