Log in

No account? Create an account

Previous Entry | Next Entry

I've had this image-processing project on a back burner for about two years, trying to decide whether to try and use one of the publicly-available image-processing libraries or roll my own (very simple, very specialized) recognition algorithms. Some of the libraries out there don't actually do what I want -- they blur, they sharpen, they flood connected shapes with the same color, they stretch and skew and so forth. Others seemed like they might have some functions that I wanted, but every single threshold algorithm and edge-finder and subregion-matcher and feature-extractor just gave me another damn image, when what I wanted was data about where things in the image were.

And we all know that images are impenetrable black boxes, suitable only for putting pretty pixels up on a screen.

Like I said, how fscking stupid I am. So now I just have to decide which library to use, which probably means who has what bindings to which language I hate the least, including something with the ability to snarf webcam images without too great an excess of plumbing. Oy. I coulda done this back before I broke half the equipment I'm trying to extract data from.