Compare images based on dimension or shape similarity index?

I have a CAD design of a simple model that is being 3D printed. To evaluate the print quality we used optical microscopy images of the model and qualitatively decide on the printing quality. I need to make this procedure quantitative, but I am not sure which index, or algorithm to use. My main attribute of quality is dimension and shape similarity. I tried the structural similarity index (SSIM), but since I did it on gray scale images and sometimes the microscopy images are not identifcal in contrast and exposure the results were not convincing. So is SSIM good for what I want but with more care to image processing may be by comparing binary images rather than grayscale? or is there another procedure/index better?

Answer 1

[yanqi liu]

sir，may be use psnr、ssim、mssim、mse to make some compare

Answer 2

[Image Analyst]

I'd probably threshold and use dice().