Calculo Integral Imagenes

What are Integral Images?

An integral image, also known as a summed-area table, is a data structure that allows for rapid computation of the sum of pixel values in any rectangular region of an image. It was introduced by F.C. Crow in 1984 as a way to speed up image processing operations.

The integral image is constructed by computing the cumulative sum of pixel values from the top-left corner of the image to each pixel location. This means that each pixel in the integral image represents the sum of all pixels above and to the left of it in the original image.

This recursive definition allows the integral image to be computed efficiently in a single pass through the image. The value at any point (x,y) in the integral image is the sum of all pixels in the original image that are above and to the left of (x,y).

How to Interpret Integral Images

Interpreting integral images involves understanding how the cumulative sum values relate to the original image. The key insight is that the sum of any rectangular region in the original image can be computed using just four values from the integral image.

This formula works because the integral image values account for the overlapping regions that would otherwise be double-counted if we simply subtracted the sums of the individual regions.

Applications of Integral Images

Integral images have numerous applications in computer vision and image processing. Some of the most common applications include:

• Face detection: Integral images are used in the Viola-Jones face detection framework to quickly compute features for classification.
• Object detection: They are used in sliding window approaches to efficiently compute features for object detection.
• Image segmentation: Integral images can be used to quickly compute region properties for segmentation algorithms.
• Image enhancement: They can be used to compute local statistics for adaptive filtering and enhancement.
• Feature extraction: Integral images are used to compute Haar-like features for machine learning applications.

These applications leverage the ability of integral images to quickly compute sums over rectangular regions, which is essential for many image analysis techniques.

Calculating Integral Images

Calculating an integral image involves computing the cumulative sum of pixel values from the top-left corner of the image to each pixel location. This can be done efficiently using dynamic programming.

Algorithm

1. Initialize the integral image II with the same dimensions as the original image I.
2. Set II(0,0) = I(0,0).
3. For each row i from 1 to height-1, set II(i,0) = II(i-1,0) + I(i,0).
4. For each column j from 1 to width-1, set II(0,j) = II(0,j-1) + I(0,j).
5. For each pixel (i,j) where i > 0 and j > 0, compute II(i,j) = I(i,j) + II(i-1,j) + II(i,j-1) - II(i-1,j-1).

This algorithm efficiently computes the integral image in O(n) time, where n is the number of pixels in the image.

Example

Consider the following 3x3 image:

The corresponding integral image would be:

For example, the value at (2,2) in the integral image is 27, which is the sum of all pixels in the original image (1+2+3+4+5+6+7+8+9).