Is there the mis-typing for integral image's document?

scorpiozj · July 10, 2022, 10:36am

I’d like to calculate integral image by using the API cv::integral. When I’m trying to understand how the tiled image is calculated, I think there may be something mis-typing in the document. I looked the v4.4.5, and the link of the document is here.

In the text of the screenshot, it uses a Rect(3,3,3,2), but it shows P0 = {4,4} in the left.
While for the right part for the titled image, the P0 equals {1,5}, and it meets the titled rect(5,1,2,3).

So I think that for the left chart, the Rect should be (4, 4, 3, 2).

crackwitz · July 10, 2022, 10:54am

yeah, probably.

if you want, you could submit a pull request. just change the text here:

github.com

opencv/opencv/blob/4.x/modules/imgproc/include/opencv2/imgproc.hpp#L2812


      
          
          
Using these integral images, you can calculate sum, mean, and standard deviation over a specific
          up-right or rotated rectangular region of the image in a constant time, for example:
          
          
\f[\sum _{x_1 \leq x < x_2,  \, y_1  \leq y < y_2}  \texttt{image} (x,y) =  \texttt{sum} (x_2,y_2)- \texttt{sum} (x_1,y_2)- \texttt{sum} (x_2,y_1)+ \texttt{sum} (x_1,y_1)\f]
          
          
It makes possible to do a fast blurring or fast block correlation with a variable window size, for
          example. In case of multi-channel images, sums for each channel are accumulated independently.
          
          
As a practical example, the next figure shows the calculation of the integral of a straight
          rectangle Rect(3,3,3,2) and of a tilted rectangle Rect(5,1,2,3) . The selected pixels in the
          original image are shown, as well as the relative pixels in the integral images sum and tilted .
          
          
![integral calculation example](pics/integral.png)
          
          
@param src input image as \f$W \times H\f$, 8-bit or floating-point (32f or 64f).
          @param sum integral image as \f$(W+1)\times (H+1)\f$ , 32-bit integer or floating-point (32f or 64f).
          @param sqsum integral image for squared pixel values; it is \f$(W+1)\times (H+1)\f$, double-precision
          floating-point (64f) array.
          @param tilted integral for the image rotated by 45 degrees; it is \f$(W+1)\times (H+1)\f$ array with
          the same data type as sum.

scorpiozj · July 10, 2022, 12:40pm

Thanks!
Will do it later.

Topic		Replies	Views
Get rectangle from a document Python imgproc	0	417	May 23, 2022
The behaviour of cv2.rectangle() for the overloaded method is not what I expected Python imgproc	1	621	June 2, 2021
Calculate the sum of pixel in a BW image per row and col C++ imgproc , core	1	567	March 15, 2022
Assertion error when running `rotatedRectangleIntersection` Python imgproc , core	8	1170	April 3, 2021
Simple question on cv::remap() C++ imgproc	1	287	April 16, 2024

Is there the mis-typing for integral image's document?

Related topics