My problem: I need to check if the rendered image on the screen is correct compared to the ground truth. There will be tons of ground truth cases,like rendering a triangle, a line, a circle or more complicated cases. And the rendered image may be scaled due to different screen settings.
My thoughts: I can't store all the ground truth in its original form due to the space they took up. I'm thinking of store them in the form of arrays of parameters, then compare the rendered picture parameters to the ground truth parameters.
What I'm trying to achieve: Targeted rendering issues will include shift in position, color, or rough edge of a circle, missing pixels. Basically any pixel level difference to ground truth will be considered.
I need help: Can you suggest parameters or algorithms to tackle this?