- GREAT Day
- (Day of student research talks, art exhibits/performances, etc.)
- Tuesday, April 19
- Extra credit (up to 2 lab points per summary) for math session summaries
- 1 paragraph to 1 page discussing whatever connects the presentations in a session
- Max 3 sessions per person can get extra credit
- Subject and background pictures can be anything, from anywhere
- Joe Dolce, CIT video specialist, has a professional green screen
- Willing to help anyone who wants to try it do so; also willing to help with other image processing
- But do it after GREAT Day, please!
- dolce at geneseo.edu
- Algorithm
- Can work with a white screen if test for background is that all color intensitities are above some threshold, and red, green, and blue intensities are all within some small fraction of each other
- x is within fraction f of y:
- x > y-fy & x < y+fy
- x > (1-f)y & x < (1+f)y
- Can keep track of which pixels are background in a mask matrix
- Algorithm with most details worked out looks like this:
chromakey( subject, background, low, f ) % "low" = lowest intensity to be considered white % high level step 1: look for subject pixels with RGB in certain range % track background pixels w/ mask matrix % mask has 1 (aka true) on white background pixels % Still need to figure out how to combine the 4 tests below into one mask mask = subject > low red > (1-f)*green & red < (1+f)*green green > (1-f)*blue & green < (1+f)*blue red > (1-f)*blue & red < (1+f)*blue % high level step 2: replace found pixels in subj with same pixels f/ bkgnd % C is composite image C = mask .* background + (1-mask) .* subject % Or... C = subject C( mask ) = background( mask ) % In either case... return C- Biggest thing not addressed here is the fact that image matrices actually have three dimensions but mask only has two
- Chroma key lab day
- No reading