Talk:Summed-area table
 | This article has not yet been rated on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | |||||||||||||||||
| ||||||||||||||||||
This article needs a lot of clearing up. Although the formulas are correct its notation is unclear for people who don't know the Viola/Jones paper. What is A, B, C, D? A sum, area or a point? What is the function A(x)? What is i(x,y)? What is the one argumented function I(X) opposed to I(x, y)? All these are essential points missing. --85.177.99.146 (talk) 13:41, 26 January 2013 (UTC)
The formula seems to be incorrect, the sum I(C)+I(A)-I(B)-I(D) calculates rectangle NOT including A, that is x0<(!)<=x1 and y0<=y<(!)y1.--46.223.253.194 (talk) 12:08, 26 January 2014 (UTC)
Actually, the sum I(C)+I(A)-I(B)-I(D) calculates rectangle not including (A, B) and (A, D) edges, that is x0 < x and y0 < y. I tested that one in my c++ project. — Preceding unsigned comment added by 188.124.228.2 (talk) 06:48, 11 April 2014 (UTC)
Popularization
The statement "In computer vision it was first prominently used within the Viola–Jones object detection framework in 2001" is not correct. I recall publications by Lewis (and others e.g. Sun) in the 1990s that used this technique. Checking, I see two 1995 publications by Lewis (Fast Normalized Cross Correlation and Fast Template Matching) that have 970 and 600 citations respectively on google scholar. The computer vision conference they appeared in (Vision Interface) is obscure now, but with the publications themselves were well known. — Preceding unsigned comment added by 216.9.110.15 (talk • contribs)
See the textbook Concise Computer Vision, by Reinhard Klette, Springer, 2014. On p.52, a section The Introduction of Integral Images into Computer Vision: "Integral images have been introduced in the Computer Graphics literature in [F.C. Crow. Summed-area tables for texture mapping. Computer Graphics, vol. 18, pp. 207–212, 1984] and then popularized by [J.P. Lewis. Fast template matching. In Proc. Vision Interface, pp. 120–123, 1995] and [P. Viola and M. Jones. Robust real-time object detection. Int. J. Computer Vision, pp. 137–154, 2001] in the Computer Vision literature." The current wording of the page highlights only Viola&Jones while ignoring the earlier popularization by Lewis. I may try to edit the page 203.176.188.16 (talk) 05:46, 15 February 2016 (UTC)
- The Frank Crow paper has over 1200 citations. Why isn't that considered the popularization of this idea? Dicklyon (talk) 06:49, 15 February 2016 (UTC)
- Oh, I see, it's "in computer vision"; missed your point. Dicklyon (talk) 06:50, 15 February 2016 (UTC)
Relationship to mip-maps
Sorry can't investigate it right now (I'd have to find the original papers but don't have access to them), but I don't think summed-area tables were introduced "for use with mip-maps". I seem to recall that summed-area tables and mip-maps were alternative methods. Raichu2 (talk) 02:32, 25 October 2018 (UTC)
2D prefix sum
Surely someone has mentioned this is just a 2D version of