Learning OpenCV, 2nd Edition

When using the integral image to compute a region, we can see by Figure 6-9 that, in order to compute the

central rectangular area bounded by the 20s in the original image, we’d calculate 398!– !9!– !10! + !1! =

!380. Thus, a rectangle of any size can be computed using four measurements (resulting in 𝑂(1)

computational complexity).

1

2

5

1

2

0

0

0

0

0

0

2

20

50

20

5

0

1

3

8

9

11

5

50

100

50

2

0

3

25

80

101

108

2

20

50

20

1

0

8

80

235

306

315

1

5

25

1

2

0

10

102

307

398

408

5

2

25

2

5

0

11

108

338

430

442

2

1

5

2

1

0

16

115

370

464

481

0

18

118

378

474

492

Figure 6-9: The 7-by-5 image of

Figure 6-8 shown numerically at left (with the origin assumed to be the upper-left ) and converted to an (8-

by-6) integral image at right

cv::integral(), for Standard Summation Integral

The different forms of integral are (somewhat confusingly) distinguished in the C++ API only by their

arguments. The form that computes the basic sum has only three.

void cv::integral(

cv::InputArray image, // Input array

cv::OutputArray sum, // Output sum results

int sdepth = -1 // Depth for results (e.g., cv::F32)

);

The first and second are the input and result images. If the input image is of size 𝑊-by-!𝐻, then the output

image will be of size (𝑊 + 1)-by-(𝐻 + 1).

22

The third argument sdepth specifies the desired depth of

the sum (destination) im ...