haar transformation on an image

Question

hello, i need some help with haar transformation, i have to apply it on an image. My math is bad, my english not all that awesome and i find it hard to understand from articles on the internet. I found this page http://www.cs.ucf.edu/~mali/haar/haar.cpp where the haar transformation is applied on 2d matrix. I suppose if i give image pixels matrix in there, it should work? im confused about this stuff, could someone enlighten me a bit please? Thank you!

Answer 1

There is an example at the end of the URL you point to. Look inside the main() function.

• the 2D variant takes a float** and two parameters, height and width
• the float** points to rows of grayscale pixels
• each row is a float*, a pointer to the first pixel in the row
• each float value is the intensity value of the pixel
• in the example code, the dimensions are 4x4.

This is where the memory is allocated:

float **mat = new float*[4];
for(int m=0;m<4;m++)
  mat[m] = new float[4];

This is where the pixel values are set:

mat[0][0] = 5; mat[0][1] = 6; mat[0][2] = 1; mat[0][3] = 2;
mat[1][0] = 4; mat[1][1] = 2; mat[1][2] = 5; mat[1][3] = 5;
mat[2][0] = 3; mat[2][1] = 1; mat[2][2] = 7; mat[2][3] = 1;
mat[3][0] = 6; mat[3][1] = 3; mat[3][2] = 5; mat[3][3] = 1;

This is where the haar2 function is called:

haar2(mat,4,4);

All you need to do is provide the data as needed by the function (float**) with the right dimensions. You probably want to store the results to an output file that you can open in an image viewing application.

Look for the PGM format for a really easy solution. Note that the results of the haar function will give you floating point values, which you may have to compress down to 8 bit to view the image.

Answer 2

If you're trying to do object detection using Haar features, pay a look at OpenCV: http://opencv.willowgarage.com/documentation/cpp/object_detection.html